First Horizon Corporation (NYSE: FHN) — Back at the Price a Buyer Offered Four Years Ago, Without the Buyer

Independent fundamental equity research. Report date: 26 July 2026.

⚡ Claude’s Take

This block is the author’s own subjective opinion, and it is the only part of this article that takes a position. It is general information, not investment advice. No recommendation and no price target appears anywhere else in the piece — the analysis that follows discusses valuation only as embedded expectations and scenarios.

Verdict: HOLD / trim-into-strength — an average bank at a full price, not a bargain and not a short. Accumulate only on a de-rate toward ~$19–22 (~1.30–1.50x tangible book). Conviction: medium.

First Horizon is a competently run, honestly reported, average regional bank whose one genuinely distinctive feature — the takeover option — its own management deflated on 15 October 2025, and whose share price now sits within a few cents of the $25.00 in cash TD Bank Group agreed to pay for it in February 2022. The operating record over that span is the thing to internalise: adjusted pre-provision net revenue was $1,374M in FY2022 and $1,372M in FY2025 — zero growth in four years on a balance sheet that added $6B of loans, while the adjusted efficiency ratio went backwards from 56.6% to 60.0% and adjusted EPS ($1.89) remains below FY2021’s $2.07. Every dollar of per-share progress since 2023 came from retiring 15.3% of the shares — a genuinely well-executed buyback at a blended $19.55, but one that has now run out of road: at 1.76x tangible book, every $100M repurchased destroys about $43M of tangible book to buy $8.4M of earnings. Meanwhile the reported 15.2% return on tangible common equity becomes 13.98% stripping out the quarter’s reserve release and roughly 13.0% credit-normalised — against a price that, on a 10.5% cost of equity, requires a sustained 16.2%. That ~3.15-point gap survives every accounting adjustment, including the AOCI argument usually deployed to close it (adjusting for AOCI lowers the return and raises the book; it is a wash).

What is genuinely good deserves saying: the best charge-off rate in an eight-bank peer set (19–20bp), a mortgage-warehouse book with one basis point of ten-year losses and physical note custody, office exposure of only ~4% of loans and half of that medical, no purchase-accounting or adjusted-EPS games (FY2025 notable items were two cents), a 6.6–7.9% total payout yield, and a CEO with $44.5M of his own money in the stock. But the moat isn’t there, and its absence is measurable: FHN’s deposits cost 2.05% against Regions’ 1.37% in the same states — 68bp on $65.4B is roughly $445M a year, about five points of ROTCE. Deposit share has fallen five consecutive years in Tennessee and the #1 rank likely flips to Pinnacle in the September 2026 FDIC data. The framing, grounded in the factor read, is “regional-bank beta in a quiet uptrend, priced as though it were something better” — Momentum is L1-zeroed in all four models, DividendYield +0.97 and Value +0.40 dominate, and its 0.945 similarity to the KRE ETF says plainly that this is the index wearing a ticker. It is not a falling knife and not a crowded momentum trade; it is a name that has already fully re-rated. The takeout hope many investors think they hold for free is worth $0.21–$1.49 a share against a ~$6.14 Gordon residual — because at 1.757x FHN already trades inside this cycle’s actual deal range for healthy franchises (1.52x–2.00x, median ~1.77x) rather than at a discount to it, and every acquirer in that cycle was marked down on announcement. Conviction: medium. Flips bullish on the adjusted efficiency ratio printing below 57% for two consecutive quarters with revenue growth outpacing loan growth — the one observable proof that the $100M+ self-help programme is real. Flips bearish on charge-offs above 35bp for two quarters while the allowance is rebuilt, or on an acquisition struck above ~1.76x tangible book without 30% credible cost saves. Tag: “The buyback was the business — and at this price the buyback has stopped working.”

📈 Stock Price Action — Five-Year Event Map

Over the trailing five years First Horizon has traded a violent round trip: from ~$15 in mid-2021 to a low of $8.99 on 4 May 2023 (the session TD Bank Group walked away from its buyout) to a five-year closing high of $26.23 on 6 February 2026. The stock closed at $25.53 on 24 July 2026 — 3.9% below its 52-week high of $26.56 (6 Feb 2026) and 28.9% above its 52-week low of $19.80 (16 Oct 2025). The arc’s defining fact: after four and a half years, two drawdowns of nearly 40% or worse, and a collapsed merger, FHN trades almost exactly at the $25.00 per share in cash a strategic acquirer agreed to pay in February 2022. (Price moves are FACT, from the AZI five-year price series; attributed causes are INTERPRETATION.)

1 — Rate-hike repricing (Sep 2021 – Feb 2022). FHN bottomed at $14.67 on 20 September 2021 and ran 27.6% to $18.72 by 18 January 2022 as the market repriced the Fed’s tightening path and bid asset-sensitive regional banks; it had faded to $18.25 by 25 February 2022.

2 — The TD deal (28 February 2022). In a single session the stock gapped +28.7%, $18.25 → $23.48, on TD Bank Group’s agreement to acquire First Horizon for $25.00 per share in cash — the second-largest one-day move of the five-year window.

3 — The merger-arb year (Mar 2022 – Feb 2023). For twelve months FHN stopped trading as a bank and traded as a spread: mean close $23.40 inside a $20.74–$24.88 band, an average 6.4% discount to the $25.00 consideration, for a net +5.5% price return. The widening spread through late 2022 was the market handicapping regulatory-approval risk.

4 — The March 2023 regional banking crisis. From $24.77 on 28 February 2023 the stock fell 39.7% to $14.93 by 17 March 2023, including a −20.2% session on 13 March. FHN traded roughly 40% below a $25.00 cash deal price — the market pricing a high probability of break, not a bank valuation.

5 — TD terminates (4 May 2023). The merger was terminated and the stock fell 33.2% in one session, $15.05 → $10.06, touching an intraday $8.99 — the largest single-day move and the absolute five-year low. It flushed further to $9.67 on 12 May. Combined with event 4, the February–May 2023 peak-to-trough drawdown was −60.8% over 86 days, and the February 2023 peak was not regained on a closing basis until 23 July 2025 — 889 days later.

6 — Standalone recovery and re-rating (May 2023 – Dec 2024). From the $9.67 trough the stock rose 108% on price (+137% total return) to $20.14 by year-end 2024 as deposits stabilised, the termination fee landed and buybacks resumed; the single largest contribution was a +16.5% session on 6 November 2024 on the US election result and the associated deregulation/M&A trade.

7 — The 2025 round trip. FHN fell 19.5% to $16.21 by 4 April 2025 in the tariff shock (−12.4% on 3 April), rebounded 39.5% to $22.61 by 30 September, then gave back 11.8% into 16 October, including a −9.4% session on 15 October 2025 — the day management signalled it would rather be an acquirer than a target. Full-year 2025 maximum drawdown was −27.1%, recovered within 97 days of the trough.

8 — Back to the deal price (Oct 2025 – Jul 2026). The stock rallied 31.5% off the October 2025 low to a five-year closing high of $26.23 on 6 February 2026, having closed at or above the old $25.00 TD price for the first time on 3 February 2026 — three years and nine months after the deal died. It pulled back 17.2% to $21.73 by 19 March 2026, then recovered 17.5% to $25.53. On 15 July 2026 FHN reported 2Q26 EPS of $0.54 against $0.52 consensus and net income available to common of $260M, up 12% year-over-year; the shares fell 3.0% that session.

1. Executive Summary

First Horizon Corporation is a Memphis, Tennessee–headquartered regional bank with $84.4B of assets at 30 June 2026, over 450 business locations across 23 states including more than 400 banking centres in 12, and roughly 7,400 employees. Founded in 1864, roughly doubled by the 2020 merger of equals with IBERIABANK, it operates through three segments restated in 2024 — Commercial, Consumer & Wealth (89% of revenue and effectively all of the profit), Wholesale (mortgage-warehouse lending, franchise finance, correspondent banking and the FHN Financial fixed-income dealer), and Corporate. Revenue is heavily spread-weighted: FY2025’s $3,419M comprised $2,622M of net interest income against $797M of fees, a 23.3% fee contribution that is bottom-third among peers.

The central tension of this report is that the operating record and the share price have been telling different stories for four years. On the reported numbers First Horizon looks like a bank in good health: FY2025 return on tangible common equity of 14.01% rising to 15.2% in 2Q26, a 3.47% net interest margin above five of eight covered comparables, the best net charge-off rate in its peer group at 19–20bp, tangible book value per share compounding to $14.53, and 15.3% of the share count retired since year-end 2023. On the underlying numbers it is a bank standing still: adjusted pre-provision net revenue was $1,374M in FY2022 and $1,372M in FY2025 — zero growth across four years on a balance sheet that added $6B of loans — while the adjusted efficiency ratio deteriorated from 56.6% to 60.0% and adjusted EPS of $1.89 remains below FY2021’s $2.07. Every dollar of per-share progress has come from the buyback, executed well at a blended $19.55 but now decaying: at 1.76x tangible book each $100M repurchased destroys roughly $43M of tangible book to buy $8.4M of earnings.

There is no durable competitive advantage here, and its absence is measurable rather than merely arguable. Deposit share has fallen for five consecutive years in Tennessee (15.93% → 13.29%), in Nashville (8.32% → 5.97%, now fifth), in Chattanooga and in Lafayette; roughly 75% of the headline Memphis share is booked to a single headquarters branch record, and normalised for that, FHN’s Memphis position has been flat for a decade and ranks second to Regions. The deposit base is 58% commercial, only 23.5% noninterest-bearing and 42% uninsured, and it reprices symmetrically with rates — roughly 63% up, 66% down — which is the signature of commodity funding rather than captivity. The cost is quantifiable: FHN’s all-in deposit cost of 2.05% against Regions’ 1.37% in the same Deep South states is worth about $445M of pre-tax income a year, roughly five points of return on tangible common equity. With Regions’ funding franchise, First Horizon would earn 19% rather than 14%.

Earnings quality cuts both ways, and the honest version is mixed. To management’s credit, the gap between GAAP and adjusted earnings has collapsed to two cents, IBERIABANK purchase accounting is fully run off, and held-for-sale balances show no evidence of manipulation — this is cleaner reporting than most of the cohort. Against that, the allowance has been released for three consecutive years to 1.24% of loans, the thinnest in the peer set, contributing roughly $0.08 of FY2025 EPS; $283M of modified commercial real estate — 2.09% of that book — sits outside the non-performing numbers, roughly doubling the true CRE problem book to ~3.8%; and reported CET1 of 10.5% is approximately 9.5% fully AOCI-inclusive.

On valuation, the market requires a return the company has never sustained. At $25.53 — 1.757x tangible book, the 89.6th percentile of FHN’s own decade and above every year-end reading of the last nine except the TD deal year — the Gordon identity implies a sustained 16.2% return on tangible common equity against a credit-normalised 13.0%, a gap of roughly 3.15 points robust to the AOCI convention. Cross-sectionally First Horizon sits exactly on the peer regression line, which is the point: it is not mispriced relative to its peers, but the whole cohort is priced on returns flattered by the same benign credit and the same reserve drawdown, and with a 0.945 factor similarity to the KRE ETF, FHN carries the full weight of that error. The takeout option many investors believe they hold for free is worth $0.21–$1.49 a share against a ~$6.14 valuation residual — because at 1.757x the stock already trades inside this cycle’s actual deal range for healthy franchises — 1.52x–2.00x, median ~1.77x — not at a discount to it. Scenario work puts the case that requires nothing to go wrong — simple continuation of the four-year record — at approximately zero total return over two and a half years.

The sections below carry no recommendation and no price target; they argue the embedded expectations and the scenarios. (No recommendation; no price target — those appear only in the opinion block above.)

2. Business Overview

What the company does. First Horizon Corporation is a financial holding company headquartered at 165 Madison Avenue, Memphis, Tennessee, founded in 1864 and today one of the larger U.S. regional banks with $84.4B in total assets at 30 June 2026. Its principal — and only — banking subsidiary is First Horizon Bank. At 31 December 2025 the company operated over 450 business locations in 23 states, including over 400 banking centers in 12 states, and employed approximately 7,400 associates. (FACT — FY2025 10-K, Item 7 MD&A Introduction; 2Q26 earnings release, 15 July 2026.)

Like every deposit-funded spread lender, FHN earns money three ways: net interest income (the spread between what it earns on loans and securities and what it pays for deposits and borrowings), noninterest income (fixed-income sales and trading, wealth and brokerage, service charges, card, mortgage banking), and operating leverage on a largely fixed cost base. The mix is heavily spread-weighted: in FY2025 total revenue of $3,419M comprised $2,622M of net interest income and $797M of noninterest income — a fee contribution of 23.3%, down from 26.8% in FY2023. (FACT — FY2025 10-K, Table 7.1 and Note 19.) FHN is therefore a more spread-dependent, less fee-diversified franchise than the better-balanced super-regionals, and correspondingly more exposed to the rate cycle and to deposit competition.

Three segments — restated in 2024. An important housekeeping point that stale data sources still get wrong: during 2024 FHN reorganized its internal management structure and reclassified its reportable segments. The long-familiar “Regional Banking / Specialty Banking / Corporate” cut is retired. Prior periods have been restated to conform. The current segments are (FACT — FY2025 10-K, Note 19 “Business Segment Information”):

Where the money is actually made. The segment economics are lopsided, and the lopsidedness matters for how the business should be valued (FACT — FY2025 10-K, Note 19, Table 8.19.1, $M):

Three observations follow. First, CC&W is the company: it generates 89% of revenue and $1,590M of the $1,345M consolidated PPNR — more than 100%, because Corporate is a −$414M drag. Second, Wholesale is not the diversifier it appears to be. It converts $489M of revenue into only $169M of PPNR, a 65% expense-to-revenue ratio against CC&W’s 48%. The segment that supposedly smooths the earnings stream is the one that earns the thinnest margin on its revenue — and it houses the most volatile revenue line in the company. Third, the Corporate drag is large and persistent, reflecting funds-transfer-pricing residuals, unallocated capital costs and the legacy run-off books that still, sixteen years on, have not fully disappeared.

The loan book. Total loans and leases were $65.3B at 30 June 2026 ($64.2B at year-end 2025). The composition at 31 December 2025 (FACT — FY2025 10-K, Item 1, Tables 1.1, 1.3a, 1.3b):

Within C&I, the disclosed line-of-business split is: loans to mortgage companies 13%, finance and insurance 12%, real estate and rental/leasing 11%, wholesale trade 7%, healthcare and social assistance 7%, accommodation and food service 7%, manufacturing 6%, retail trade 5%, transportation and warehousing 5%, other C&I 27%. (FACT — Table 1.3a.)

Two features of this book deserve emphasis up front, because they cut in opposite directions. The office exposure is small and unusually defensive — 20% of a $14B CRE book is roughly $2.8B, and with nearly half of it medical office, traditional office is only about $1.4B, or ~2.2% of total loans. That is a genuine and underappreciated positive. Against it, roughly a quarter of the $36B C&I book — call it ~$9B — is lent to other financial intermediaries. That is the single largest concentration in the company, and it is the channel through which mortgage-sector and private-credit stress would reach First Horizon. the financial-quality section and the risk section return to both.

The deposit franchise. Total deposits were $67B at year-end 2025 ($67.5B period-end). The composition is the most important disclosure in the filing for anyone assessing whether this bank has a moat (FACT — FY2025 10-K, Table 1.4):

Total estimated uninsured deposits were $28.1B, or 42% of deposits, at 31 December 2025, up from $26.7B / 41% a year earlier; $5.2B (8% of deposits) was collateralized. (FACT — FY2025 10-K, MD&A deposits section.)

Recurring versus cyclical revenue. The spread book and the wealth/treasury fee lines are genuinely recurring. The fixed-income business inside Wholesale is not. Its output is measured in average daily revenue (ADR), which was $594K in 2Q26 — up 8% year-over-year but down sequentially, a decline management attributed to “macroeconomic volatility amidst a changing geopolitical environment and uncertain rate environment.” (FACT — 2Q26 earnings call, 15 July 2026.) The 10-K is explicit that this business is a partial hedge: “our fixed income business tends to perform better when rates decline or markets are moderately volatile, which tends to partially offset net interest margin compression,” while “a flat or inverted yield curve … tends to decrease net interest margin … and it generally reduce[s] FHN’s revenues from its fixed income bond trading.” (FACT — FY2025 10-K, Item 1A.) Both effects occurred from 2022 through 2024.

Verdict on the business model. First Horizon is a conventional, spread-dominated Southeast/Gulf-South commercial bank with a modest, low-margin capital-markets appendage and a comparatively commercial, comparatively rate-sensitive funding base. It is neither exotic nor fragile. What it is not is differentiated: strip away the segment labels and this is a $84B balance sheet earning a 3.47% margin on a 60%-plus efficiency ratio — a description that fits a dozen other listed banks.

3. Industry Dynamics

3.1 Structure: a consolidating industry with a mediocre through-cycle return

There were 4,278 FDIC-insured institutions at 31 March 2026, down 60 in the quarter, against a post-1940 peak of 14,496 in 1984 — a 70% reduction in charters over four decades, running at 40–60 disappearances a quarter and almost entirely merger-driven (the first quarter of 2026 produced three new charters and one failure). (FACT — FDIC Quarterly Banking Profile.) First Horizon, at $84.4B, sits in the largest sub-$100B tier: above the $10B Durbin/CFPB threshold, above the $50B resolution-planning threshold, and below the $100B Category IV line.

Industry profitability is at a cyclical high that has already begun to roll over. First-quarter 2026 net income was $80.5B, ROA 1.26% (up from 1.16% a year earlier), efficiency 56.0%. But net interest margin fell 8bp sequentially to 3.31% as asset yields dropped 21bp against a 13bp fall in funding costs, and net interest income declined $1.6B in dollar terms. The earnings beat came from noninterest income (+$5.0B, concentrated at the largest banks) and from reserve releases — industry coverage fell from 171.2% to 166.8%. (FACT — FDIC QBP, Q1 2026.)

The long-run base rate is the number to keep in view. Industry return on assets has exceeded 1% in only fourteen years since 1935; the long-term average is roughly 0.75%, and industry return on common equity since 1935 has averaged about 10%. (FACT — Deutsche Bank Securities bank primer, 2011 — third-party research, framework rather than current data; directionally corroborated by current FDIC series.) Today’s 1.26% ROA is therefore roughly 1.7x the through-cycle norm.

3.2 The rate regime turned in June 2026 — and this changes the sector setup

The single most important macro fact for this report, and one that post-dates several of our own prior bank reports: the FOMC held at 3.50–3.75% on 17 June 2026 — a fourth consecutive hold — and revised the 2026 dot plot upward. Nine of nineteen participants now see a hike, eight no change, one a cut; the inflation forecast was raised to 3.6% headline / 3.3% core. The two-year note trades at 4.21% and the ten-year at 4.56%, leaving 2s10s at just +36bp with the two-year some 50–60bp above the funds rate. (FACT — FOMC materials and Treasury yields, June–July 2026.)

This matters directly. The entire 2025 regional-bank earnings story was falling deposit costs — First Horizon’s own deposit cost fell 42bp year over year, and that alone drove the 12bp of margin expansion. That tailwind has stopped. FHN’s interest-bearing deposit cost already rose 5bp sequentially in 2Q26, to 2.33% average and a 2.43% spot rate. A 66% cumulative deposit beta is a gift while the Fed cuts and a liability if it hikes.

3.3 The capital cycle: capacity is leaving, but credit supply is expanding

Applying Marathon’s supply-side lens honestly produces an uncomfortable answer. Bank M&A in the first half of 2026 hit a seven-year high: 83 deals, roughly $15.1B of announced value. Closed in January and February 2026 alone: Pinnacle/Synovus ($8.6B, creating a $117–123B institution), Huntington/Cadence, PNC/FirstBank and Fifth Third/Comerica — with Santander/Webster announced in February 2026 and still pending. (FACT.)

But this is the wrong kind of consolidation. Marathon’s framework requires capacity to exit so that survivors gain pricing power. Banking consolidation removes charters and branches — it does not remove credit or deposit supply, both of which are national and substitutable. Removing a competitor in Nashville does not let the survivor charge more; the loan gets bid by a bank in Charlotte or a fund in New York. Meanwhile the Federal Reserve’s H.8 series shows bank loans of $13.9T growing 6.1% annualised, with C&I at $2.9T growing 10.9% annualised. Supply is expanding, not contracting.

Private credit is the dominant supply-side development, and it is now large enough to quantify. The global market is roughly $1.96T (2026), growing at about 12% a year; US direct lending at $1.5–2T now matches the broadly-syndicated loan market and is forecast to reach $3T by 2028. Fourth-quarter 2025 pricing tells the story: syndicated loans at roughly SOFR+300 against direct lending at SOFR+480–500, with spreads compressing amid what participants describe as heated competition. Private-credit default rates rose from 8.1% (2024) to 9.2% (2025), and “bad PIK” reached 6.4% of volume in the first quarter of 2026. (FACT.)

The banking system’s response has been to lend to the funds. Loans to non-depository financial institutions reached $1.4T at 31 December 2025 — 5.6% of industry assets, compounding at 22.7% a year since 2010, and up 35.2% in 2025 alone; H.8 shows $1.99T by 24 June 2026, +15.4% year over year, the fastest-growing category in the system. (FACT.)

The structural problem for First Horizon is that it gets the disintermediation without a competitive share of the offsetting pool: 86% of NDFI balances sit at banks above $100B, and ten banks hold roughly two-thirds. The middle-market and asset-based credits FHN underwrites are precisely what private credit has been taking, and it takes the borrowers who most value speed, covenant flexibility and hold size — disproportionately the profitable ones. A lender whose spreads hold while direct-lending spreads compress 180bp toward it is not demonstrating pricing power; it is being handed a worse cohort. (INTERPRETATION.)

3.4 Regulation: the one cycle running in the sector’s favour

The regulatory cycle is, unusually, a tailwind — and it deserves to be stated precisely because it is real, quantifiable and near-dated (all FACT):

• Basel III Endgame was rescinded and re-proposed as relief. On 19 March 2026 the Fed, OCC and FDIC issued three notices of proposed rulemaking formally withdrawing the 2023 proposal. Estimated risk-weighted-asset relief: ~4.8% for GSIBs, ~5.2% for Category III/IV, ~7.8% for community banks. Comments closed 18 June 2026; finalisation is expected in the fourth quarter, implementation in 2027. Management expects roughly a 10% reduction in FHN’s standardised risk-weighted assets.
• The FDIC special assessment is fully off the run-rate — final payment 30 March 2026 at a reduced 2.97bp. The Deposit Insurance Fund stands at $157.4B, a 1.43% reserve ratio, the highest in over twenty years. A July 2026 proposal would cut base assessment rates and raise the large/small dividing line from $10B to $30B, saving the industry an estimated ~$3.4B a year.
• The brokered-deposit proposal was withdrawn on 3 March 2025 and the FDIC has said it does not intend to reissue one.
• Stress-test capital buffers are frozen at 2025 levels through 2027 pending framework revision.
• The resolution-planning threshold is proposed to move from $50B to $100B — which would take First Horizon out of that perimeter entirely.
• Merger approval friction has eased sharply. The 2024 merger guidelines were rescinded and the OCC restored streamlined procedures effective 15 May 2025; Pinnacle/Synovus went from announcement (24 July 2025) to close (1 January 2026) in about 5.3 months. Market participants tie the window to the November 2026 midterms.

But note what this tailwind is and is not. It is a one-time transfer from the regulator to the shareholder, reversible by an election. A better industry lets survivors charge more; this one lets survivors hold slightly less capital.

The $100B threshold is the strategically decisive regulatory fact for First Horizon. Crossing it attaches an annual capital plan, biennial supervisory stress testing with a model-driven Stress Capital Buffer replacing the fixed 2.5% conservation buffer, FR Y-14/Y-15 reporting build-out, monthly liquidity reporting, recovery planning, a triennial FDIC resolution plan — and, under the current proposal, mandatory inclusion of most AOCI in CET1 over a five-year phase-in. Three independent estimates of the cost: FHN’s own 10-Q (3Q23) put it at “roughly on the order of $100 million per year”; Pinnacle, which actually crossed on 1 January 2026, reports ~$45M one-time plus $35M a year ongoing; and Webster — at $84.1B, First Horizon’s near-exact structural twin — told investors it expected significant compliance costs approaching the tier, a curve that market commentary identified as a core rationale for its sale to Santander. (FACT.)

$100M a year is roughly 7–8% of FHN’s pre-tax income and about 12bp of ROA. But at 1–2% organic asset growth ($83.9B → $84.1B → $84.4B over the last three quarters) First Horizon does not cross $100B organically for a decade. The threshold is therefore irrelevant organically and decisive strategically: because the cost is a step function, the only rational way to cross is with a transaction large enough that $100M a year is a small share of the combined expense base. A bolt-on that lands FHN at $95–105B is the worst possible outcome — full cost, no scale. (INTERPRETATION, and a point the capital-allocation section returns to.)

3.5 First Horizon’s end markets: a demographic barbell with the moat in the wrong half

(FACT — Census Bureau metropolitan population and migration estimates.)

Shelby County — Memphis — lost 3,379 residents in 2024, more than any county in the United States. Louisiana led the nation in population loss for a second consecutive year, with job growth of +0.18% since 2015 against +10.4% nationally. Nashville, by contrast, runs ~2.7% unemployment and added 28,000 jobs year over year.

Here is the structural asymmetry that defines First Horizon. The company holds a 33.35% deposit share in the Memphis MSA ($14.01B at June 2025) — a genuine local scale-plus-captivity position in Greenwald’s taxonomy, and the strongest competitive asset it owns — in a metro that is smaller than it was in 2020. Its growth markets (Nashville, Houston, Dallas, Florida, the Carolinas, Atlanta) are precisely where every other bank is expanding, where Pinnacle/Synovus, Fifth Third/Comerica and Huntington/Cadence have just added scale, and where FHN is a small, contested price-taker. This is Greenwald’s dictum that market growth is the enemy of economies-of-scale advantages in its purest form: the moat sits in the shrinking market, and the growth is where there is no moat. (INTERPRETATION, on FACT.)

3.6 Commercial real estate: stabilising at the property level, deteriorating at the maturity wall

Office vacancy was 14.0% at year-end 2025, just 4bp above 2024 — still the worst of the four major property types, but the rate of deterioration has collapsed. Construction and development lending fell 5.8% in 2025. Against that, non-owner-occupied CRE noncurrent loans reached 1.59%, the highest since Q4 2013, and the benign headline is partly manufactured: modified CRE totalled $11.6B (0.38% of CRE-secured loans), 82% of it still performing — and banks above $100B, holding only 29% of CRE loans, did more than half of all 2025 modification volume. The unmodified market mark is much worse: CMBS office delinquency hit an all-time high 12.34% in January 2026, roughly eight times the bank CRE delinquency rate, and 2026 CMBS maturities total $146.2B including $76.6B of hard maturities, about 39% of them in the fourth quarter. The Federal Reserve’s own 2026 stress test assumed an 8.8% CRE portfolio loss rate. (FACT.)

On the regulatory concentration tests, the relevant benchmark is unflattering to the tier: banks in the $10–100B cohort run a median CRE concentration of 289% of tier 1 capital plus allowance — the highest of any size group. First Horizon’s $13.6B of CRE against roughly $9.7B of total risk-based capital would put it well below that median — but note that FHN does not disclose its CRE-to-risk-based-capital ratio directly, so this is a derived figure and an open question (the open-questions section) rather than a reported one.

3.7 The two specialty end markets

Mortgage warehouse. The Mortgage Bankers Association forecasts $2.2T of single-family originations in 2026 (+8%) against a forecast 4.2% average ten-year yield. Warehouse line spreads run roughly 175–250bp on agency product. It is a good business — FHN calls it its highest-spread lending segment — but it is structurally moat-free: volume is a pure derivative of the mortgage rate, originators multi-bank their lines by design to guarantee capacity, so switching costs are essentially zero, pricing is a public grid, and capacity is expanding. (FACT / INTERPRETATION.)

FHN Financial’s market. Demand for securities from depository institutions is a residual of deposit growth in excess of loan demand, the loan/deposit ratio, and the shape of the curve. All three currently point the wrong way: deposits are growing, but loan growth accelerated to 7.1% annualised, absorbing that liquidity into loans rather than bonds; and with 2s10s at +36bp and the two-year above the funds rate, nobody is extending duration into a possible hiking cycle. Worse, the customer base is itself disappearing at 50–60 institutions a quarter, and the acquirers run their own desks.

Verdict — Industry Dynamics

Structurally a below-average industry, and one whose structure has deteriorated since 2022. The product is perfectly undifferentiated. There is no supply/cost advantage — funding is priced by the Fed and by money-market funds paying ~4.50% against a record $8.28T of assets, and technology is bought from the same vendors by everyone. There is no demand-side captivity of consequence — retail deposit stickiness was permanently repriced by the 2023 failures and instant digital transfer (industry uninsured deposits grew 7% in 2025), and the commercial borrower now has a $2T direct-lending alternative. Scale economies exist but the evidence they convert into efficiency is weak: across roughly fifty US banks from $50B to over $4T there is no significant correlation between asset size and efficiency ratio. The only genuine barrier is regulatory — and Greenwald’s point applies exactly: government protection here protects the industry, not any participant. Every insured bank holds the same charter; it excludes neither the other 4,277 banks nor the private-credit funds that need no charter at all.

Add 10–12x leverage on a commodity product, non-linear correlated credit losses, a 0.75% through-cycle ROA, a permanent regulatory tax with a step function at $100B, and terminal-value risk from a curve that has removed 70% of US charters since 1984. The narrow 12–17% ROTCE band at 1.3–2.3x tangible book across the entire super-regional cohort is itself the evidence: a narrow return band across many participants is the signature of weak differentiation.

Cyclically, the setup is unfavourable. Earnings are at a cyclical high (1.26% ROA against a 0.75% norm) and already inflecting; the rate regime turned against the cohort in June; supply is expanding; and credit is the one thing not yet deteriorating and the thing most likely to. The regulatory tailwind is genuine but is a one-time capital transfer, not a structural improvement.

4. Competitive Position

The question this section has to answer is simple: does First Horizon possess a durable competitive advantage, and if so, what is its mechanism? For a bank there are only two candidate answers that survive scrutiny — local deposit-scale density with customer inertia (Greenwald’s economies-of-scale-plus-captivity, and the only genuine moat in banking), and a differentiated specialty franchise with switching costs. First Horizon claims both. Neither survives the evidence.

4.1 The hometown share is substantially a booking entry

First Horizon’s strongest apparent asset is its Memphis franchise — a headline 33–40% deposit share of the Memphis MSA, depending on the geographic cut, which on its face is exactly the local density Greenwald identifies as a real advantage. (FACT — FDIC Summary of Deposits, 30 June 2025.)

But roughly 75% of FHN’s Memphis MSA deposits — about $10.50B of $14.01B — are booked to a single branch record: the 165 Madison Avenue headquarters. (FACT — FDIC SOD branch-level data.) Headquarters branch records absorb corporate, treasury, brokered and non-retail balances that have no local franchise character at all. Normalise for them and the ranking inverts: Regions 16.32% (#1), First Horizon 12.86% (#2) — and FHN’s ex-headquarters Memphis share has been 11.00% in 2015 and 11.15% in 2025: flat for a decade.

Two caveats in fairness. First, this normalisation threshold is a judgement, though the underlying fact — three-quarters of the “share” in one branch record — is not. Second, Pinnacle does exactly the same thing in Nashville, so the adjustment must be applied symmetrically across the peer set and does not single FHN out for bad practice. The point is not that FHN is misleading anyone; it is that the headline number that appears to be the company’s best evidence of a moat mostly is not measuring one.

4.2 Greenwald test one — market-share stability: fails

A genuine barrier to entry shows up as stable market share. First Horizon’s is falling almost everywhere (FACT — FDIC Summary of Deposits):

Tennessee deposit dollars were flat ($30.76B → $31.03B) while the Tennessee market grew 20.9%. Post-IBERIABANK, deposits fell 21.7% in Louisiana, 20.2% in Arkansas and 12.4% in Florida; total Summary-of-Deposits balances are down 1.7%.

And the last remaining bragging right is about to go. First Horizon holds 13.29% of Tennessee deposits against Pinnacle’s 12.94% — and Pinnacle closed its $8.6B all-stock merger with Synovus on 1 January 2026, creating a $117.2B institution. On 30 June 2025 data, pro-forma Pinnacle is roughly $30.60B against FHN’s $31.03B — about 19 basis points behind. Pinnacle plans to hire 225–250 revenue producers in 2026 and 250–275 in 2027, and its CEO Kevin Blair is explicit about method: “Disruption is our friend. We’re not hiring headhunters… It’s identifying who the best bankers are in each market.” (FACT.) FHN’s number-one rank in its home state most likely flips in the September 2026 Summary of Deposits — a clean, dated falsification test (the What Must Be True section).

4.3 Greenwald test two — the deposit moat: fails, and the cost is quantifiable

This is the decisive test, because in banking the deposit franchise is the only thing that can be a moat. Two facts settle it.

First, the structure. FHN’s deposits are 58% commercial / 42% consumer, only 23.5% noninterest-bearing (down from 37.2% in 2021 and never recovered), and 42% uninsured. Commercial balances are larger, more rate-aware, more concentrated and far more mobile than granular retail money.

Second, the behaviour. FHN’s interest-bearing deposit beta through the hiking cycle was roughly 63–64% — the worst in its peer set: Regions 35%, Truist 46%, Pinnacle 48%, Citizens 50%, M&T and East West 56%. (FACT — computed from filed deposit costs.)

Management presents the 66% down-beta as evidence of franchise strength, and the financial analysis initially read it the same way. On reflection that reading is wrong, and I side with the competitive-position analysis. A deposit base that follows rates up at ~63% and down at ~66% is symmetric — which is the signature of commodity money, not captivity. A real franchise is asymmetric: it reprices slowly on the way up and quickly on the way down. FHN’s skill here is treasury execution, and it is genuine; but it is not a moat. The CEO concedes the direction of travel himself: “the cost of deposits is drifting slowly towards wholesale cost of funds just with the transparency of interest rates.”

Now price it. FHN’s all-in deposit cost is 2.05% against Regions’ 1.37% — in the same Deep South states. That 68bp gap on $65.4B of deposits is roughly $445M of pre-tax income a year: about 35% of FY2025 pre-tax income, or approximately 5.1 points of return on tangible common equity. (INTERPRETATION, arithmetic on filed inputs.) With Regions’ deposit franchise, First Horizon would earn roughly a 19% ROTCE instead of 14%. That single comparison is the most complete answer to the moat question in this report: the moat is absent, the absence is measurable, and it costs about a third of pre-tax earnings.

Corroborating detail: core fee income is shrinking in nominal dollars — deposit and cash-management fees $165M → $158M, card fees $67M → $64M. A franchise with pricing power does not show falling nominal service-charge revenue.

4.4 Greenwald test three — persistent returns above the cost of capital: fails on Greenwald’s own threshold

Greenwald’s diagnostic for a franchise is a persistent return well above the cost of capital — his working band is 15–25%. First Horizon’s return on total common equity was 11.01% / 8.80% / 11.30% in 2023–2025 against a cost of equity of roughly 10–11%. Normalised return on tangible common equity, adjusting for the AOCI-shrunken denominator, runs roughly 12.4–13.5%. (FACT / INTERPRETATION.)

In Greenwald’s terms, earnings power value approximates asset value — there is no franchise margin to explain. That is the arithmetic definition of a business without a moat. It is not a disaster; it is a company earning a modest, positive, unremarkable spread on its capital.

4.5 The specialty franchises: one real edge, no durable moat

Mortgage warehouse lending is the best-run thing First Horizon does. The record is genuinely excellent — “over the last 10 years our mortgage warehouse business has averaged about 1 basis point annual net charge-offs” — and the operational explanation is specific and credible: “We hold the underlying collateral. We take each note at closing… We pick the closing attorney. We take physical ownership of the actual loan docket. It sits in the same vault as the mortgages that are on our balance sheet… On the fraud piece of it, we have seen collateral double pledged elsewhere. We do not have that situation.” FHN has taken real share as Comerica (2023), Flagstar (2024) and Independent Financial (2024) exited. (FACT.)

But it is not a moat. Originators multi-bank their warehouse lines by design to guarantee capacity, so switching costs are structurally zero; pricing is a public grid (roughly 175–250bp on agency product); volume is a pure derivative of the mortgage rate; and capacity is expanding again. FHN’s operational discipline earns it a lower loss rate, not a higher price. It is a well-executed commodity business.

FHN Financial is a cyclical trading business, and the market-position claim cannot be verified. Fixed-income revenue ran $423M (2020) → $133M (2023), −69% → $206M (2025) — still 51% below the peak; average daily revenue has gone $1,436K (FY2021) → $594K (2Q26). Distribution is shrinking: 29 offices in 18 states down to 25 in 17. (FACT.)

Critically, First Horizon makes no market-position claim for this business in its own filings. The FY2025 10-K says only: “We have many competitors within both groups, including major U.S. and international securities firms as well as numerous regional and local firms.” No ranking, no client count, no named competitors. No independent league table validating a top-three or top-five position could be located. Any “one of the largest fixed-income dealers to depositories” claim in circulation is marketing, not filing-grade — record it as unverified rather than as fact.

And the economics settle the matter regardless: the business runs a ~60% compensation ratio, so only about forty cents of each incremental revenue dollar reaches pre-provision net revenue. That is Greenwald’s classic case of the star talent owning the franchise, not the shareholder. The Wholesale segment as a whole runs a 65% expense-to-revenue ratio against Commercial, Consumer & Wealth’s 48%.

4.6 Structural position: density in the wrong markets, and a technology gap

• First Horizon is #1 in four MSAs — Memphis, Knoxville, Chattanooga and Lafayette — every one of them a slow-growth market. (FACT.)
• It is below 2% share in every Sun Belt growth metro: Raleigh 1.42%, Miami 1.71%, Charlotte 0.22%, Houston 0.40%, Dallas 0.08%. Texas is served by seven or eight branches — a loan-production presence, not a franchise.
• The technology gap is not closeable. FHN’s computer-software expense was $138M in FY2025. JPMorgan’s 2026 technology budget is $19.8B — its annual increase of roughly $2B is fourteen times First Horizon’s entire software line. FHN buys parity from third-party vendors (the LPL wealth platform conversion is the current example), and Greenwald’s observation applies directly: technology bought from vendors available to everyone confers advantage on none.
• The value proposition is people-dependent, which is the same thing as saying it is rentable. 53 bankers were hired in 2Q26 alone, with named external hires across consumer banking, deposit product, Charlotte, Atlanta, healthcare and Memphis retail. A market in which relationships are acquired with compensation packages is, by definition, a market without customer captivity. What First Horizon hires in, Pinnacle can hire out — and Pinnacle has said it intends to.

(In fairness: we found no evidence of First Horizon losing teams to Pinnacle or anyone else, and FHN is not named in the trade coverage of Pinnacle’s hiring, which cites Fifth Third, Huntington, Truist, PNC, JPMorgan, Morgan Stanley and Santander as its sources. The risk is structural, not yet realised. Record it as an open question, not a fact.)

Verdict — Competitive Position

No durable competitive advantage. A crowded market with weak differentiation.

Every test fails. Market share is falling in the core state, the core city on a normalised basis, Nashville, Chattanooga and Lafayette. The deposit franchise is structurally commercial, only 23.5% noninterest-bearing, and reprices symmetrically with rates — and the gap to a genuine Deep South franchise is worth roughly $445M a year, about 5 points of ROTCE. Returns sit at the cost of capital rather than above it, so earnings power value approximates asset value. The specialty businesses are, respectively, a well-run commodity (warehouse), a shrinking cyclical with no verifiable market position (FHN Financial), and a set of verticals under direct attack from a $2T private-credit alternative.

What remains is real but modest: operational competence — best-in-peer-set charge-offs, skilful deposit repricing, disciplined credit selection — and a genuine, if flat, retail density in four slow-growing cities. That is a decent bank. Applying CLAUDE.md the risk section’s test directly: if a moat claim cannot be tied to a financial outcome that would deteriorate without it, it is not a moat. Remove First Horizon’s claimed moat and what deteriorates? Almost nothing, because the financial outcome it should produce — a low deposit beta and a premium margin — is already absent.

5. Growth History and Forward Opportunities

The five-year record is close to no growth at all. Loans went from $58,232M (2020) to $64,156M (2025) — +10.2% over five years, about 2.0% a year. Total assets are lower today ($84.4B) than in FY2021 ($89.1B). Average assets in 2Q26 ($84.1B) remain below FY2021’s ($87.6B). Revenue rose from $3,070M (FY2021) to $3,420M (FY2025), a 2.7% compound rate — below nominal GDP. (FACT.)

Composition matters more than the total, and the composition is deteriorating. Within that modest loan growth (FACT — 10-K loan tables):

• Loans to mortgage companies: $5,404M (2020) → $2,019M (2023) → $4,703M (2025) → $4,759M (2Q26) — up 136% in ten quarters and the single largest growth engine. It is a genuinely low-risk book (about one basis point of annual charge-offs over ten years) but it is also low-margin, refi-cyclical and entirely moat-free. Strip it out and 2Q26 period-end loans grew roughly 1% sequentially.
• CRE: $12,275M (2020) → $14,421M (2024 peak) → $13,563M (2025) — deliberately cut 6% off the peak, then restarted in 2026 (+$175M in 2Q26, with new commitments reportedly up more than 50% year over year).
• Consumer real estate installment +28%; credit card −49%; home-equity lines −11%.

Deposits are the weaker half of the picture. Average deposits were flat year over year in FY2025, and 2Q26’s $1.6B of period-end deposit growth was, in management’s own words, “primarily driven by brokered deposit growth,” with average client interest-bearing deposits “roughly flat.” The loan-to-deposit ratio has climbed from 73.3% (2021) to 96.0% (2Q26). (FACT.) A bank funding loan growth with brokered money is not growing its franchise; it is renting a balance sheet.

Forward opportunities, and how to weigh them.

1. The “$100M+ PPNR opportunity.” Management has quantified an incremental pre-provision-net-revenue target built on treasury management, wealth cross-sell (helped by an LPL platform conversion), commercial-real-estate repricing, and repricing underpriced relationships. Two things are worth knowing. First, it is entirely revenue — “there is nothing in there about expenses. That is all deepening relationships and about revenue.” Second, management declines to report progress against it: “that is not intended to signal that we are at any percentage point of completion.” It is embedded in the FY2026 guidance of 3–7% revenue growth. The only external proxy offered is revenue growth outpacing loan growth — net interest income +6% against loans +3%. (FACT — 2Q26 call.) This is the single most important falsification test in the report (the What Must Be True section).
2. Banker hiring. 53 bankers were added in 2Q26, with leadership build-outs in Atlanta, the Gulf States and middle Tennessee. This is the cheaper, lower-risk version of the growth management has hinted at buying through M&A — and, given the segment economics, the more sensible one.
3. Wealth and treasury management. Brokerage, trust and insurance income reached $45M in 2Q26, up $14M year over year. This is the highest-quality growth in the company: capital-light, relationship-locked, genuinely recurring. It is also small.
4. M&A. Discussed in the capital-allocation section and the variant-perception section. Note only that management’s own framing has cooled: by 2Q26 the environment felt “more benign today than it did call it, 12 months ago,” and focus remains on “driving the profitability in this business.”

Verdict — Growth

Low-quality growth, and not much of it. A 2% five-year loan compound growth rate, flat average deposits, a shrinking balance sheet, and a growth engine concentrated in the least-differentiated book the company owns. The genuinely attractive growth — wealth, treasury management — is real but too small to move an $84B balance sheet. The honest characterisation is a mature franchise defending share in slow markets and renting growth in the mortgage-warehouse market, with a revenue-improvement programme that may well be working but which management has structured so that no outsider can verify it.

6. Financial Quality

This is the section where the First Horizon thesis is settled, because the company’s reported numbers and its underlying economics point in different directions — and the gap between them is not an accounting scandal but something more mundane and more useful: the business has not grown in four years, and every dollar of per-share progress has been manufactured by retiring shares.

6.1 The revenue and earnings spine

(FACT — FY2021–FY2025 10-Ks, 1Q26 10-Q, and quarterly earnings releases furnished on Form 8-K. $M unless noted.)

Two numbers in that table do more work than all the others.

Adjusted pre-provision net revenue was $1,374M in FY2022 and $1,372M in FY2025. Four years; no growth. Over the same period the loan book grew roughly $6B and the company hired hundreds of bankers. The 2Q26 annualised figure of $1,456M is +6% against FY2022 — four and a half years for six percent. Pre-provision net revenue is the cleanest available measure of a bank’s underlying earning power precisely because it strips out the credit-cycle judgement; on that measure First Horizon has been running to stand still.

Adjusted diluted EPS of $1.89 in FY2025 is still below the $2.07 earned in FY2021. The headline GAAP progression ($1.36 → $1.87, +37%) is a genuinely misleading number: most of it is the absence of FY2024’s $91M after-tax securities-repositioning loss. On the adjusted series the improvement was $1.55 → $1.89, about 22% — and even that overstates operating progress, because of what produced it.

6.2 What actually produced the earnings growth

Not revenue. Not margin. The share count. (FACT — 10-K/10-Q share data and repurchase disclosures.)

Since 1 January 2024 First Horizon has spent $1,831M retiring 93.7M shares at a blended ~$19.55 under its announced repurchase programmes — approximately 15.1% of today’s $12.1B market capitalisation, bought back in ten quarters. (The cash-flow statement shows a slightly larger $1,879M / 96.3M because it also captures shares withheld for employee taxes and excise tax; the programme figure is the right basis for a capital-allocation judgement, and it reconciles to the company’s own disclosure of “$894 million… at an average price of $21.16 per share including commissions” for 2025.) Period-end shares fell from 559M to 474M, −15.3%. Note the FY2023 line: the count rose to 559M despite the buyback, because TD’s $494M Series G convertible preferred converted into 19,742,776 common shares on 26 June 2023 (the recent-changes section).

The strategic judgement embedded in this table cuts both ways, and honesty requires stating both halves. The buyback has been well executed: a blended $19.51 against a $25.53 share price is a demonstrably profitable use of shareholder money, and 2024’s purchases at 1.24x tangible book were outright cheap. But the accretion is decaying fast. The price paid has gone from 1.24x tangible book to ~1.69x, and at today’s 1.76x a repurchase is roughly neutral-to-dilutive to tangible book value per share while remaining accretive only to earnings per share. A buyback is a capital-allocation decision that gets worse as the stock gets better, and First Horizon is now well past the point where it was obviously creating value. The quarterly pace tells the same story: $360M → $9M → $190M → $335M → $230M → $100M.

6.3 Margin, funding and the one genuinely impressive number

(FACT — 10-K MD&A rate/volume tables and quarterly releases.)

Here is the standout: First Horizon’s deposit down-beta exceeds its up-beta. Through the hiking cycle (4Q21 to the 3Q24 peak, +518bp of Fed funds) the interest-bearing deposit cost rose 333bp — a 64% beta. Through the cutting cycle since (−163bp) it has fallen 116bp — a 71% beta. Repricing deposits down faster than you were forced to reprice them up is unusual, and it is why the margin expanded 12bp in a rate-cutting year. Management’s framing is that this was deliberate: “we were maximizing the decrease in our deposit cost knowing that we give some back once rates stop cutting.” (2Q26 call.)

That is a real operating achievement and it deserves credit. It is also, on the evidence, a treasury achievement rather than a franchise one — it reflects skilful repricing of a rate-sensitive book, not the possession of a rate-insensitive one. The 2Q26 margin already slipped 3bp as time deposits jumped $2.9B to $10.0B; the incremental funding is brokered. And the structural direction of travel is the one the CEO himself describes: “the cost of deposits is drifting slowly towards wholesale cost of funds just with the transparency of interest rates.” (FACT — management quotation; INTERPRETATION — that this is a secular headwind, not a cyclical one.)

6.4 Fee income, and the trouble with FHN Financial

Noninterest income is 23.3% of revenue — bottom-third among peers — and its most distinctive component has not recovered. FHN Financial’s average daily revenue (FACT, $000):

FY2025 ADR is 47% of FY2021’s, and the business has shrunk from 13.2% of revenue to 6.0%. It fell 20% sequentially in 2Q26. Management’s own 1Q26 description of its scorecard — “we have it in red and green, and all but one factor is in red” — is more candid than most sell-side commentary on it. Worse, the segment runs a ~60% compensation ratio, so only about 40% of any incremental revenue reaches pre-provision net revenue. (FACT — management commentary, 2Q26 call.)

The bull framing is that FHN Financial is a countercyclical hedge: the 10-K states the fixed-income business “tends to perform better when rates decline or markets are moderately volatile, which tends to partially offset net interest margin compression.” INTERPRETATION: that hedge is currently degraded, and management has said so. Jordan, 2Q26: “given where fixed income ADRs are today, a rate increase would be incrementally more positive because ADRs have already been at a relatively low level … it would look more like the interest sensitivity that we have described as opposed to the aggregate sensitivity of the balance sheet.” In plain terms — the offset is near the bottom of its range, so the balance sheet’s rate exposure is currently running closer to naked than the “balanced countercyclical model” implies.

6.5 Expenses: the synergies were spent, not banked

The adjusted efficiency ratio is worse in FY2025 (60.0%) than it was in FY2022 (56.6%). The post-IBERIABANK cost synergies were real, and they were handed straight back to reinvestment — outside services and occupancy together rose roughly $94M from FY2023 to FY2025 — while revenue barely moved. FY2026 expenses are guided flat, but this is a one-off: it reflects the completion of a three-year, roughly $100M technology programme. The CFO’s own normalised expense-growth rate is “2.5% to 3%,” which reasserts itself from FY2027. (FACT.)

Verdict on operating leverage: there hasn’t been any. A bank that adds $6B of loans, holds revenue roughly flat in real terms, and lets its efficiency ratio deteriorate 340bp has not demonstrated that its economics improve with scale. Note also that average assets are lower today ($84.1B) than in FY2021 ($87.6B). Pre-provision net revenue as a share of average assets did improve (1.39% → 1.73%), but that improvement is fully explained by redeploying pandemic-era cash into loans as rates rose — the margin went from 2.48% to 3.49% — a cycle effect available to every bank in the country, not evidence of a widening operating margin on a larger base.

6.6 Credit: genuinely the best in the peer set — and the thinnest reserve

(FACT — 10-K credit tables and quarterly releases.)

The credit performance is excellent and should not be grudged. Net charge-offs of 19–20bp are the best in an eight-bank peer set — better than Huntington (23bp), Webster (33), KeyCorp (41), M&T (~41), Citizens (49), Regions (54) and Fifth Third (60). Criticized and classified commercial loans fell 21% during 2025. The mortgage-warehouse book, now the largest single growth engine, has averaged about one basis point of annual net charge-offs over ten years, and management’s explanation of why is specific and credible: “We hold the underlying collateral. We take each note at closing… We pick the closing attorney. We take physical ownership of the actual loan docket… On the fraud piece of it, we have seen collateral double pledged elsewhere. We do not have that situation.” (FACT — 2Q26 call.)

But First Horizon is simultaneously the lowest-reserved and the fastest-releasing bank in its cohort. Provision has been set below net charge-offs in five of the last six quarters, including a literal zero provision in 4Q25. Held at the year-end 2024 coverage ratio of 1.43%, the 2Q26 allowance would be $934M against the actual $808M — a $126M cushion consumed. The earnings contribution is quantifiable: roughly $0.08 of FY2025’s $1.87 and $0.05 of the $1.07 earned in 1H26. And the release is running while charge-offs tick up sequentially (17bp → 18bp → 20bp) and while the commercial-real-estate allowance specifically has been cut from 1.53% to 1.10% in twelve months.

(INTERPRETATION: management’s justification — better credit mix and $2.2B of non-pass resolutions in 2025 — is evidence-backed rather than naked, and I do not regard the reserve as inadequate. But a reserve that is the thinnest in the peer group is a reserve with the least room to absorb a surprise, and the earnings tailwind it has provided is by definition not repeatable.)

What isn’t disclosed matters too. First Horizon publishes no office loan-to-value distribution, no office maturity schedule, no office-specific reserve rate, and no shared-national-credit exposure. Office is $2,694M — 4.1% of loans but ~39% of tangible common equity — and has barely moved in three years. Roughly half is medical office, which is materially more defensive than central-business-district space. The honest characterisation is therefore not “clean” but “unverifiable”: the composition looks good, the disclosure is too thin to confirm it.

6.7 Capital — and the AOCI question that changes the verdict

First Horizon is not a Category IV institution — Category IV standards would apply only above $100B of assets, and FHN is at $84.4B. It is a non-advanced-approaches firm that made a one-time, permanent election to opt out of including accumulated other comprehensive income in regulatory capital. There is no phase-in and no reversal. (FACT — FY2025 10-K.)

That election is doing significant work. Reconciling to year-end 2025 CET1 of $7,760M (10.63%), the capital stack adds back $512M of unrealised AFS losses, $256M of pension/OPEB losses and $42M of cash-flow-hedge losses. Include them and CET1 is 9.92% (AFS only) or 9.52% (all AOCI) — roughly 111bp lower than reported. (FACT — 10-K capital reconciliation; ratio derivation is arithmetic.)

The same adjustment reframes the return. Average tangible common equity in 2Q26 was $6,865M including −$842M of average AOCI. Add it back and return on tangible common equity is approximately 13.5%, not 15.2% — and FY2025’s is roughly 12.4%, not 14.0%. (INTERPRETATION, from filed inputs.) Against a cost of equity of roughly 10–11%, that is a spread of 250–300bp: real, positive, and thin.

One further caution on tangible book. TCE rose from $6,733M to $6,882M during FY2025 — +$149M — of which AOCI pull-to-par contributed +$319M, more than twice the entire increase. In 1H26 tangible common equity was flat at ~$6,888M, and the whole of the reported TBVPS gain came from the share count falling. The “+7% year-over-year tangible book value growth” management cites is therefore not an earnings-retention rate and should not be extrapolated as one.

6.8 Earnings quality, on the other hand, is good

It is worth ending on the credit side of the ledger, because it is substantial and it is unusual.

• The GAAP-to-adjusted gap has collapsed to nothing. Notable items moved from −$0.32 per share (FY2021) to −$0.19 (FY2024) to −$0.02 (FY2025) and $0.00 (1H26). The 10-K presents only four non-GAAP measures — PPNR, ROTCE, TCE/TA and TBVPS — and no “adjusted EPS” at all. Reported earnings are, at present, economic earnings.
• IBERIABANK purchase-accounting accretion is fully run off, last quantified in 4Q22 at +$1M of net interest income. There is no accretion tailwind flattering the margin and no accretion cliff ahead.
• Held-for-sale balances are clean ($406M at YE25, $501M at 2Q26), almost entirely government-guaranteed SBA/USDA paper. No evidence of loans being migrated to held-for-sale to avoid charge-offs.
• The deferred-compensation revenue and expense lines are a near-perfect wash and do not distort.

The one genuine opacity is CECL: the allowance is built on weighted Moody’s baseline/upside/downside scenarios, and the weights are not disclosed — so an outsider cannot separate real credit improvement from an assumption change inside the release.

Verdict — Financial Quality

Structurally average, and the economics do not improve with scale.

The affirmative case is credit and funding discipline: best-in-peer-set charge-offs across a full cycle, a deposit down-beta that exceeds its up-beta, zero remaining purchase accounting, and reported numbers that need almost no adjustment. Those are real and they are why 2Q26 printed a 15.2% headline return on tangible common equity.

The case against is structural and, on balance, weightier. Adjusted pre-provision net revenue has not grown in four years. Adjusted EPS is still below FY2021. The adjusted efficiency ratio has deteriorated. The most differentiated fee business is at 47% of its FY2021 revenue run-rate. The loan growth that exists is concentrated in a zero-loss, low-margin, refi-cyclical mortgage-warehouse book that is up 136% since YE23 — strip it out and 2Q26 period-end loans grew about 1% sequentially. The headline return is flattered by an AOCI-shrunken denominator; on the honest measure it is ~13.5%. And the entire per-share record is a buyback whose accretion is decaying as the multiple rises.

First Horizon earns roughly 250–300bp over its cost of equity on the measure that does not flatter it. That is enough to say the business is not destroying value. It is not enough to call it a compounder.

7. Capital Allocation

First Horizon’s capital-allocation record over six years contains one enormous stroke of luck, one well-handled disaster, one clear and costly error, and an incentive structure that points at the next mistake. It deserves to be worked through in that order.

7.1 The IBERIABANK merger of equals (July 2020): a windfall, not a negotiation

(FACT — FY2020 10-K business-combination note.) Total consideration was $2,502M (roughly 243M FHN shares worth $2,243M, $28M of equity awards and $231M of Series B/C/D preferred) against net assets acquired at fair value of $3,033M — producing a $531M non-taxable bargain-purchase gain and, remarkably, zero goodwill. That gain was $1.23 of 2020’s $1.89 in diluted earnings per share. First Horizon paid 0.82x the fair value of the net assets it acquired.

This is the single best acquisition outcome in the peer group, and honesty requires saying how it happened. The exchange ratio was fixed on 4 November 2019, when the transaction was worth roughly $3.9B. The COVID crash intervened before the July 2020 close, cutting realised consideration by about 36% while IBERIABANK’s net assets held. This was luck transmitted through deal structure. Management deserves credit for using a fixed exchange ratio and for closing rather than renegotiating — real decisions with real risk — but not for foresight.

The execution afterwards was solid: a $200M pre-tax annualised cost-synergy target by 4Q22 against roughly $400–450M of integration cost, about a 2.2x cash payback, with no tangible book dilution and no earn-back period (TBVPS ran $10.02 at YE2019 → $10.23 → $11.00 → $14.53 today). One caveat that the financial-quality section already made: the synergies were spent rather than banked — the adjusted efficiency ratio is worse now than in FY2022. Goodwill has been $1,510M and flat since 2022, all of it from the 2017 Capital Bank deal, legacy positions and $78M from the 2020 purchase of thirty Truist branches. None from IBERIABANK.

7.2 The TD Bank episode: better handled, and worth more, than it looks

The bare facts (FACT — 8-Ks of 28 February 2022, 3 March 2022 and 4 May 2023; FY2023 10-K Note 11):

• TD agreed on 27–28 February 2022 to acquire First Horizon for $25.00 per share in cash — a 37% premium to the $18.25 pre-announcement close — with a ticking fee accruing $0.0017808 per share per day after 27 November 2022.
• TD simultaneously bought 4,935.694 shares of Series G perpetual convertible preferred for $493,569,400. The Series G paid no dividend — its annual rate was disclosed as “N/A.”
• The outside date was extended from 27 February to 27 May 2023. TD walked on 4 May 2023, stating it had no timetable for regulatory approvals “for reasons unrelated to First Horizon” — the OCC and Federal Reserve had stalled over TD’s own anti-money-laundering control deficiencies. TD paid $200M in cash plus the $25M expense reimbursement already due — $225M in total. The stock closed at $10.06, down 33%.
• The Series G was not returned. It converted on 26 June 2023 into 19,742,776 common shares — at $25.00 a share — on a day FHN traded near $10.

That last point is the one most commentary misses, and it runs strongly in shareholders’ favour. Issuing 19.74M shares at the then-market price would have raised roughly $198M; First Horizon had received $493.6M. The Series G therefore transferred roughly $270–296M of value to continuing shareholders. Adding the $225M termination payment, the episode’s total economic benefit was approximately $500–520M pre-tax, or about $0.94 a share — and the conversion moved $493M from Tier 1 into common equity tier 1, which the FY2023 10-K names as the main driver of CET1 rising from 10.17% to 11.40%. (FACT / INTERPRETATION.)

Against that: $50M of the $200M — a quarter of the break fee — was donated to the First Horizon Foundation; $51M of 2023 deal expense was incurred; a $68M FDIC special assessment landed in the same year; and the softer costs were real — twenty-two months with no buyback, a frozen strategy, and a 33% one-day collapse that forced an emergency deposit campaign.

Was taking the cash the better outcome? The honest answer is more favourable to management than the “stock is flat since 2022” framing implies. From the 25 February 2022 close to 24 July 2026, holding FHN produced $18.25 → $29.76 of total-return value, +63.1%, about 11.7% a year — against the KRE regional-bank ETF at +16.3% (3.5% a year) and against $25 of cash rolled in Treasury bills at $27.93–$28.81. Holders who sat through beat both the index and the cash. Cash was superior only if it had been redeployed into regional-bank equity at the 2022–23 lows. (FACT — computed from the AZI adjusted price series.) The offsetting datapoint, from FHN’s own 10-K performance table: from the deal-inflated year-end 2022 mark, three-year total shareholder return was +9.1% against the KRX index’s +20.1%.

7.3 Buybacks: good arithmetic, procyclical timing, decaying accretion

(FACT — 10-K/10-Q repurchase disclosures.) Period-end shares went 558.8M (YE2023) → 474M (2Q26), −15.2%. Against today’s $25.53 the programme carries roughly a $560M mark-to-market gain, and the implied earnings yield of about 9.7% comfortably exceeded the ~4.5% available on parked capital. Execution within each year tracked the volume-weighted average price almost exactly (2024: realised $15.98 against a $16.01 VWAP) — neither skill nor damage in the tape.

The error is what did not happen. First Horizon repurchased nothing in 2023 — the one year the stock traded below tangible book — while carrying the highest CET1 of the period (11.40%). The 2024 proxy concedes it verbatim: “we suspended our stock purchase program… even though our stock price often was below tangible book value per share.” Post-termination 2023 VWAP was $11.81. Deploying 2024’s $605M at that price would have bought 51.2M shares rather than 37.9M — 13.4M more, roughly $341M or $0.72 per share of forgone value. (INTERPRETATION, arithmetic on filed data.)

And the accretion is decaying. The price paid has climbed from 1.28x tangible book to ~1.71x. At today’s 1.76x, a repurchase is roughly neutral-to-dilutive to tangible book value per share and accretive only to earnings per share. The quarterly pace confirms management sees it: $360M → $9M → $190M → $335M → $230M → $100M.

Authorisation history (FACT): $500M (Jan 2021) → $1.0B (Oct 2021) → moratorium on the TD announcement → expired October 2023 with $599M unused → $650M (Jan 2024, retired with $174M unused) → $1.0B (Oct 2024, retired with $180M unused) → $1.2B (27 October 2025, expiring 31 January 2027), with roughly $665M remaining.

7.4 Dividends, preferred, and a financing decision worth flagging

The common dividend was $0.60 a share, flat for six years (2020–2025), then raised 13.3% to $0.17 a quarter in April 2026 — the first increase since 2020, an annualised $0.68 and a ~2.7% forward yield. FY2025 dividends of $314M were a 32.8% payout; with buybacks, total distributions of $1,208M were 126% of net income available to common, funded by running CET1 from 11.20% down to 10.63%. (FACT.)

The preferred stack tells a subtler story. FHN redeemed Series A (2021), Series D (6.10%, $100M, 2024), Series B (6.625%, $80M, 2025) and Series C (6.60%, ~$58M, May 2026) — and then, in March 2026, issued $400M of Series H at 6.75%. (FACT.) The company retired roughly $238M of 6.10–6.625% preferred and replaced it with $400M at a higher coupon — a larger stack costing roughly $27M a year more — while running CET1 at target and buying common stock at 1.71x tangible book. Preferred dividends have already jumped to $10M in 2Q26 from $5M in 1Q26. (INTERPRETATION: the Series H issue funded the common buyback as much as it funded capital. That is a legitimate transaction, but it is leverage, and it should be recognised as such rather than counted as capital strength.)

7.5 Organic investment: the Marathon test, mostly passed

Marathon’s asset-growth anomaly warns against banks that grow the balance sheet into weak returns. First Horizon largely passes: loans grew only 2.0% a year over five years, CRE was deliberately cut 6% from its 2024 peak, credit card is down 49% and home-equity lines down 11%. Assets are smaller than in 2021. This is not an empire-builder.

The qualification is where the growth went: mortgage warehouse from $2,019M (2023) to $4,759M (2Q26), +136% in ten quarters — the least-differentiated book in the company, growing fastest.

7.6 Incentives: the red flag, and the partial offset

The annual bonus is the problem. (FACT — 2026 DEF 14A, quoted verbatim.) FY2025’s plan was “75%: Adjusted Pre-Tax Income (PTI) – target payout at budget performance; threshold at 75% of budget, maximum at 125% of budget” plus “25%: Strategic – non-quantitative assessment of strategic outcomes.” Achieved PTI of $1,308M against a $1,111–1,227M target range scored 112%; strategic scored 108%; the corporate rating was 115%.

A 75%-weighted absolute pre-tax-dollar target, measured against an internally set budget, with no capital or return denominator, pays management for balance-sheet growth regardless of the return earned on it. For a bank about to receive a Basel III capital release of $1.66–$3.07 a share, that is precisely the wrong incentive. Note also the metric churn: pre-tax earnings (2023) → PPNR plus credit ratios (2024) → pre-tax income (2025) — three different primary income metrics in three years.

The 2023 bonus deserves to be recorded plainly. The formulaic outcome was zero. The single metric was adjusted pre-tax earnings; the grid paid nothing below $1,130M and the company achieved $1,028M, 68% of budget. The Compensation Committee then discretionarily added back the $225M TD termination fee — a payment received because a deal failed — and excluded the FDIC special assessment, producing “fully adjusted” pre-tax earnings of $1,319M, then added a further 10%, for a final 85% payout. (FACT.) Without adding back a break fee from a collapsed transaction, no named executive would have received a bonus.

The May 2023 retention reset is the second discretion event. Restricted cash units granted in January 2023 were cancelled and rewritten over the weekend of 5–6 May 2023 — within 48 hours of the termination — with a $10.58 base price, a floor at target, and a kicker of “an extra 5%, up to a maximum of 25% of target, … for each $1 that our stock price, at vesting, exceeds the base price.” Targets: LoCascio $1.5M, Restel $1.0M, Popwell $1.0M, Dmuchowski $0.8M — $4.3M at target, $5.375M at the cap. Jordan was excluded, which is to his credit. The awards vested on 12 May 2026 at the full cap. (FACT.) It is defensible — the collapse was not management’s doing — but it is functionally an underwater-award reset with no downside participation in the value destruction that triggered it, and shareholders were re-based nowhere.

The offset is genuine and should not be ignored. The long-term plan is “ROTCE Rank – target payout at median performance vs KRX index banks over 3-yr period” with a relative-TSR modifier, benchmarked against the roughly fifty banks of the KRX index rather than a curated peer set — a real governance positive, and a harder test than most. Payouts have been earned rather than gifted: the 2020 grant paid 187.5% on a 18.67% ROTCE ranking second of fifty; the 2022 grant paid 128.3%.

Alignment at the top is strong. Jordan holds 1,743,677 shares, roughly $44.5M, about 40x his salary against a 6x guideline, with a career-long 50% net-share retention requirement. (One caveat worth recording: that figure is the proxy beneficial-ownership number at 31 January 2026 and includes 125,786 exercisable options; his Form 4 direct-plus-indirect balance at the same date is roughly 1,078,000. The gap traces to ~1.13M of shares moved to a family trust and to his spouse in 2024–25, where he “remains the beneficial owner” — but no corresponding indirect line was ever added to the Form 4 record. The two disclosures do not reconcile on their face; see the open-questions section.) Say-on-pay support has run 96.8–97.2%. (FACT.) Two blemishes: hedging requires CEO/general-counsel approval rather than being flatly prohibited (none has been granted), no anti-pledging policy is disclosed, and Restel retains a legacy IBERIABANK arrangement guaranteeing $5,766,018 payable whenever his employment ends for any reason, plus roughly $3M of tax gross-ups — which directly contradicts the proxy’s “no tax gross-ups” best-practice claim.

7.7 Insider behaviour: the most telling single table in this report

Every open-market purchase by a First Horizon insider in five years occurred in one four-week window (FACT — all 308 Form 3/4/4A filings, July 2021 – May 2026):

Against that: 128 sales totalling 1,912,641 shares for $42.5M. The dollar ratio of insider selling to open-market buying is 79 : 1. And decisively — no executive officer has bought a single share on the open market in five years. Not the chief executive, not the chief financial officer. Only two of twenty-five sale filings reference a 10b5-1 plan.

The directors who bought at $10.33 in the panic are up 120–150% and deserve the credit. But the pattern overall is unambiguous: the people who run First Horizon receive stock, they do not buy it. Indeed the chief executive’s largest economic gain in the window came from being granted 570,112 shares during 2023 — including a special award struck at $13.41 — at prices set by the collapse he presided over. Because the long-term plan is dollar-denominated, a lower share price mechanically increases the share count granted. Directors and all executive officers together own 0.91% of the company.

Verdict — Capital Allocation

Adequate, not intelligent — and the structure points at the next error.

The credit side is real: a merger of equals that created $531M of value and no goodwill; a failed sale handled well enough to extract roughly half a billion dollars from a walking acquirer; 15% of the share count retired in ten quarters at a blended $19.55 for a ~$560M gain; an unsentimental approach to redeeming expensive preferred; a genuinely low-asset-growth balance sheet; and a chief executive with $44.5M of his own money in the stock.

The debit side is more structural. They bought no stock in the only year it was cheap, with the highest capital of the period, and then scaled up as it got expensive — roughly $341M forgone. They spent the IBERIABANK synergies rather than banking them. They issued $400M of 6.75% preferred to keep buying common at 1.71x tangible book. The bonus plan pays 75% on absolute pre-tax dollars against a self-set budget, and in 2023 a zero formulaic outcome was converted into an 85% payout by adding back a break fee from a deal that failed. The chief financial officer ranks loan growth ahead of buyback in the stated capital priority stack. And the fastest-growing book is the least-differentiated one.

Now put those together with what is arriving: a Basel III capital release worth $1.66–$3.07 a share, a management team that has signalled appetite to acquire, an incentive plan that rewards pre-tax dollars, and a $100M-a-year step-cost at $100B of assets that makes any sub-scale deal the worst possible outcome. That is where the next mistake comes from, and the What Must Be True section makes it a falsification test.

8. Changes and Headwinds — Last Two Years

Strategic and organisational.

• October 2024: segment-president titles eliminated; during 2024 the reportable segments were reorganised from Regional Banking / Specialty Banking / Corporate to Commercial, Consumer & Wealth / Wholesale / Corporate, with prior periods restated. (FACT.)
• 15 October 2025 — the single most consequential event of the period. On the 3Q25 call Jordan said he was “increasingly confident in our ability to integrate a well-structured merger with a strong cultural fit in our existing footprint, if such an opportunity arises in 2026 or beyond.” The stock fell as much as 13% intraday to $19.99 and closed down 9% — its worst session since the TD collapse. Bloomberg Intelligence’s read: “investors had believed the bank could be a takeover target.” Evercore ISI downgraded to In Line. (FACT.)
• 16 January 2026 — the walk-back. CFO Dmuchowski: “We’re not for sale. But we’ve already proven that if an offer comes our way, our board’s going to do the right thing.” She sized any acquisition appetite at the “billion-dollar range, maybe two or three, but not $30 billion” and named Raleigh as a target market with “probably a dozen” private lenders with retiring owners, plus a plan for four or five Raleigh branches within twelve months. (FACT.)
• 15 July 2026: Jordan describes the M&A backdrop as “more benign today than it did call it, 12 months ago.”
• Leadership: Chief Risk Officer (Argo) and Chief Credit Officer (Hung) both took their seats in 2024–January 2025; David Popwell, President–Specialty Banking, retired 31 December 2025; new directors Moehn (Ameren CFO) and Mody (Kinder Morgan) joined in August and October 2025; the board shrinks from 14 to 12 after the April 2026 meeting, with Lead Director Colin Reed — who chaired both the Risk and Executive Committees — departing. Compton is the incoming Lead Director; no successor Risk Committee chair has been disclosed. (FACT.)
• No CEO succession has been disclosed anywhere. Jordan is 64, has been chief executive since 2008, and his contract runs to 3 August 2028 — at which point all non-performance awards accelerate and service requirements on performance awards are waived. (FACT — 2026 proxy.)

Competitive. The Southeast consolidated around First Horizon in a single quarter. Pinnacle/Synovus closed 1 January 2026 ($8.6B, $117.2B pro forma); Fifth Third/Comerica closed 1 February 2026 ($10.9B); Huntington/Cadence closed 1 February 2026 ($7.4B, adding 118 Texas branches). Santander/Webster — announced 3 February 2026 at $12.3B and roughly 2.0x tangible book for an ~$84B bank — remains pending, expected to close in the second half of 2026. 2025 produced 181 bank M&A announcements, the most since 2021, and median approval-to-close fell to about four months. (FACT.)

Capital and balance sheet. The $1.2B buyback authorisation (October 2025); the first dividend increase in six years (+13.3%, April 2026); $400M of 6.75% Series H preferred (March 2026); $500M of 5.514% senior notes (March 2025); Series B and C preferred redeemed.

Headwinds, stated plainly.

1. The rate regime turned in June 2026 — the Fed on hold with the dot plot revised up and nine of nineteen participants now expecting a hike. The falling-deposit-cost tailwind that produced FY2025’s margin expansion has stopped; FHN’s interest-bearing deposit cost already rose 5bp sequentially in 2Q26.
2. FHN Financial is weakening — average daily revenue down 20% sequentially to $594K, with management guiding the second half of 2026 to underperform the second half of 2025.
3. Reserve coverage is the thinnest in the peer set and has been released for three consecutive years while charge-offs tick up sequentially.
4. Deposit competition is intensifying and fragmenting — Dmuchowski describes an industry that has moved to “different rate specials in different cities” with tiered pricing by size.
5. Private credit and the NDFI channel continue to take the middle-market business FHN underwrites, while 86% of the offsetting bank-lending-to-funds pool accrues to banks above $100B.

Verdict — Changes and Headwinds

On balance these weaken the thesis. The operating changes are sensible and incremental — banker hiring, market build-outs, branch rationalisation, a first dividend increase. But the three that matter most are all negative: the rate cycle turned against the earnings model; the competitive set consolidated into three much larger balance sheets inside two months; and management reset the market’s expectation of a takeover from “likely seller” to “possible buyer,” which the market priced immediately at −9% and has not fully given back. The one genuinely favourable change — the regulatory relief package and the prospective Basel III capital release — is a one-time transfer rather than a structural improvement, and it arrives at a company whose incentive plan rewards deploying it into loans.

9. Risk Analysis

The two risks that would actually break the thesis are #1 and #2 together — a credit turn arriving at the moment the company has the thinnest reserve in its peer group and a modified-CRE book roughly twice the size of its disclosed CRE non-performers. Everything else is a matter of degree.

10. Valuation Discussion — Embedded Expectations

(No recommendation and no price target appear in this section. What follows is embedded-expectations analysis and scenarios.)

10.1 The right toolkit, and why the enterprise-value family is inapplicable

At $25.53 (24 July 2026) against 2Q26 tangible book value per share of $14.53, First Horizon trades at P/TBV 1.757x; P/B 1.426x on BVPS of $17.90; P/E of about 11.9x on annualised 1H26 earnings of $2.14; a 2.66% forward dividend yield ($0.68 annualised). Market capitalisation is roughly $12.09B. (FACT.)

EV/EBITDA, EV/Sales and P/FCF are not applicable to a deposit-funded lender, and it is worth saying why rather than simply omitting them. For a bank, borrowed money is raw material, not financing — adding $75B of deposits and borrowings to a $12.09B equity value produces a number with no economic referent. The ROIC.ai aggregator demonstrates the point by printing FHN’s enterprise value as −$13.54B in FY2021 and −$0.53B in FY2023; that is a category error, not a data error. EBITDA is empty here because interest expense is the principal cost of goods sold — FY2025’s entire margin story was a 57bp fall in the cost of interest-bearing liabilities against a 26bp fall in asset yields; add interest expense back and the business disappears. And free cash flow fails because a bank can manufacture enormous “free cash flow” by shrinking its loan book, which destroys value; the binding constraint on distributions is regulatory capital, not cash.

The correct lenses are therefore P/TBV (balance sheet), P/E on normalised earnings (earnings), ROTCE (the return that ties them together through the residual-income identity P/TBV = (ROTCE − g) / (COE − g)), and total payout yield (distribution), with CET1, ACL/loans and the charge-off rate as risk adjusters. A bank earning exactly its cost of equity trades at 1.0x tangible book by construction; everything above 1.0x is a claim about excess returns.

One genuinely attractive headline: the total payout yield is 6.6–7.9% of market capitalisation — a $300M dividend plus a $500–660M annualised buyback pace. That is real cash, not accounting.

10.2 Own-history context: above every year-end of the last nine except the deal year

AZI’s own-history percentiles put FHN at the 89.6th percentile on price-to-book and the 81.4th on the composite of its own decade. The year-end price-to-tangible-book series is more revealing still (FACT — ROIC.ai series; read the series, not the level, since its tangible-book basis runs ~4% above the filing’s): 2014 1.38x · 2015 1.49x · 2016 1.95x · 2017 2.24x · 2018 1.35x · 2019 1.51x · 2020 1.19x · 2021 1.41x · 2022 2.27x · 2023 1.12x · 2024 1.50x · 2025 1.61x.

Today’s 1.76x is above every year-end reading of the last nine except 2022 — and the 2022 reading was not a fundamental multiple, it was the TD deal price. The only genuinely comparable standalone peak is 2017’s, struck on a materially smaller, differently-mixed, pre-Capital-Bank and pre-IBERIABANK company. (INTERPRETATION: First Horizon has never sustained this multiple on standalone fundamentals.)

10.3 The peer cross-section — and why it gives the wrong answer on its own

(FACT — prices 24 July 2026; P/TBV computed on a single consistent basis from 1Q26 balance sheets; fundamentals from published peer filings where available, otherwise author-derived estimates marked “e”.)

† multiple distorted by a deal closed January/February 2026 (goodwill created shrinks tangible book while the price reflects combined earnings). ‡ Webster trades at Santander deal value, not fundamental value.

Regressing price-to-tangible-book on return on tangible common equity across nineteen non-deal-distorted names gives P/TBV = 0.7154 + 0.0747 × ROTCE(%), R² = 0.322 — each incremental point of ROTCE is worth about 0.075x of tangible book. First Horizon’s fitted value at its estimated 14.3% trailing ROTCE is 1.78x against 1.78x actual: a residual of −0.2%. FHN sits exactly on the line. (FACT — author-derived.)

(In fairness, a parallel exercise using nine hand-picked names fitted a steeper line — 0.632 + 0.0964 × ROTCE, R² 0.62 — on which FHN would look about 12% cheap. The two disagree because they use different P/TBV and efficiency conventions. We report ours as primary because it is struck on one date with one methodology and reconciles on the target; we report theirs because it moves the answer. Note also that an R² of 0.32 is itself a finding: ROTCE explains only a third of the cross-sectional variation in regional-bank price-to-book, so the regression is a weak instrument and should not carry the argument alone.)

Here is the point that matters most in this section, and it is easy to miss. The cross-sectional test and the absolute test give opposite answers, and the reason they diverge is the whole finding. Cross-sectionally, FHN is priced exactly where its reported return says it should be — no premium, no discount. But a regression run inside a cohort that shares an unsustainable input cannot detect that the input is unsustainable. Every bank in that table is reporting returns flattered by the same benign credit environment and the same reserve drawdown. First Horizon is not expensive relative to its peers; FHN and its peers are together priced for a credit environment with no historical precedent for persistence. That is a beta statement rather than an alpha one — and it fits the factor evidence exactly: FHN’s statistical similarity to KRE, the regional-bank ETF, is 0.945, higher than to all but two individual banks. First Horizon essentially is the index, and it will de-rate with it.

10.4 The justified multiple — and the gap that survives every adjustment

Using P/TBV = (ROTCE − g) / (COE − g) (FACT — grid computation):

Solving for the return the current price requires, at a 10.5% cost of equity and 3% growth: 16.18%. Across the plausible grid the requirement runs 15.3% to 17.9%; only at the single most generous corner (COE 10.0%, g 4.0%) does it fall to 14.54%, roughly FY2025’s reported figure. Everywhere else the market requires First Horizon to sustain a return it has never sustained through a cycle — reported ROTCE was 14.10% (FY2023), 10.99% (FY2024) and 14.01% (FY2025), a three-year average of 13.03%.

The AOCI objection is a wash, and this is worth stating because it is the sensitivity most likely to be argued over. Adjusting for the AOCI-shrunken equity base lowers the return and raises the book — both numerator and denominator move together. Ex-AOCI tangible book per share is $16.30, so the ex-AOCI multiple is 1.566x, not 1.757x, and the required return falls to 14.74%:

The gap between what the price requires and what a credit-normalised First Horizon delivers is roughly 3.15 points of return on tangible common equity — and it is robust to the accounting convention. Any analysis that haircuts the return for AOCI without also lifting the tangible book is simply wrong; done properly, the adjustment neither rescues the multiple nor condemns it.

Why 2Q26’s 15.2% is not the run-rate — three independent flatterers. (FACT.) First, reserve release: removing the 2Q26 quarter’s release alone takes reported ROTCE from 15.15% to 13.98%. Second, a 20bp charge-off rate against FHN’s own 0.28% in FY2023; normalising to a mid-cycle 35bp and provisioning for reserve build on 4% loan growth costs roughly $0.25 of annual EPS, taking normalised earnings to about $1.89 and normalised ROTCE to about 13.0%. Third, First Horizon carries the lowest ACL/loans in its peer set (1.24%) while reporting the best charge-off rate — peers run RF 1.68%, KEY 1.63%, MTB 1.53%, CFG 1.52%. Thinnest armour plus best-in-class current credit is precisely the combination that produces the largest negative surprise when the cycle turns; the chief credit officer’s own flag on trucking, auto and restaurants is the disconfirming evidence sitting inside the good number.

10.5 What must be true at $25.53

At a 10.5% cost of equity and 3% growth, the price requires a sustained ~16.2% ROTCE, which on a tangible common equity base of $6,896M compounding at ~4% implies roughly $1.12–1.20B of annual net income available to common — about $2.35–2.50 of earnings per share on a share count falling ~3.5% a year. That is against a $2.14 annualised 1H26 run-rate and a ~$1.89 credit-normalised figure. Decomposed, the market is underwriting all of the following simultaneously:

1. Net interest margin holds in the guided “mid to high 3.40s” — i.e. FY2025’s 57bp fall in funding cost proves durable rather than cyclical, in a regime where the Fed’s own dot plot now leans toward a hike.
2. Loan growth of 4–6% converts from the >50% year-over-year rise in new CRE commitments and the 53 bankers hired in 2Q26, on a book that grew 3%.
3. Revenue growth of 3–7% is delivered and flows through, with expenses flat — positive operating leverage that finally moves the efficiency ratio out of the 60s.
4. Credit costs stay at 20bp and the allowance is never rebuilt.
5. The buyback continues at $450–660M a year — requiring either the ~10% Basel III risk-weighted-asset relief or a willingness to run CET1 at the floor.

Break any one and the required 16.2% is not reached.

And here is the decisive disconfirming evidence: adjusted pre-provision net revenue was $1,374M (FY2022), $1,370M (FY2023), $1,299M (FY2024), $1,372M (FY2025) — approximately zero growth over four years on a balance sheet that added $6B of loans — while the adjusted efficiency ratio deteriorated from 56.6% to 60.0%. The base-case operating-leverage assumption embedded at $25.53 is not an extrapolation of the record; it is a break from it.

What the market is underwriting correctly, and it is a real list: a genuine Sunbelt franchise with a 3.47% margin above KeyCorp (2.82%), Citizens (2.97%), Truist (3.03%), Fifth Third (3.11%) and Huntington (3.13%); a credible revenue setup; large-scale, disciplined capital return (15.3% of the share count retired since YE2023, a 6.6–7.9% total payout yield); best-in-cohort current asset quality at 19–20bp against peers at 23–60bp; and a prospective Basel III capital release that is genuinely underappreciated.

What it may be underwriting incorrectly: a mid-teens return that is rented rather than owned; a 60%-plus efficiency ratio treated as fixable when it has moved the wrong way for four years; Wholesale treated as diversification rather than as a low-margin cyclical rental; a takeout option treated as free; and a buyback treated as perpetually accretive when at 1.76x tangible book every $100M of repurchase destroys roughly $43M of tangible book to buy $8.4M of annual earnings — an 8.4% earnings yield against a 10.5% cost of equity.

10.6 The M&A option, quantified — and it is far smaller than the narrative implies

This is the distinctive feature of First Horizon’s valuation and it deserves an explicit number rather than a hand-wave.

The decisive fact: First Horizon already trades inside — and below the midpoint of — this cycle’s observed deal multiples. The 2025–26 record (FACT — primary deal documents):

Three things follow, and they are more useful than the folklore.

First, the clearing range is known, and First Horizon is already in the middle of it. Stripping the two special cases — FirstSun/First Foundation at 0.65x (a loss-making, distressed target) and Columbia/Pacific Premier at 0.99x (an in-market consolidation) — the six healthy $10B+ franchises cleared at a median and mean of ~1.77x tangible book, in a 1.52x–2.00x range. FHN trades at 1.757x. The remaining takeout premium is real but thin; it is not the large uncaptured option the narrative implies.

Second, high multiples paid do not mean rich premiums — the targets were already expensive. One-day premiums across this set ran 9.0%–23.5%, median 17.2%, against KBW’s own disclosed precedent benchmarks of a 1.40x median P/TBV and an 11.0% median premium. Huntington paid 1.74x for Cadence at only a 9.0% premium because Cadence already traded at 1.60x tangible book. This cycle is being struck far above the historical top quartile on price-to-book while paying roughly historical premiums — nothing like the 37% TD paid FHN in 2022. Applying a median 17% premium to a stock already at 1.757x implies roughly 2.05x, i.e. the very top of the observed range.

Third — and this is the part that bears directly on management’s stated appetite — the acquirers were punished, almost without exception. Eight of nine fell on announcement; the median reaction was −2.68% and the mean −5.10%. The only rewarded acquirer was Columbia, which paid below tangible book. Acquirer day-one return moves inversely with both the multiple paid and the tangible-book dilution accepted: the two worst reactions (Pinnacle −12.10%, FirstSun −16.54%) were the two deals with the most dilution, and the only tangible-book-accretive deal (Fifth Third) drew one of the two mildest. A year on, Pinnacle holders are −10.8% since the undisturbed date and roughly 30 points behind the regional-bank index — despite the deal tracking or beating its announced plan. (FACT.) That is the asset-growth anomaly showing up cleanly, and it is the best available evidence for what the market would do to First Horizon’s shares if management acted.

(Worth noting for completeness: an EDGAR full-text enumeration of S-4 and 425 filings by bank registrants across this window found no qualifying 2025–26 transaction involving First Horizon in any capacity — neither as acquirer nor as target.)

Implied takeout values against this record: 1.73x = $25.14, 1.5% below the current price; 1.77x (the healthy-franchise median) = $25.72 (+0.7%); 1.86x = $27.03 (+5.9%); 2.00x = $29.06 (+13.8%); 2.20x = $31.97 (+25.2%).

Taking 1.90–2.10x as a realistic ceiling, the gross premium available is $2.08–$4.98 a share. At a post-October-2025 implied probability of 10–30%, the residual embedded option is worth roughly $0.21–$1.49 a share, or 0.8%–5.9% of the price. Against that, the Gordon residual — the gap between today’s 1.757x and the ~1.33x a credit-normalised 13.0% return justifies — is $6.14 a share, 24.0% of the price.

And note the reflexive trap in the acquirer case. The tangible-book breakeven is exactly First Horizon’s own trading multiple because that multiple is elevated; if the market marks the shares down 3–12% on announcement — as it did to eight of nine acquirers above — the currency that made the deal work is impaired the moment the deal is struck.

The M&A option is therefore four to ten times too small to explain the premium. The remaining $4.50–5.50 a share is not idiosyncratic takeout value; it is sector re-rating that First Horizon inherits by being the index. It will de-rate with the sector, and no takeover hope will cushion it.

The acquirer case, modelled. For an all-stock purchase of a $15B-asset in-footprint bank ($1.5B tangible book, $150M of earnings, $300M of expenses), against a standalone EPS baseline of $2.19 (ASSUMPTION-based model on FACT inputs):

The tangible-book breakeven is exactly First Horizon’s own trading multiple — which is precisely why the chief executive wants to buy now: his currency sits at the 89.6th percentile of its own decade. But the deal is only EPS-neutral-to-accretive if roughly 30% cost saves are realised; without them, a 1.73–1.90x transaction is 5–7% dilutive to earnings. The entire acquirer case rests on synergy execution — from a management team whose last transformational deal produced zero adjusted-PPNR growth in four years and a 3.4-point deterioration in the adjusted efficiency ratio. In Marathon’s terms, a wave of large in-market bank mergers struck at the top of a valuation cycle with elevated acquirer currencies is the textbook condition under which acquirer returns disappoint.

10.7 Sum-of-the-parts: a quality-of-earnings finding, not a large adjustment

Wholesale is 13.9% of gross segment revenue but only 9.3% of pre-corporate earnings, running a 34.6% PPNR margin against Commercial, Consumer & Wealth’s 52.4%. Valuing Wholesale at a cyclical broker-dealer 8.5x rather than the blended 12.1x costs roughly $286M, about $0.60 a share, 2.4% of market capitalisation.

Even a full return of FHN Financial to its FY2021 boom at $1,436K of average daily revenue would add only about $0.14 of EPS (~6.5%), because the 60% compensation ratio caps the drop-through. It is not a call option worth paying for. The directional point is what matters: a bank with 14% of revenue in a cyclical, comp-heavy trading business, mortgage-warehouse lending and correspondent banking should trade at a discount to a pure-play commercial bank of the same return — not a premium.

10.8 Scenarios — two and a half years (2H26 → FY2028)

(All figures are ASSUMPTIONS except the 2Q26 starting point. Ranges and zones, never targets. Every tangible-book path includes AOCI pull-to-par of ~$525M cumulative, ~$1.15 a share — mechanical balance-sheet normalisation, not franchise compounding.)

Bear assumes the margin compresses to 3.30% as the deposit beta proves asymmetric, loans grow 1%, charge-offs reach 55bp as the flagged sectors deteriorate, the allowance is rebuilt toward 1.45%, expenses do not flex (the 53 banker hires are fixed cost), and the buyback halts from FY27. Trend simply extrapolates the realised record — flat adjusted PPNR, efficiency stuck near 60%, charge-offs normalising to 30bp, buyback continuing at ~$425–450M. Base delivers guidance: revenue +4–5%, expenses +3.5%, efficiency finally reaching ~58%, charge-offs ~32bp. Bull requires a 54% efficiency ratio the company has never achieved and a 16.8% return it has never sustained.

The distribution is asymmetric to the downside relative to the base case’s own probability. The Base case requires a break from a four-year flat-earnings record. The Trend case — which requires nothing but continuation — produces approximately zero total return over two and a half years. That is the honest centre of gravity. Meanwhile the Bear requires only that credit normalise to a still-benign 55bp while revenue stalls.

Sensitivities, ranked: the charge-off rate dominates (each 10bp ≈ $0.11 of EPS, ~0.75 points of ROTCE, ~0.10x of justified P/TBV); then the efficiency ratio (each 100bp ≈ $36M of PPNR ≈ $0.06 of EPS); then NIM (each 5bp ≈ $38M ≈ $0.06). FHN Financial’s average daily revenue, by contrast, is worth only ±$0.03–0.05.

(Recorded as third-party market-expectations datapoints only, never adopted as ours. Sell-side consensus sits at FY2026 adjusted EPS of $2.13–$2.15 and FY2027 of $2.31–$2.34, with revisions skewing negative, against mean published price targets of $26.89–$28.11 across the major aggregators (high $29–30, low $24–26) and a hold-majority rating distribution — S&P Global counts 5 Strong Buy / 3 Buy / 11 Hold. Named recent moves span BofA and RBC at $30 down to Wells Fargo at $26 and JPMorgan trimming to $27. Note also that FHN has now beaten consensus for four consecutive quarters but by a compressing margin: +13.3% → +13.0% → +8.2% → +3.9%. This article publishes no price target and adopts none of these.)

Verdict — Valuation

At 1.757x tangible book the market requires First Horizon to sustain roughly a 16.2% return on tangible common equity against a credit-normalised 13.0% — a gap of about 3.15 points that survives every accounting adjustment, including the AOCI argument that is usually deployed to close it. The company sits exactly on the peer regression line, but the whole cohort is priced on returns flattered by the same benign credit and the same reserve drawdown. The takeout option that many investors believe they are getting for free is worth at most 0.8–5.9% of the price and is dwarfed by a 24% Gordon residual — and at 1.757x the stock already sits at the median of what actual acquirers paid for healthy franchises this cycle (~1.77x, range 1.52x–2.00x) rather than at a discount to it. The buyback that produced the last three years of per-share progress has become value-neutral at this multiple. And the case that requires nothing to go wrong — simple continuation of the four-year record — returns approximately zero.

11. Variant Perception

Consensus. The prevailing view is that First Horizon is a cleanly-run Southeast regional with best-in-class credit, a repaired balance sheet, an aggressive and well-executed buyback, and a residual takeover premium — a reasonable, if unexciting, way to own Sunbelt banking beta at an in-line multiple. Sell-side positioning is genuinely split, which is itself informative: DA Davidson calls it “the top takeover target for banks above $75B in assets,” yet the rating distribution is a hold majority (S&P Global: 5 Strong Buy, 3 Buy, 11 Hold) with published targets spanning $24 to $30 and estimate revisions skewing negative. (FACT — third-party expectations, recorded not adopted.)

The strongest bull case, stated as well as it can be. First Horizon has the best charge-off rate in its peer group (19–20bp against peers at 23–60bp), a margin above five of the eight covered comparables, a deposit franchise that repriced down faster than it repriced up, and no remaining purchase-accounting or adjusted-EPS games — reported earnings are economic earnings, which is rarer than it should be. Its office exposure, the sector’s great fear, is only ~4% of loans and roughly half medical office. It has retired 15.3% of its shares in ten quarters at a blended $19.55 and returns 6.6–7.9% of its market capitalisation annually. A ~10% Basel III risk-weighted-asset relief is arriving, worth $1.66–$3.07 a share of capital. Management has a $100M-plus revenue self-help programme, 53 new bankers, and three large competitors — Fifth Third/Comerica, Huntington/Cadence, Pinnacle/Synovus — all simultaneously distracted by integrations in its own footprint. If credit holds and operating leverage finally appears, the Base case returns ~23% and the Bull ~60%.

The strongest bear case. Earnings power has not grown in four years. Adjusted pre-provision net revenue was $1,374M in FY2022 and $1,372M in FY2025; adjusted EPS of $1.89 is still below FY2021’s $2.07; the adjusted efficiency ratio went backwards from 56.6% to 60.0%. All per-share progress is a buyback that has become value-neutral at 1.76x tangible book. The reported 15.2% return becomes 13.98% ex-reserve-release and ~13.0% credit-normalised against a price requiring 16.2%. The company carries the thinnest allowance in its peer set while releasing reserves for a third consecutive year, and $283M of modified CRE — 2.09% of the book — sits outside the non-performing numbers, roughly doubling the true CRE problem book. The deposit franchise is not a franchise: 23.5% noninterest-bearing, 58% commercial, symmetric ~63%-up/66%-down betas, and a cost 68bp worse than Regions’ in the same states — worth about $445M a year, or 5 points of ROTCE. Deposit share is falling in Tennessee, Nashville, Chattanooga and Lafayette. And the rate regime turned in June.

The three to five assumptions that actually matter.

1. Does credit normalise, and when? Everything else is second-order. Each 10bp of charge-offs is $0.11 of EPS and 0.10x of justified multiple. The bull needs 20–25bp to persist; the bear needs only 55bp — still historically benign.
2. Does operating leverage ever appear? Four years of flat PPNR and a deteriorating efficiency ratio say no; management’s $100M+ programme and flat FY2026 expense guide say yes. This is the single cleanest disagreement in the name.
3. Is the deposit beta symmetry cyclical or structural? If the CEO is right that deposit costs are “drifting slowly towards wholesale cost of funds,” the margin has a permanent ceiling.
4. Does management buy something? The tangible-book breakeven is exactly FHN’s own multiple, the incentive plan pays on absolute pre-tax dollars, and a capital release is arriving. The market told them what it thinks on 15 October 2025.
5. Does the sector hold its multiple? With a 0.945 factor similarity to KRE, this is mostly not FHN’s decision.

The positioning read (FACT — FactorsToday, 756-day window). First Horizon has no measurable momentum loading — Momentum is L1-zeroed in all four nested models. Its dominant exposures are Regional Banks +1.22, Market +1.04, DividendYield +0.97, Value +0.40, Quality −0.18, LowVolatility −0.23. Idiosyncratic volatility is 14.4% against an R² of 0.77 — roughly 77% of the variance is factor-explained. Relative strength is unremarkable (rs_12m +16.3, rs_6m +3.8). The risk-adjusted record is sobering: the three-year Sharpe of 0.90 is measured off the May-2023 termination trough, while the ten-year Sharpe is 0.19 and the lifetime Sharpe is negative (−0.029) on an −87.7% lifetime drawdown.

What that means for where consensus may be offsides. This is not a crowded momentum trade and not a falling knife — it is a high-dividend-yield, value-tilted regional-bank beta in a quiet, orderly uptrend, whose two dominant style factors (DividendYield z +1.72, Value z +1.57) are the two best-performing style factors in the market right now. The tape is a mild tailwind, not a signal. The variant perception is therefore not that the market has mispriced First Horizon relative to its peers — it hasn’t; FHN sits exactly on the regression line. It is that the market is mispricing the peer group’s normalised earning power, and First Horizon, being statistically almost indistinguishable from the index, carries the full weight of that error while offering less earnings quality than most of the cohort to cushion it.

Falsifying evidence. The bull case is falsified if charge-offs exceed 35bp for two consecutive quarters while the allowance is rebuilt, or if FY2027 adjusted PPNR again fails to grow. The bear case is falsified if the adjusted efficiency ratio prints below 57% for two consecutive quarters with revenue growth exceeding loan growth — the one externally observable proxy management has offered for the $100M+ programme — or if charge-offs remain below 25bp through a full year of rising rates.

12. Fact vs. Interpretation

13. Open Questions

1. How much of the ~$9B C&I exposure to financial-services borrowers is self-liquidating warehouse versus term lending to non-bank lenders? Management’s 2Q26 decomposition (NDFI $8.6B, ~55% warehouse with sub-20-day dwell, private credit under 1% of loans) is reassuring but is call commentary, not disclosure. The 10-K collapses both into one bucket.
2. What are the office book’s loan-to-value distribution, maturity schedule and specific reserve rate? None is disclosed, while the CRE allowance was cut from 1.53% to 1.10% in twelve months. Record this as unverifiable, not as clean.
3. What is First Horizon’s shared-national-credit exposure? Not disclosed anywhere in the corpus.
4. What are the CECL scenario weights? Disclosed sensitivities are wide — 100% baseline would cut the allowance 7%; 100% adverse would raise it 31% (~$230M pre-tax, ~$0.35 a share) — but the weighting itself is undisclosed, so an outsider cannot separate genuine credit improvement from an assumption change.
5. Who succeeds Bryan Jordan, and when? He is 64, has been chief executive for eighteen years, holds the Chairman, President and CEO titles simultaneously, and his contract expires 3 August 2028 — at which point all non-performance awards accelerate. No successor, bench or process appears anywhere in the 2026 proxy.
6. Who chairs the Risk Committee after Colin Reed’s departure? He chaired both Risk and Executive; the incoming Lead Director has been named, the Risk chair has not.
7. Is FHN Financial actually a top-tier dealer to depositories? The company makes no market-position claim in its own filings, and no independent league table validating one could be located.
8. Has First Horizon lost banker teams to Pinnacle or the other acquirers? No evidence was found either way, and FHN is not named in trade coverage of Pinnacle’s hiring. Structural risk, unrealised.
9. What is the FY2026 buyback pace? The largest single swing factor in the scenarios, and dependent on whether the ~10% Basel III RWA relief is finalised as proposed.
10. Does the reserve-release cycle end in 2026? The CFO said in 4Q25 “we’re done in that building phase,” but the allowance has fallen for three consecutive years and again in 2Q26.
11. What is First Horizon’s commercial-real-estate concentration as a percentage of total risk-based capital, and its construction/land-development ratio against the 100% supervisory criterion? Neither is disclosed. The $10–100B cohort median is 289%, and our own estimate puts FHN well below it — but it is derived, not reported, and it is the standard supervisory lens on a $13.6B CRE book.
12. Why does the CEO’s proxy beneficial-ownership figure (1,743,677 shares) not reconcile to his Form 4 balance (~1,078,000) at the same date? The gap traces to 2024–25 estate-planning transfers where he remains the beneficial owner, but no offsetting indirect-ownership line was ever added to the Form 4 record.

14. What Must Be True

The bull case requires:

The bear case requires:

15. Source Appendix

The full source appendix — every primary filing, transcript, dataset and third-party source relied on, with URLs and access dates — is provided as Appendix B to this report.

Independent fundamental research. No recommendation and no price target appears anywhere in the analysis above; the single labelled exception is the opinion block at the head of this article. This is general information and not investment advice.

APPENDIX A — Standard Diligence Questionnaire

First Horizon Corporation (NYSE: FHN) — Report date 2026-07-26

A supplemental diligence questionnaire covering the standard checklist an investor would work through on this business. Labels: FACT / INTERPRETATION / ASSUMPTION.

General

What thoughtful questions have other investors asked about this company?

The question set that dominates FHN’s earnings calls and the sell-side commentary is unusually well-defined, and it is worth stating plainly because it frames everything below.

1. “Are you a buyer or a seller?” This is the first question about First Horizon, not the fifth. TD Bank Group agreed to buy the company for $25.00 a share in cash in February 2022 and walked away in May 2023 for reasons entirely to do with TD’s own anti-money-laundering supervisory problems. Ever since, a portion of FHN’s valuation has been an embedded takeover option. On the 3Q25 call (15 October 2025) CEO Bryan Jordan said he was “increasingly confident in our ability to integrate a well-structured merger with a strong cultural fit in our existing footprint if such an opportunity arises in 2026 or beyond” — and the stock fell roughly 9%, its worst session since the TD collapse. He spent the following two quarters walking it back. FACT.
2. “When does the reserve release stop?” The allowance for credit losses has fallen from 1.45% of loans to 1.24%, and provision has run below net charge-offs in five of the last six quarters — including a literal zero provision in 4Q25. Analysts have repeatedly pressed on how much of the earnings trajectory this represents. FACT.
3. “Where do deposit costs go from here?” Multiple analysts on the 2Q26 call pressed on the third- and fourth-quarter funding-cost outlook and the sustainability of the 66% cumulative downside deposit beta. FACT.
4. “What is FHN Financial actually worth?” The fixed-income broker-dealer’s average daily revenue has fallen from $771K (3Q25) to $594K (2Q26), and management conceded that on its own scorecard “all but one factor is in red.” FACT.
5. “What is the $100M+ PPNR opportunity, and how would we know it was working?” Management has quantified an incremental pre-provision-net-revenue opportunity but declines to report progress against it. Jordan, 2Q26: “that is not intended to signal that we are at any percentage point of completion.” FACT.
6. “Does crossing $100B of assets matter?” At ~$84B, FHN is roughly $16B from the Category IV threshold. Jordan’s own framing is that “the bright line … seems to be a little less bright.” FACT.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? INTERPRETATION: closer to a cyclical high than a low, on three independent counts. (i) Credit is at an unsustainably benign point — net charge-offs of 19–20bp are roughly half a normalised through-cycle regional-bank loss rate, and the reserve ratio is being drawn down on top of that. (ii) The net interest margin of 3.47% (FY2025) is being earned in a positively-sloped-curve environment after a 42bp year-over-year fall in deposit costs; management guides to “mid to high 3.40s,” i.e. no further expansion. (iii) Against that, one component is at a cyclical low: FHN Financial’s ADR at $594K is near the bottom of its range, and management has said the back half of 2026 “is not gonna be as strong as the back half of last year.” Net: the spread and credit engines are running hot; the capital-markets engine is running cold; the aggregate is above mid-cycle.

Driven by the external environment or internal actions? INTERPRETATION: predominantly external. The FY2024→FY2025 EPS move from $1.36 to $1.87 looks like a 37% internal improvement but is substantially the absence of the prior year’s $91M after-tax AFS-restructuring loss; on an adjusted basis the improvement was from roughly $1.55 to $1.89, about 22%. Of that, the largest identifiable drivers were a 42bp fall in deposit costs (external — the Fed) and a $55M pre-tax reserve release (an accounting judgement, not an operating result). The genuinely internal contributions — the mortgage-warehouse market-share gain of ~$1.2B, the wealth cross-sell, the banker hiring — are real but second-order. FACT for the components; INTERPRETATION for the attribution.

How stable are revenues? Moderately. ~77% of revenue is net interest income, which is stable in the sense that it is contractual but rate-sensitive in the sense that over half the loan portfolio is indexed to short-term rates (FY2025 10-K). The 23% fee component is bifurcated: treasury management, wealth and service charges are genuinely recurring; the fixed-income business is not — management has described weekly ADR swinging from “a little over $700,000 … the next week we’re a little over $450,000.” FACT.

Outlook for products/services; how big will this market be? Commercial and consumer banking in the Southeast/Gulf-South/Texas footprint is a mature, GDP-plus market — growing with Sunbelt population and job migration in Nashville, Atlanta, Florida, the Carolinas and Texas, and stagnant in Memphis and parts of Louisiana. It is domestic; the international banking line is a service to domestic clients, not a geographic expansion. FY2026 guidance is 3–7% revenue growth. FACT (guidance) / INTERPRETATION (market characterisation).

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. Three vectors: the national banks’ balance-sheet and technology scale in FHN’s own growth markets; non-bank/private-credit encroachment on middle-market and asset-based lending; and deposit competition that has become explicitly granular — CFO Dmuchowski described an industry that has moved beyond uniform rates to “different rate specials in different cities” with tiered pricing by deposit size. Jordan’s own summary: “the demand for deposit and lending continue to probably put a little bit of pressure on relative spreads on both sides of the balance sheet.” FACT (quotes) / INTERPRETATION (conclusion).

How profitable is the business (ROIC, ROE)? ROIC in the industrial sense is not a meaningful metric for a bank — the correct analogues are return on assets, return on common equity and, above all, return on tangible common equity. FY2025: ROA 1.22%, ROCE 11.30%, ROTCE 14.01%. 2Q26: ROCE 12.3%, ROTCE 15.2%. FACT (FY2025 10-K Table 7.1; 2Q26 release.) INTERPRETATION: this clears a ~10–11% cost of equity, but not by much, and not by as much as the better regionals — peer disclosures put Regions near 18%, Fifth Third ~17%, Huntington 16–17%, M&T ~16%, KeyCorp ~15%. FHN is a mid-pack earner.

How profitable is the industry — how many competitors, what barriers to entry? US banking is a licence-protected industry with thousands of institutions and, at the industry level, a long-run return that is unremarkable. The Deutsche Bank primer held internally makes the normalisation point starkly: since 1935 there have been only fourteen years in which the banking industry earned an ROA above 1%, the long-term average industry ROA is about 0.75%, and industry return on common equity since 1935 has averaged about 10%. Barriers to entry are regulatory (a charter, capital, supervision) rather than economic — which is why the industry’s structural profitability is mediocre despite the licence. FACT (primer, third-party research, 2011 — framework not current data).

Can the business be easily understood? Yes, with one qualification. The spread bank is transparent. The Wholesale segment — mortgage-warehouse lending against physically-held collateral, franchise finance, correspondent banking and a fixed-income dealer — requires more work, and the disclosure collapses distinct risks (self-liquidating warehouse advances versus term lending to non-bank financial companies) into single line items. INTERPRETATION.

Can it be undermined by foreign low-cost labour? No, not in any direct sense. Deposit-taking and relationship lending are geographically bound. Back-office and technology functions are exposed to offshoring, but that is a cost opportunity rather than a competitive threat.

Do brands matter? Marginally. “First Horizon” carries genuine recognition in Memphis and Tennessee and the IBERIABANK heritage carries it in Louisiana, but bank brand equity converts into pricing power only weakly. The evidence: FHN’s deposits are 77% interest-bearing and repriced with a 66% cumulative beta. A brand that commanded loyalty would show up as a lower beta and a higher noninterest-bearing share. INTERPRETATION, grounded in FACT.

What is the nature of competition? Price and relationship. On loans, Jordan describes pricing as “very competitive, particularly for larger transactions.” On deposits, it is explicit rate competition, city by city. The differentiator management claims is service and speed — “big bank capabilities with a community bank touch” — delivered through relationship bankers, of whom 53 were hired in 2Q26 alone. FACT. INTERPRETATION: a moat delivered through individual bankers is a moat that can be hired away, and FHN’s own hiring strategy proves the point in both directions.

Customers’ switching costs? Genuine but modest, and asymmetric. A commercial operating account with integrated treasury management, payroll and lending covenants is sticky — the ACH/lockbox/reporting rewiring is a real cost. A consumer savings balance or a brokered CD is not sticky at all. Given that 58% of FHN’s deposits are commercial, the sticky half is the larger half — but commercial balances are also larger, more rate-aware and more concentrated, which is why they reprice. INTERPRETATION.

Financial Condition & Balance Sheet

Assets not fully recognised on the balance sheet? The principal one is the deposit franchise itself — a core-deposit intangible is recognised only when purchased (as in the IBERIABANK merger), never when built. A second is the servicing and client-relationship value inside wealth and treasury management. Neither is quantified by the company. INTERPRETATION.

Off-balance-sheet liabilities? The standard banking set: unfunded lending commitments (against which a separate reserve is held — the gap between the 1.15% allowance-to-loans ratio in the 10-K and the 1.31% ACL-to-loans ratio in the earnings release is precisely this unfunded-commitment reserve), standby letters of credit, and derivative counterparty exposure. Also relevant: mortgage-warehouse commitments, where FHN takes physical possession of the underlying note. FACT.

How conservative is the accounting? On the available evidence, more conservative than most peers — with one significant exception. In FY2025 total notable items were only −$12M after tax, or −$0.02 per share, so GAAP EPS of $1.87 sits within two cents of adjusted EPS; the 10-K presents only four non-GAAP measures (PPNR, ROTCE, TCE/TA, TBVPS) and no “adjusted EPS” at all. That is unusual and creditable. The exception is reserving. The ACL-to-loans ratio has fallen from 1.45% to 1.24% while the loan book grew, and provision has been set below net charge-offs in five of the last six quarters. Management attributes this to “criticized and classified loan resolutions” — $2.2B of non-pass resolutions in 2025 — and CFO Dmuchowski said in 4Q25 “we’re done in that building phase.” That may well be right. It is nonetheless a judgement that has flattered reported earnings by roughly $0.08 per share in FY2025 and is continuing into 2026. FACT for the numbers; INTERPRETATION for the characterisation.

How CapEx-hungry is the business? Not capital-expenditure-hungry in the industrial sense; FY2025 depreciation was $64M against $3.4B of revenue. It is regulatory-capital-hungry, which is the binding constraint: every dollar of loan growth consumes CET1. FHN has just completed a three-year, roughly $100M technology programme, which is why FY2026 expenses are guided flat — a one-off harvest. Dmuchowski’s own normalised expense-growth rate is “2.5% to 3%.” FACT.

Capital Allocation & Management

How much free cash flow does the business generate, and how is it used? “Free cash flow” is not the right lens for a bank; the correct analogue is distributable earnings after funding regulatory capital for loan growth. Management states the priority stack explicitly: “loan growth being the priority … Second is what is the right level of dividend … and third is share buyback.” In FY2025 FHN returned $894M via buyback plus roughly $300M of common dividends — approximately $1.19B against $956M of net income available to common, i.e. it distributed more than it earned, funded by running CET1 down from 11.20% to 10.64%. FACT.

Significant acquisitions recently? None since the 2020 IBERIABANK merger of equals. The defining recent M&A event was the failed sale to TD Bank Group (agreed February 2022 at $25.00 cash, terminated 4 May 2023, $225M received). Management has since signalled appetite to be an acquirer of in-footprint banks, then softened it — by 2Q26 Jordan said the environment “does feel more benign today than it did call it, 12 months ago” and that focus remains on “driving the profitability in this business.” FACT.

Buying back shares? Aggressively, and — this is the single strongest fact in management’s favour — well. FY2025 repurchases totalled $894M at an average price of $21.16 including commissions, against a year-end tangible book value of $14.20 and a current share price of $25.53. 2Q26 repurchases were $100M at an average of $24.52. A $1.2B authorisation was approved on 27 October 2025, running to 31 January 2027. The quarterly pace has nonetheless decelerated markedly — $360M → $9M → $190M → $335M → $230M → $100M — as CET1 reached the 10.5% near-term target. FACT.

Issuing large amounts of new shares to insiders? No evidence of unusual dilution; the share count has fallen materially. FHN did issue $400M of Series H preferred in 1Q26 (adding roughly 44bp of Tier 1), which is a financing decision rather than insider issuance — and, notably, a lever that lets the company hold its regulatory capital line while continuing to retire common equity. FACT.

Compensation policy of directors/management; motivations of management? See the capital-allocation section of the report for the detail. The structural point for diligence: Bryan Jordan has been chief executive since 2008 — an eighteen-year tenure spanning the financial crisis, the Capital Bank and IBERIABANK acquisitions, an agreed sale of the company and its collapse. That is an unusually long run, and it cuts both ways: deep institutional knowledge and a demonstrated willingness to sell the company at a good price, against the ossification risk that attends any two-decade incumbency.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No. FHN is a US domestic C-corporation filing 10-Ks with the SEC, incorporated in Tennessee, listed on the NYSE. Common stock is $0.625 par. Also NYSE-listed are depositary shares for the Series E, Series F and Series H non-cumulative perpetual preferred. Holders receive a Form 1099, not a K-1. FACT.

Dividend policy? A quarterly common dividend, raised 13% to $0.17 per share with the January 2026 declaration (from $0.15), payable 1 April 2026. Common dividends declared were $0.60 per share in each of FY2023, FY2024 and FY2025; the annualised rate is now $0.68, a forward yield of roughly 2.7% at $25.53. Preferred dividends run roughly $26–37M a year across the outstanding series. FACT.

How profitable is the business? Answered above: FY2025 ROA 1.22%, ROCE 11.30%, ROTCE 14.01%; 2Q26 ROTCE 15.2%; net interest margin 3.47%; efficiency ratio 60.66%. FACT.

Is net income diverging from cash from operations? The industrial version of this question does not translate cleanly to a bank, where operating cash flow is dominated by balance-sheet flows. The banking analogues are the ones that matter, and two are worth flagging: (i) the provision-versus-charge-off gap — FY2025 provision of $65M against $120M of actual net charge-offs, i.e. reported earnings exceeded cash credit losses absorbed by roughly $55M pre-tax; and (ii) purchase-accounting accretion from the IBERIABANK transaction, which is non-cash-recurring income that runs off over time. Both make reported earnings modestly higher-quality-looking than the underlying cash economics. FACT / INTERPRETATION.

Risks & Downside

What factors would cause the stock to decline? In rough order of probability-weighted impact: (i) credit normalisation — charge-offs reverting from 20bp toward a 40–50bp through-cycle rate while the reserve ratio has already been drawn down, which would hit provision twice; (ii) NIM compression from further Fed easing, given that over half the loan book is indexed to short-term rates and deposit costs are, in Jordan’s words, “drifting slowly towards wholesale cost of funds”; (iii) a dilutive acquisition — management has signalled appetite, and the market’s 9% one-day response on 15 October 2025 is a direct measurement of how it would be received; (iv) multiple compression from a price-to-book already in the 90th percentile of its own decade; (v) a further leg down in fixed-income ADR; (vi) a deposit event, to which a 42%-uninsured, 58%-commercial, 97%-loan-to-deposit balance sheet is not immune.

Risk of a catastrophic loss? Low but non-zero, and structurally higher than for a non-financial business of similar quality. Banks are leveraged roughly 10:1 on tangible common equity (FHN’s TCE/TA is 8.37%), so a loss of about 8% of assets would exhaust tangible equity. The relevant historical evidence is unflattering and should not be waved away: FHN’s lifetime maximum drawdown is −87.7% (the 2008–09 crisis, when its pre-2009 mortgage business nearly destroyed it — remnants of which still sit in the Corporate run-off segment), and within the last five years it has suffered a −60.8% drawdown in 86 days. FACT.

Chance of a total loss? Remote absent a systemic event. CET1 of 10.5% against a well-capitalised minimum of 7.0% (including the conservation buffer), charge-offs at 20bp, office exposure of only ~2.2% of loans, and access to FHLB, discount-window and repo liquidity make an equity wipe-out a tail scenario. The honest caveat is that this was also true of several banks in early 2023; bank solvency is a confidence variable as much as a capital variable.

Recent News & Events

Has the business environment changed recently? Yes, in two directions. Favourably: the yield curve has resumed a normal upward slope, deposit costs have fallen 42bp year-over-year, credit has stayed benign, and the proposed Basel III standardised approach would cut FHN’s risk-weighted assets by roughly 10% — a prospective capital release. Unfavourably: deposit competition has intensified and become geographically granular; the fixed-income business has weakened materially; and the October 2025 regional-bank credit scare re-priced the sector’s non-bank-financial exposure, an area where FHN carries ~$8.6B. FACT.

Significant acquisitions? None completed. See above.

Change in accounting policies? No change in accounting policy identified. There was a change in segment reporting: during 2024 FHN reorganised its internal management structure and reclassified its reportable segments from Regional Banking / Specialty Banking / Corporate to Commercial, Consumer & Wealth / Wholesale / Corporate, restating prior periods. Anyone working from pre-2024 segment data — or from third-party data providers, several of which still publish the old structure — will misread the business. FACT.

Recent changes — new markets, facilities, management? A steady organic build-out rather than any step change: 53 bankers hired in 2Q26; new commercial, private-client, trust and retail leadership in Atlanta (June 2026); a Gulf States Regional President appointment (January 2026); new market leadership in Cookeville and Sparta, Tennessee (July 2026); and a new Head of Consumer Distribution and ATM Strategy (January 2026) charged with optimising the physical and ATM network — which reads as branch rationalisation. On management: the Chief Risk Officer and Chief Credit Officer both took their seats in 2024–January 2025, immediately before the company began growing commercial-real-estate commitments more than 50% year-over-year and drawing down reserve coverage. That is not evidence of a problem; it is a configuration worth monitoring. FACT / INTERPRETATION.

APPENDIX B — Source Appendix

Every primary filing, transcript, dataset and third-party source relied on in the First Horizon Corporation (NYSE: FHN) research report dated 2026-07-26, with publisher, date, URL where one exists, local mirror path where applicable, and a one-line note on what each was used for. All sources accessed 2026-07-26 unless otherwise stated.

A. SEC filings (primary)

First Horizon Corporation, CIK 0000036966. EDGAR company page and full filing history: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000036966&type=&dateb=&owner=include&count=40 Archives root for all documents cited below: https://www.sec.gov/Archives/edgar/data/36966/

The trailing 60-month corpus (filings since 2021-07-01) was enumerated with scripts/edgar.sh since FHN 2021-07-01 — 538 filings — and mirrored with the data-pull scripts into the local filing mirror (149 primary documents plus all 308 Form 3/4/4A). Inventory: the local filing mirror and the local filing mirror.

A.1 Annual reports on Form 10-K (FY2021–FY2025)

A 10-K/A is also present in the corpus (the local filing mirror), reviewed and immaterial.

A.2 Quarterly reports on Form 10-Q (15 filings, 2Q21 through 1Q26)

All mirrored at the local filing mirror. Filed: 2021-08-05, 2021-11-05, 2022-05-06, 2022-08-04, 2022-11-07, 2023-05-08, 2023-08-04, 2023-11-07, 2024-05-03, 2024-08-02, 2024-11-07, 2025-05-07, 2025-08-07, 2025-11-06, 2026-05-07 (1Q26).

• 1Q26 10-Q, filed 2026-05-07 (2026-05-07_fhn-20260331.htm) — Table I.2.1 revenue mix; Table I.2.16 share repurchases; Note 7 Preferred Stock (Series H, $400M, 6.75%, first call 2031-04-10); Note 17 Subsequent Events (Series C redeemed 2026-05-01); cash-flow statement (buybacks, term-borrowing issuance); 1Q26 loan and deposit detail.
• 3Q23 10-Q, filed 2023-11-07 — the source of management’s own estimate that crossing $100B of assets would raise compliance cost by “roughly on the order of $100 million per year.” https://www.sec.gov/Archives/edgar/data/36966/000003696623000120/fhn-20230930.htm
• 3Q24 10-Q, filed 2024-11-07 — peak-quarter average deposit cost used to compute the hiking-cycle deposit beta.

A.3 Current reports on Form 8-K (59 in the window; all mirrored at the local filing mirror)

The TD Bank Group transaction

Quarterly earnings releases (Item 2.02, with Exhibit 99.1) — the source of all quarterly EPS, NIM, ROTCE, TBVPS, CET1, ADR, buyback and credit figures, and of the “Notable Items” reconciliation tables used to build the adjusted-EPS and adjusted-PPNR series.

• 2026-07-15 (2Q26) — 2026-07-15_fhn-20260715.htm; Exhibit 99.1 at https://www.sec.gov/Archives/edgar/data/36966/000003696626000143/a2q2026earningsrelease.htm — EPS $0.54, NIAC $260M (+12% y/y), ROTCE 15.2%, TBVPS $14.53, BVPS $17.91, CET1 10.5%, RWA $74.6B, ACL/loans 1.24%, NCOs 20bp, buyback $100M at $24.52, 474M shares.
• 2026-04-15 (1Q26), 2026-01-15 (4Q25/FY2025) — 2026-01-15_fhn-20260115.htm; the FY2025 notable-items bridge (−$13M pre-tax / −$0.02 EPS) and the disclosure that FHN “returned $894 million of capital to shareholders in 2025 through share repurchases at an average price of $21.16 per share including commissions.”
• Prior-year releases used for the multi-year series: 2025-10-15 (3Q25), 2025-07-16 (2Q25), 2025-04-16 (1Q25), 2025-01-16 (4Q24/FY2024), 2024-10-16, 2024-07-17, 2024-04-17, 2024-01-18 (4Q23/FY2023), 2023-10-18, 2023-07-19, 2023-04-18, 2023-01-18 (4Q22/FY2022), 2022-10-18, 2022-07-19, 2022-04-19, 2022-01-20, 2021-10-20, 2021-07-16.

Capital, buyback and financing

Executive, board and organisational

Noted as evidence in its own right: no 8-K was filed between 2023-03-13 and 2023-05-03 — through the entire regional-banking crisis. The deposit trajectory had to be reconstructed from the 1Q23 and 2Q23 earnings releases.

A.4 Proxy statements

A.5 Insider filings — Forms 3, 4 and 4/A

All 308 filings (292 Form 4, 10 Form 3, 6 Form 4/A) covering 2021-07-06 through 2026-05-14 were downloaded and machine-parsed into 597 transaction rows. Raw XML: the local filing mirror; parsed table: the local filing mirror. Reproducible from the manifest.

Used for the the recent-changes section insider read: 12 open-market purchases (code P), 52,036 shares, $537,634, all by directors, all between 2023-05-04 and 2023-06-07; against 128 sales, 1,912,641 shares, $42,502,593; 158 grants (4,543,864 shares), 124 tax-withholding dispositions (1,688,284 shares), 72 option exercises. Also the finding that only two of twenty-five sale filings reference a Rule 10b5-1 plan, and that no executive officer bought a share on the open market in five years.

A.6 Other filings reviewed

• SC 13G / 13G/A — 14 filings, no SC 13D. BlackRock (2024-11-08), Vanguard (2024-02-13, 2026-03-26), Vanguard Capital Management (2026-04-29), BNY Mellon (2025-01-23), State Street (2025-02-04, 2025-05-13), Millennium Management (2025-10-23, 2026-01-29). Used to establish that the register is passive index money and that no activist position exists.
• Forms 11-K (5), S-8 / S-8 POS (22), S-3ASR (2), 8-A12B (1), ARS (4) — reviewed for completeness; the 424B2 / FWP / 144 flow was excluded as structured-note noise per the standing methodology.

B. Earnings-call transcripts

Retrieval source of record: the ROIC.ai MCP (get_earnings_call_transcript / get_latest_earnings_call, identifier NYSE:FHN). Six consecutive calls were retrieved and read in full. Local copies of four are stored at the local filing mirror (git-ignored, shared across runs).

Published 2Q26 transcript (public cross-check for the ROIC body): “First Horizon (FHN) Q2 2026 Earnings Call Transcript,” The Motley Fool, published 2026-07-22 — https://www.fool.com/earnings/call-transcripts/2026/07/22/first-horizon-fhn-q2-2026-earnings-call-transcript/

Speakers across all six calls: D. Bryan Jordan (Chairman, President & CEO), Hope Dmuchowski (CFO), Thomas Hung (Chief Credit Officer), Tyler Craft (Head of Investor Relations).

Principally used for: the 66% cumulative down-cycle deposit beta and the 2.33%/2.43% interest-bearing deposit cost; the “mid to high 3.40s” NIM guide and the 3–7% FY2026 revenue guide; the mortgage-warehouse “1 basis point over ten years” and physical-note-custody claim; the NDFI decomposition ($8.6B, ~55% warehouse, sub-20-day dwell, private credit under 1% of loans); the FHN Financial ADR series and the ~60% compensation ratio; the “$100M+ PPNR opportunity” and management’s explicit refusal to report progress against it; the CET1 target ratchet and the loan-growth-first capital priority stack; the ~10% Basel III standardised RWA relief; the 3Q25 M&A comment that moved the stock; and the CEO’s “cost of deposits is drifting slowly towards wholesale cost of funds” concession.

Two caveats recorded for the reader. (i) The 2Q26 ROIC transcript is visibly machine-transcribed with ASR errors; every load-bearing number quoted from it was reconciled to the 2Q26 earnings release before use. (ii) ROIC’s coverage is earnings-call-only — no investor-day, conference-presentation or fireside-chat transcripts were available for FHN in ROIC or Google Drive, and management references several such appearances. Per the standing rule, all transcript content is management commentary, i.e. hypothesis, not evidence, and was validated against filings before entering the report.

C. Quantitative datasets

Every source in this section is third-party or unofficial and carries the same bind: it accelerates and cross-checks the filing, it does not replace it. Where an aggregator and a filing disagreed, the filing won.

• AZI price history (default price/OHLCV source). https://azitrading.com/controls/download-data.php?t=FHN (and =KRE for the index counterfactual), full history to 2026-07-24; saved to the local filing mirror. Split- and dividend-adjusted closes, EMAs, beta (1.114) and alpha (0.106). Source of the five-year event map, every drawdown figure, the total-return counterfactuals, and the annual buyback VWAP checks. Caveat: a commercial feed, not an exchange source; used for price history only, never for fundamentals.
• AZI fundamentals — valuation_index own-history percentiles only. scripts/azi.sh fundamentals FHN, .valuation_index — composite 81.4th, P/E 65.7th, P/B 89.6th, P/S 89.0th percentile of FHN’s own ~10-year range. Caveat: percentiles are own-history context only and are never comparable cross-sectionally; the rest of the AZI fundamentals feed (statement arrays, snapshot EV/market cap) was not used.
• ROIC.ai MCP. Tools used: get_company_profile, get_income_statement, get_balance_sheet, get_cash_flow, get_profitability_ratios, get_per_share_data, get_valuation_multiples, get_enterprise_value, get_company_news, get_earnings_call_transcript / get_latest_earnings_call.

⚠ MATERIAL DATA WARNING — ROIC.ai IS WRONG FOR FHN ON THREE FIELDS, AND NONE OF THEM WAS USED. (1) return_com_eqy prints 19.04% (FY2025), 16.36% (FY2024), 21.41% (FY2023), 24.84% (FY2022), 32.99% (FY2021) against the filed return on average common equity of 11.30 / 8.80 / 11.01 / 11.81 / 12.53%. It appears to be computed on a non-common-equity base. Every return figure in this report is filing-derived. (2) Per-share book values are internally impossible: ROIC reports FY2025 book_val_per_sh of $11.00 against tang_book_val_per_sh of $14.81 — tangible book cannot exceed total book. Company-reported figures are BVPS $17.53 (FY2025) / $17.91 (2Q26) and TBVPS $14.20 / $14.53. (3) Enterprise value is a category error for a deposit-funded lender: ROIC prints FHN’s EV as −$13.54B (FY2021), −$7.92B (FY2020) and −$0.53B (FY2023). No EV-based multiple appears anywhere in this report. What was used from ROIC: the year-end price-to-tangible-book series 2014–2025 (read as a series, not a level, because its tangible-book basis runs ~4% above the filing’s); ROA and effective tax rate, which reconcile to the filings (1.18% vs 1.22%; 22.03% vs 22.06%); the news feed; the company profile (with its retired segment names corrected against Note 19); and the transcripts in §B. ROIC’s FY2025 consolidated revenue of $3,323M was rejected in favour of the filing’s $3,419M. ROIC’s list_earnings_calls ignores its identifier argument and returns a global cross-ticker list; it is unusable for enumeration.

• FactorsToday factor model (https://www.factorstoday.com/api, no-auth), pulled 2026-07-26; JSON saved to the local filing mirror:
• /api/stock-loadings/FHN (models dated 2026-06-30 and 2026-07-24) — the four nested models, R² 0.7689 (All-Factors), and the finding that Momentum is L1-zeroed in all four.
• /api/leaderboard/FHN — annualised return, vol, max drawdown, Sharpe and Sortino by horizon (3-year Sharpe 0.90; 10-year 0.19; lifetime −0.029 on an −87.7% lifetime drawdown). All figures are annualised at every horizon; the mandatory de-annualisation check was run and reconciles to the AZI series.
• /api/stock-info/FHN — beta 1.114, alpha 0.106, relative strength, market cap $12.089B.
• /api/stock-specific-vol/FHN — idiosyncratic vol 14.37% (252-day).
• /api/related-stocks/FHN — factor-similar peers; KRE similarity 0.945, third behind WBS 0.966 and CFG 0.956.
• /api/factor-returns/historic and /intraday — the factor-regime table (DividendYield z +1.72, Value z +1.57 on 252 days). Caveats applied per the model’s own methodology: loadings are L1-sparse (absent = zeroed, not missing); betas are comparable only within a single nested model; the SmallSize loading (+0.48) is discounted because FHN issued heavily around the 2020 merger; and the intraday endpoint returned no signal at a weekend pull. These are third-party statistical estimates — the loadings and realised returns are reportable facts, any persistence claim is labelled interpretation.
• the data-pull scripts — SEC EDGAR XBRL companyfacts and submissions history for CIK 0000036966. Used for the corpus enumeration and for us-gaap:PaymentsForRepurchaseOfCommonStock, WeightedAverageNumberOfDilutedSharesOutstanding, CommonStockDividendsPerShareDeclared (confirming the quarterly dividend rose $0.15 → $0.17 in 1Q26), PaymentsOfDividendsCommonStock, ShareBasedCompensation, and us-gaap:Goodwill for Cullen/Frost. Authoritative — this is the filer’s own XBRL.
• the data-pull scripts (yfinance) — closing prices for 2026-07-24, market caps, book value per share, TTM ROE, dividend yields, and 2026-03-31 quarterly balance sheets (common equity, goodwill and other intangibles, shares outstanding) for 23 US regional banks, from which the peer TBVPS, P/TBV and the ROTCE→P/TBV regression were derived. Explicitly unofficial. The derivation was reconciled on FHN itself — it returns 1Q26 TBVPS of $14.34 against the filed $14.20 (FY2025) and $14.53 (2Q26) — but its tails are noisy: the goodwill line appears overstated for WAL and was missing entirely for CFR (corrected from EDGAR). Peer figures derived this way are labelled as estimates, not reported figures (see §G).

D. Industry, regulatory and market data

• FDIC, Quarterly Banking Profile, First Quarter 2026 — press release 2026-05-27, https://www.fdic.gov/news/press-releases/2026/fdic-insured-institutions-reported-return-assets-126-percent-and-net ; Chairman’s remarks, https://www.fdic.gov/news/speeches/2026/fdic-quarterly-banking-profile-first-quarter-2026 — 4,278 insured institutions, industry ROA 1.26%, NIM 3.31% (−8bp q/q), net interest income −$1.6B, efficiency 56.0%, reserve coverage 171.2% → 166.8%, loan growth 7.1% annualised, deposit and uninsured-deposit trends.
• FDIC, 2026 Risk Review (full report) — https://www.fdic.gov/analysis/2026-risk-review-full.pdf — the business-overview section.2 loan performance, the industry section funding and interest-rate risk, the competitive-position section.1 CRE (office vacancy 14.0%, non-owner-occupied noncurrent 1.59%, modified CRE $11.6B/0.38%, the $10–100B cohort’s 289% median CRE concentration, ADC −5.8%), the competitive-position section.2 NDFI lending ($1.4T, 5.6% of industry assets, 22.7% CAGR, 86% at banks >$100B).
• FDIC, Summary of Deposits, 30 June 2025 — annual survey released 2025-09-19, https://www.fdic.gov/news/press-releases/2025/fdic-releases-results-summary-deposits-annual-survey ; branch-level data pulled from the public API https://api.fdic.gov/banks/sod and https://api.fdic.gov/banks/institutions?filters=CERT:4977 (First Horizon Bank, FDIC Cert #4977). All deposit-share, rank, branch-level and trend figures in the competitive-position section are our own computation from the raw 2025 SOD branch file (76,120 records) — the Memphis HQ-booking finding, the Tennessee 2015–2025 share series, the state and MSA maps, the pro-forma Pinnacle+Synovus comparison, and the post-IBERIABANK erosion table. Methodology validated to the basis point against Pinnacle’s own FDIC-sourced published figures on every overlapping datapoint. Note: the 30 June 2026 SOD is not yet published (a YEAR:2026 query returns zero records); the “>$1B per branch” HQ-normalisation cutoff is a reasonable but arbitrary threshold and the resulting ex-HQ ranking is threshold-sensitive.
• FDIC — special assessment, assessment rates and brokered deposits. https://www.fdic.gov/deposit-insurance-assessments/special-assessment-pursuant-systemic-risk-determination (final payment 2026-03-30 at 2.97bp); FIL, “FDIC Withdraws Proposed Rules Related to Brokered Deposits,” 2025-03, https://www.fdic.gov/news/financial-institution-letters/2025/fdic-withdraws-proposed-rules-related-brokered-deposits ; Sullivan & Cromwell, “FDIC Proposes Revisions to Insured Depository Institution Resolution Planning Rule and Deposit Insurance Assessments,” 2026-07, https://www.sullcrom.com/insights/memo/2026/July/FDIC-Proposes-Revisions-IDI-Resolution-Planning-Rule-Deposit-Insurance-Assessments (the $50B → $100B resolution-planning threshold and the ~$3.4B/yr assessment saving).
• Federal Reserve, H.8 Assets and Liabilities of Commercial Banks, release of 2026-07-02 (data week ending 2026-06-24) — https://www.federalreserve.gov/releases/h8/20260702/ — total loans $13.9T (+6.1% annualised), C&I $2.9T (+10.9%), NDFI lending $1.99T (+15.4% y/y).
• Federal Reserve, FOMC materials, 17 June 2026 — projections https://www.federalreserve.gov/monetarypolicy/fomcprojtabl20260617.htm ; minutes https://www.federalreserve.gov/monetarypolicy/fomcminutes20260617.htm ; CNBC coverage https://www.cnbc.com/2026/06/17/fed-interest-rate-decision-june-2026.html — the fourth consecutive hold at 3.50–3.75%, the upward-revised dot plot (nine of nineteen see a hike), and the 3.6%/3.3% inflation forecasts. Treasury yields (2Y 4.21%, 10Y 4.56%) from Advisor Perspectives’ 2026-07-10 snapshot, https://www.advisorperspectives.com/dshort/updates/2026/07/10/treasury-yields-snapshot-july-10-2026 .
• Federal Reserve / OCC / FDIC, joint capital re-proposals, 19 March 2026 — https://www.federalreserve.gov/newsevents/pressreleases/bcreg20260319a.htm ; analyses: Holland & Knight, 2026-06, https://www.hklaw.com/en/insights/publications/2026/06/us-banking-agencies-propose-new-rules-to-reduce-regulatory ; Freshfields, “Basel III Endgame, Take Two,” https://www.freshfields.com/en/our-thinking/blogs/a-fresh-take/basel-iii-endgame-take-two-8-key-takeaways-from-the-federal-banking-agencies-c-102mnm3 ; Sullivan & Cromwell, 2026-03, https://www.sullcrom.com/insights/memo/2026/March/Fed-Vice-Chair-Bowman-Previews-Basel-III-GSIB-Surcharge-Proposals — the rescission of the 2023 proposal, the ~4.8% / ~5.2% / ~7.8% RWA relief estimates, and the mandatory AOCI inclusion for Category III/IV over a five-year phase-in.
• Federal Reserve, 2026 DFAST results, 24 June 2026 — https://www.federalreserve.gov/newsevents/pressreleases/bcreg20260624a.htm ; Bank Policy Institute, “The 2026 Federal Reserve Stress Test Results: A Framework in Transition,” 2026-07, https://bpi.com/the-2026-federal-reserve-stress-test-results-a-framework-in-transition/ — the 1.6pp aggregate CET1 decline, the SCB freeze at 2025 levels through 2027, and the 8.8% CRE portfolio loss rate assumption.
• Congressional Research Service — R47876, “Enhanced Prudential Regulation of Large Banks,” https://www.congress.gov/crs-product/R47876 (what attaches at $100B / Category IV); IN12680, “Recent Regulatory Changes to Bank Capital Requirements,” https://www.everycrsreport.com/reports/IN12680.html .
• OCC — Interim Final Rule on Bank Mergers, NR 2025-44, effective 2025-05-15, https://www.occ.gov/news-issuances/news-releases/2025/nr-occ-2025-44.html ; Bulletin 2006-46, Interagency Guidance on CRE Concentration Risk Management (the 100%/300% tests), https://www.occ.gov/news-issuances/bulletins/2006/bulletin-2006-46.html .
• Mortgage Bankers Association — “Total Single-Family Mortgage Originations to Increase 8 percent to $2.2 Trillion in 2026,” 2025-10-19, https://www.mba.org/news-and-research/newsroom/news/2025/10/19/mba-forecast--total-single-family-mortgage-originations-to-increase-8-percent-to--2.2-trillion-in-2026 ; CREF Forecast, 2026-02-09, https://www.mba.org/news-and-research/newsroom/news/2026/02/09/mba-cref-forecast--total-commercial-mortgage-originations-to-increase-27-percent-to--805-billion-in-2026 .
• US Census Bureau, Vintage 2025 metropolitan population and migration estimates, released 2026-03-26 — https://www.census.gov/newsroom/press-releases/2026/population-growth-slows.html — the the industry section.5 metro table (Nashville +1.60% / +16,967 net domestic migration; Memphis −0.31% / −10,302 / −0.30% since April 2020; Miami −113,724; all 387 MSAs +0.58%). Supporting: Population Reference Bureau on Shelby County depopulation, https://www.prb.org/news/how-depopulation-is-slowing-economic-growth-in-shelby-county-tennessee/ ; Nashville Area Chamber, 2026 Regional Economic Overview, https://s3.us-east-1.amazonaws.com/nashvillechamber.com/Economic-Development/NACC_ED_Nashville+Region+Economic+Overview_2026_Digital.pdf ; NAR state migration data, https://www.nar.realtor/news/economists-outlook/top-15-states-for-population-and-migration-trends-in-2025-the-south-leads-the-midwest-gains .
• Louisiana economy — KSLA, “Louisiana leads nation in population loss for second consecutive year,” 2026-02-12, https://www.ksla.com/2026/02/12/louisiana-leads-nation-population-loss-second-consecutive-year/ ; Pelican Institute, https://pelicanpolicy.org/tax-budget/louisianas-economy-needs-bold-reforms-to-reverse-slow-growth-and-out-migration/ ; The Data Center, “Pathways to Prosperity: Louisiana 2026,” https://www.datacenterresearch.org/reports_analysis/pathways-to-prosperity-louisiana-2026/ .
• CRE and CMBS — Trepp via REI Prime, “CMBS Maturity Wall: $76.6B in Hard Maturities, Office Delinquency at Record 12.34%,” 2026-05-07, https://reiprime.com/news/2026-05-07-cmbs-maturity-wall-trepp-spring-review ; CRE Daily, https://www.credaily.com/briefs/cmbs-maturity-wall-tests-refinancing-in-2026/ ; BankHealthData, https://www.bankhealthdata.com/blog/commercial-real-estate-bank-risk-2026 .
• Private credit / NDFI — Office of Financial Research Brief 26-02, “Measuring Counterparty Exposures to Private Credit,” 2026-03-12, https://www.financialresearch.gov/briefs/files/OFRBrief-26-02-measuring-counterparty-exposures-private-credit.pdf ; Financial Stability Board, “Report on Vulnerabilities in Private Credit,” 2026-05-06, https://www.fsb.org/uploads/P060526.pdf ; Moody’s, “US banks’ private credit loan exposure nears $300 billion,” 2025-10-21, https://www.moodys.com/web/en/us/insights/data-stories/breakdown-of-banks-annual-reporting-on-private-credit.html ; Mordor Intelligence market sizing, https://www.mordorintelligence.com/industry-reports/private-credit-market .
• Deposit competition — Bloomberg, “Dash for Cash Sends Money Fund Assets to Record $8.3 Trillion,” 2026-05-29, https://www.bloomberg.com/news/articles/2026-05-29/dash-for-cash-sends-money-fund-assets-to-record-8-3-trillion ; Investment Company Institute money-market fund statistics, https://www.ici.org/research/stats/mmf .
• Scale-and-efficiency evidence — International Banker, “The Efficiency Myth: Why Bigger Banks Don’t Necessarily Perform Better,” https://internationalbanker.com/banking/the-efficiency-myth-why-bigger-banks-dont-necessarily-perform-better/ ; SRA Consulting, https://www.sraconsults.com/bank-efficiency-ratios-size-business-model-and-the-distribution-story/ ; CEIC/FDIC efficiency-ratio series, https://www.ceicdata.com/en/united-states/performance-and-condition-ratios/fdic-commercial-banks-efficiency-ratio .
• Technology-spend comparison — Reuters via AOL, “JPMorgan will spend almost $20 billion on technology this year,” https://www.aol.com/news/jpmorgan-spend-almost-20-billion-000403027.html (the ~$19.8B 2026 budget used against FHN’s $138M computer-software line).

E. Third-party and press

• TD Bank Group, “TD Bank and First Horizon Mutually Agree to Terminate Merger Agreement,” 2023-05-04 — https://td.mediaroom.com/2023-05-04-TD-Bank-and-First-Horizon-Mutually-Agree-to-Terminate-Merger-Agreement — the acquirer’s own account of the termination (“no timetable for regulatory approvals… for reasons unrelated to First Horizon”). Corroborated by The Globe and Mail, https://www.theglobeandmail.com/business/article-td-first-horizon-terminate-merger/ .
• Banking Dive, “TD, First Horizon terminate $13.4B merger,” 2023-05-04 — deal-size and termination-fee confirmation.
• Banking Dive, “M&A in 2026 may put more distance between big, small banks,” https://www.bankingdive.com/news/2026-bank-mergers-acquisitions-outlook-faster-approval-regionals-midterm-elections-buyer-pool/809514/ — the merger-approval window and the midterm-election timing constraint.
• Banking Dive, “First Horizon calls out Raleigh as a market of interest,” https://www.bankingdive.com/news/first-horizon-cfo-dmuchowski-southeast-bank-mergers-acquisitions/809873/ — the CFO’s January 2026 walk-back of the acquisition signal, the sizing of appetite, and the Raleigh expansion commentary.
• Banking Dive, “Inside the explosion of banking charter applications,” https://www.bankingdive.com/news/inside-the-explosion-of-banking-charter-applications/810250/ — de novo and fintech charter pipeline.
• Bloomberg, “First Horizon’s (FHN) M&A Plan Tanks Stock in Worst Day Since TD-Deal Wipeout,” 2025-10-15 — the single most thesis-relevant press item: the ~9% one-day decline on the CEO’s acquisition commentary and Bloomberg Intelligence’s read that “investors had believed the bank could be a takeover target.” Corroborated by Investing.com, “First Horizon stock falls 9% as CEO signals acquisition plans.”
• American Banker — “Lending to nonbanks is booming. Will it last in 2026?” https://www.americanbanker.com/news/lending-to-nonbanks-is-booming-will-it-last-in-2026 ; “Will the slow pace of new bank formation pick up in 2026?” https://www.americanbanker.com/news/will-the-slow-pace-of-new-bank-formation-pick-up-in-2026 .
• PRNewswire — First Horizon company releases (the the recent-changes section organisational and capital-return timeline):
• “First Horizon Corporation Delivers Strong Second Quarter 2026 Results…,” 2026-07-15, https://www.prnewswire.com/news-releases/first-horizon-corporation-delivers-strong-second-quarter-2026-results-with-net-income-available-to-common-shareholders-of-260-million-up-12-year-over-year-and-eps-of-0-54--up-0-09-from-second-quarter-2025--302825692.html
• “First Horizon Announces $1.2 Billion Share Repurchase Program; Declares Cash Dividends…,” 2025-10-27 (also via Barchart / FinancialContent).
• “First Horizon Declares Cash Dividends on Common and Preferred Stock,” 2026-01-27 (the 13% dividend increase to $0.17); also Nasdaq and Investing.com coverage.
• Leadership releases: Gulf States Regional President and Head of Consumer Distribution & ATM Strategy (January 2026); Atlanta commercial/private-client/trust/retail expansion (June 2026); East Tennessee business banking (June 2026); Cookeville and Sparta market leadership (2026-07-10).
• Zacks, “FHN Shares Fall Despite Q2 Earnings Beat on Higher NII & Fee Income,” 2026-07-16 — attribution for the −3.0% session on the 2Q26 print.
• Peer-deal sources (the 2025–26 consolidation record):
• Pinnacle / Synovus — Businesswire, “Pinnacle and Synovus Complete Merger to Become Regional Bank Growth Champion,” 2025-12-30, https://www.businesswire.com/news/home/20251230366754/en/Pinnacle-and-Synovus-Complete-Merger-to-Become-Regional-Bank-Growth-Champion ; Synovus release, 2026-01-02, https://www.synovus.com/about-us/news/2026/2026-01-02-pinnacle-and-synovus-complete-merger-to-become-regional-bank-growth-champion ; Pinnacle completion release, https://www.pnfp.com/about-pinnacle/media-room/news-releases/pinnacle-and-synovus-complete-merger-to-become-regional-bank-growth-champion/ — the $8.6B all-stock terms and the $117.2B pro-forma institution.
• Pinnacle deposit-share releases — “Pinnacle is the No. 1 bank in the Nashville MSA by deposits for the eighth consecutive year; holds No. 2 in Tennessee…,” 2025-09-25, https://www.pnfp.com/news/news-releases/pinnacle-is-the-no-1-bank-in-the-nashville-msa-by-deposits-for-the-eighth-consecutive-year-holds-no-2-in-tennessee-grows-market-share-in-22-of-27-msas/ (the “less than half a percentage point behind the leader” claim, which pointedly declines to name First Horizon); and the Memphis release, https://www.pnfp.com/news/news-releases/pinnacle-is-the-third-largest-bank-in-memphis-by-deposits-crosses-2-billion-for-the-first-time/ . Both were used to validate our own FDIC SOD methodology.
• Fifth Third / Comerica, Huntington / Cadence, Santander / Webster — terms, closing dates and paid multiples taken from the acquirers’ own transaction disclosures and merger registration statements, cross-read against prior published coverage of FITB, HBAN and WBS (§F).
• Motley Fool, “Bank Mergers Just Hit a 7-Year High,” 2026-07-02, https://www.fool.com/investing/2026/07/02/bank-mergers-just-hit-a-7-year-high-here-are-the-r/ — the H1-2026 count of 83 deals / ~$15.1B. Note: this piece also groups FHN with FNB and Webster as “likely targets”; that is market commentary and is explicitly not treated as evidence anywhere in the report.
• Reed Smith, “U.S. Bank M&A Outlook for 2026 and Beyond,” https://www.reedsmith.com/articles/us-bank-ma-outlook-for-2026-and-beyond/ .
• Private-credit stress coverage — CNBC, “From Jamie Dimon’s ‘cockroaches’ to the Blue Owl freeze,” 2026-02-24, https://www.cnbc.com/2026/02/24/private-credit-3-trillion-boom-bankruptcies-fraud-blue-owl-redemptions-tricolor-first-brands-bdc.html ; Forbes, “Rising Private Credit Defaults Are Testing Banks And Insurers,” 2026-05-24, https://www.forbes.com/sites/mayrarodriguezvalladares/2026/05/24/rising-private-credit-defaults-are-testing-banks-and-insurers/ .
• FHN Financial corporate description — https://www.fhnfinancial.com/about . Used only for the business description; note that no market-position claim from this source was accepted (see §G).

F. Internal prior work

Two distinct categories, both labelled as internal and neither treated as primary evidence.

F.1 Google Drive — sector primer

** “Banking 101 — Large Cap Bank Primer,” Deutsche Bank Securities Inc., Special Report, 11 May 2011.** Authors Matt O’Connor CFA, Adam Chaim CFA, Robert Placet CFA, David Ho CFA; 98 pages plus appendices. Drive file id 13jv6kEyGDcBc1-rMPpc88j80tB5w_WYQ, within the shared “Primers” folder (id 13hBqJuBu3gJtgLmJ55Yleh_3OgvA41vd). Extracted text staged at the local filing mirror.

This is dated third-party broker research, used for analytical framework only and never as current data. It predates CECL, CCAR/DFAST/SCB, EGRRCPA tailoring and the 2023 regional-bank AOCI episode. Used for: the NII/fee revenue architecture; the four-factor NIM decomposition; the funding-stability hierarchy (and its point that brokered deposits are the least stable funding); the efficiency-ratio definition and the non-monotonic relationship between asset size and efficiency (banks $50–100B beat banks >$100B); the provision-versus-charge-off reserve build/release signal; and above all the normalisation prior — industry ROA has exceeded 1% in only fourteen years since 1935, the long-run average is ~0.75%, and industry return on common equity since 1935 has averaged ~10%. Every one of these is directionally corroborated in the report against current FDIC series. Caveat: the Drive text extraction truncates at page 79 of 98, so the primer’s own bank-valuation chapter (pp. 85–88) was not recoverable; the valuation framework in the valuation section was built independently.

Also present in the Drive “Primers” folder and reviewed but not read in full: “Global Banking Industry Primer” (Deutsche Bank, 2011), “Global Banking Outlook” (UBS, 2010), “Profiles of US Banks” (Ceres, 2008), “Loan Market Guide” (Standard & Poor’s, 2011).

A company-specific Drive sweep returned nil. Title- and full-text queries on FHN, First Horizon, IBERIABANK, Bryan Jordan and Memphis-bank terms returned no memo, model, deck, transcript, watchlist entry or note of any kind. This is fresh coverage with no internal prior thesis on the company.

F.2 Prior internal bank reports mined for peer data

All are prior published reports by this author in , attributed as such and not re-verified externally except where a figure drives a verdict. Used for peer ROTCE, NIM, efficiency ratio, CET1, net-charge-off rate, deposit cost, noninterest-bearing mix, deposit beta, the 2025–26 deal record, and moat-verdict calibration.

A standing caution applied throughout: none of these prior reports, nor any Drive artifact, watchlist or simulation output, is evidence of a position in First Horizon. No position is stated, implied or assumed anywhere in this report.

F.3 Analytical frameworks

.claude/skills/investment-research-frameworks/SKILL.md — Greenwald & Kahn, Competition Demystified (the barriers-to-entry taxonomy, the market-share-stability and ROIC tests, EPV versus asset value) and Chancellor / Marathon, Capital Returns (supply-side capital-cycle analysis, the asset-growth anomaly, procyclical management behaviour). Applied in the industry section, the competitive-position section, the capital-allocation section and the diligence appendix.

G. Data limitations and unverified claims

Stated plainly, because the report’s credibility rests on being clear about what could not be established. Nothing in this section is a gap that was papered over — each item is flagged in the body as an open question or an estimate.

1. FHN Financial’s market position could not be verified — at all. The FY2025 10-K makes no market-position claim; it says only that “we have many competitors within both groups, including major U.S. and international securities firms as well as numerous regional and local firms.” No independent league table ranking fixed-income distributors to depository institutions (against Stifel, Piper Sandler, Raymond James, Hilltop Securities, Janney, D.A. Davidson, StoneX) could be located in public sources. Any “one of the largest dealers to depositories” claim in circulation is marketing, not filing-grade, and is recorded in the report as unverified rather than as fact.

2. Office commercial real estate: no loan-to-value distribution, no maturity schedule, no office-specific reserve rate. First Horizon discloses office as 20% of a $14B CRE book ($2,694M) and states that “nearly half” is medical office — and nothing further. Meanwhile the total CRE allowance was cut from 1.53% to 1.10% in twelve months. Office is ~4.1% of loans and ~39% of tangible common equity. The honest characterisation is therefore “unverifiable,” not “clean.”

3. Shared national credit exposure is not disclosed anywhere — not in any 10-K, 10-Q or earnings release in the 60-month corpus, and not asked or volunteered on any of the six earnings calls reviewed. With C&I at 57% of loans and ~25% of that book lent to financial intermediaries, this is a genuine unknown.

4. CECL scenario weights are not disclosed. FHN discloses that it uses Moody’s baseline, S1 upside and S3 adverse scenarios with management weighting over a three-year reasonable-and-supportable period, and it discloses the endpoint sensitivities (100% baseline = −7% allowance; 100% adverse = +31%, roughly $230M pre-tax / ~$0.35 a share). The weighting itself is not disclosed, so the portion of the reserve release attributable to genuine credit improvement rather than to an assumption change cannot be isolated.

5. CEO succession: nothing exists to cite. No successor, bench, or process appears anywhere in the 2026 proxy, against a hard 2028-08-03 employment-agreement expiry and an incumbent of eighteen years who holds the Chairman, President and CEO titles simultaneously. Likewise, no successor Risk Committee chair has been disclosed following Lead Director Colin Reed’s departure at the 2026 annual meeting.

6. No evidence was found either way on banker-team departures to competitors. We looked. First Horizon is not named in trade coverage of Pinnacle’s hiring programme, and no report of a team lift-out from FHN was located. The risk described in the competitive-position section.6 is structural and unrealised, and is recorded as an open question rather than as a fact.

7. Peer fundamentals for HWC, SSB, CFR, BOKF and PB are author-derived estimates, not filing-sourced. Their ROTCE figures in the the opinion block comp table (marked “e”) are estimated as yfinance TTM ROE × (BVPS ÷ TBVPS), with tangible book derived from 2026-03-31 balance sheets. The estimator is accurate to well within a point on the FHN control (it returns 14.3% against a filed FY2025 14.01% and a 2Q26 15.2%), but these five names carry no verified NIM, efficiency, CET1 or charge-off data and must not be quoted as reported figures. Related: the peer tangible-book derivation uses yfinance’s “Goodwill And Other Intangible Assets” line, which appears overstated for WAL and was missing entirely for CFR (corrected from EDGAR XBRL).

8. The ROTCE→P/TBV regression is a weak instrument, and a second one disagrees with it. Ours (n=19, 2026-07-24 prices, 1Q26 tangible book) fits P/TBV = 0.7154 + 0.0747 × ROTCE with an R² of only 0.322 — return on tangible common equity explains barely a third of the cross-sectional variation in regional-bank price-to-book. A parallel exercise on nine hand-picked names fits a steeper line (0.632 + 0.0964 × ROTCE, R² 0.62) on which FHN would look ~12% cheap. Both are reported in the opinion block; neither should carry the argument alone.

9. The Memphis HQ-booking normalisation rests on our own computation and a judgement threshold. The underlying fact — that $10.50B of FHN’s $14.01B of Memphis MSA deposits sits in a single branch record at 165 Madison Avenue — is raw FDIC data. The resulting ex-headquarters ranking (Regions #1 at 16.32%, First Horizon #2 at 12.86%) depends on a “>$1B per branch” cutoff that is reasonable but arbitrary, and is threshold-sensitive. No third party has published this finding; it is presented as our own analysis. It is also applied symmetrically — Pinnacle’s Nashville share falls the same way.

10. Jordan’s beneficial ownership does not reconcile between two primary sources. The Form 4 series shows roughly 1,078,000 shares direct plus indirect at 2026-01-29; the 2026 proxy reports 1,743,677 shares (including 125,786 exercisable options) at 2026-01-31. The gap is approximately the 1.13M shares transferred to a family trust and to his spouse during 2025, which the Form 4 footnotes assert he continues to beneficially own but which were never added to his reported indirect holdings. The report uses the proxy figure; the discrepancy is unresolved.

11. Three further reconciliation items left open. (i) Management’s decomposition of the $8.6B NDFI book has moved across calls — “about 60%” (3Q25), “closer to two-thirds” (4Q25), “about 55%” (1Q26) mortgage warehouse; the actual call-report split was not pinned. (ii) A 4Q25 call statement on commercial-real-estate origination yields (“534 basis points”) is almost certainly an automated-transcription garble and was not used. (iii) Jordan’s Summary Compensation Table pension/NQDC column ($3.46M cumulative 2023–2025) exceeds his entire accumulated pension present value ($1,283,294) on a plan frozen since 2012, while the proxy simultaneously states no above-market NQDC earnings accrued; the column does not reconcile.

12. The synergy claims of the three January/February 2026 mergers are unfalsified, not validated. None of Fifth Third/Comerica, Huntington/Cadence or Pinnacle/Synovus has yet reported a full post-close year. The acquirer-returns evidence used in the M&A-option discussion is therefore directional rather than conclusive.

13. Coverage gaps in the transcript record. ROIC.ai carries earnings calls only. No investor-day, conference-presentation or fireside-chat transcript was available for FHN in ROIC or in Google Drive, although management repeatedly references such appearances. Segment-level and capital-allocation detail from those venues is missing from this report.

End of Appendix B.