Jefferies Financial Group Inc. (NYSE: JEF) — The Share Gains Are Real; the Returns Belong to the Employees

An independent fundamental research note Date: 2026-07-26 · Fiscal year end: November 30 Price at analysis: $55.86 (2026-07-24) · Market capitalisation: ~$12.9bn (230.6m common-equivalent shares) Sector: Financials · Capital Markets / Investment Banking & Brokerage · CIK: 0000096223

⚡ Claude’s Take

This section is the author’s own independent opinion and is offered as general information only — it is not investment advice, and it is not a recommendation to buy or sell any security. The analytical body (Sections 1–15) below deliberately carries no recommendation and no price target; this block is the single place a view is expressed.

Verdict: AVOID at this price. Not a short. Revisit at or below ~1.0–1.1x as-converted tangible book — roughly $38–42 per share.

Jefferies has done something genuinely difficult. It converted a sprawling Leucadia conglomerate into a focused investment bank, took global advisory fee share from 3% to 4.5% and moved from eighth to sixth in the league tables, and did it while the Europeans retreated. Management deserves the credit, and the bull case — a scarce independent full-service platform gaining share into a fee pool annualising at a record ~$160bn — is not stupid. The problem is that none of it reaches the owners. In FY2025 Jefferies generated a pre-compensation profit pool of $4.73bn and paid $3.86bn of it, 81.6%, to employees. Shareholders got the remaining 18.4% before tax. The result is a return on tangible equity of 8.5% in FY2025 and 10.6% over the trailing twelve months at a cyclical peak, against a cost of equity of roughly 11–12% for a stock with a 1.61 beta, no deposits and no central-bank backstop. Jefferies has not cleared its hurdle rate in any year since 2021 — and it does not clear it even on management’s own generously-adjusted 10.1% figure. Strip out the optical illusion created by SMBC exchanging 27.6m shares into non-voting preferred and as-converted tangible book value per share has gone from $36.96 in FY2022 to $37.87 today: four years, essentially nothing.

You are asked to pay 1.47x as-converted tangible book (1.62x on the company’s own adjusted measure) for that. The stock sits 28.7% below its December-2024 high, which makes it look chastened, but on its own ten-year history it is in the 87.7th percentile on price-to-book — because the multiple re-rated while the book did not compound. So the framing is not “fallen angel.” It is cyclical peak earnings on a structural-peak multiple, in an industry where the scarce asset takes the lift home every night and can be hired by a competitor tomorrow. The factor model agrees and is unsentimental: JEF loads +1.44 on Market, +0.50 on CreditRisk, +0.49 on Broker-Dealers, roughly zero on Quality and negative on Value — it does not trade as a cheap stock, it trades as levered credit beta. That is also why the drawdowns are violent: −49.5% into April 2025, −45.2% into March 2026, with First Brands litigation the proximate trigger. I am not short it, because the balance sheet is genuinely sound — Level 3 assets are only 1.0% of assets, average daily VaR is $11.2m against $8.5bn of tangible equity, and the liquidity buffer is 23.9% of assets ex-goodwill. This is not a fragile company. It is a fairly-to-richly-priced one that does not earn its cost of capital.

What tips this from “expensive” to “avoid” is the conduct around the credit losses. Management calls First Brands “idiosyncratic” — but it is the fourth such event, after George Weiss, the 3|5|2 Capital Ponzi ($17.2m) and now Market Financial Solutions, a £103m UK facility where the collateral “may have been double-pledged,” the same failure mode as First Brands. Management guided MFS to “less than $20 million”; the Q2 10-Q booked a $58.7m gross mark-to-market loss. Third-party money in Jefferies-managed funds has fallen from $2,462m to $1,261m in six months, −49%, and the firm has never narrated it. First Brands appears nowhere in the FY2025 10-K’s risk factors or legal proceedings, and the Q2 10-Q affirmatively deleted the Eugenia and Western Alliance suits from Legal Proceedings. Jefferies also holds no quarterly earnings call — there is no forum in which any of this can be asked. And through a 54% drawdown, not one insider bought a single share: the five-year record is $1.86m of purchases against $246.4m of sales, none under a 10b5-1 plan, with roughly $89m sold by Handler, Friedman and a director in the six weeks before the all-time high. Insiders defended the stock with two open letters and with shareholders’ money, not their own.

Conviction: medium-high. The returns arithmetic is not in dispute; what is uncertain is how long this cycle runs, and a hot FY2026 could carry the shares higher before the maths asserts itself. What would flip me bullish: four consecutive quarters of ROTE above 13% with the compensation ratio falling below 50% — genuine operating leverage rather than a revenue-driven flatter — accompanied by real disclosure of concentration limits and a resumption of insider buying. What would flip me outright bearish: a fifth counterparty failure, an adverse SEC or DOJ outcome, or evidence that the Point Bonita redemptions are spreading to the wider Jefferies Credit Partners platform — which would impair the recurring credit-management franchise that is the only place a durable advantage could plausibly form.

Tag: “Sixth place, first-class pay.”

📈 Stock Price Action — Five-Year Event Map

Jefferies compounded from a $18.56 low in January 2021 to an all-time closing high of $78.36 on 2024-12-16 — a better-than-four-bagger — and has since round-tripped a third of it, closing at $55.86 on 2026-07-24, 28.7% below that high. The 52-week range is $35.74–$68.80. The path down has not been a drift; it has been three separate violent drawdowns of 30–50%, each with an identifiable trigger, and the last one bottomed only four months ago.

Cycle narrative. (1) FY2021 was the best year in Jefferies’ history and remains unmatched — net revenues of $8.01bn have not been exceeded in the five years since. (2) The 2022 rate shock hit underwriting far harder than advisory (−58.7% versus −4.8%), the first evidence that Jefferies’ earnings torque sits in leveraged finance and ECM, not in the advisory ballast. (3) The 2023–24 re-rating was the genuine article: fee share rose, the SMBC alliance deepened, and the November-2024 election added an 11.2% single-day repricing on deregulation and M&A hopes. (4) That optimism unwound fast — two consecutive earnings disappointments were followed by the April-2025 tariff shock, and the stock halved. (5) The tariff pause produced a 15.0% single-day rebound and a 74% recovery. (6) First Brands Group filed for Chapter 11 on 2025-09-24; Jefferies’ Point Bonita Capital held ~$715m of purported receivables in what the bankruptcy estate alleges was a massive fraud, and the stock fell 30.6% as the market re-priced credit and fiduciary risk. (7) The deepest and most recent drawdown was litigation, not earnings: the Eugenia funds sued on 2026-02-26 for $18.4m and Western Alliance sued on 2026-03-06 for $126m, each landing on the day of a 9–14% single-session decline, taking the stock to $35.74 on 2026-03-12. (8) The recovery since is partly fundamental and partly strategic: Q1 and Q2 FY2026 net revenues rose 26.6% and 35.0% respectively and H1 diluted EPS rose 75%, but the rally began on 2026-03-24 when the Financial Times reported SMFG was weighing a takeover — the shares gapped to $41.02 and closed +2.5% on more than double normal volume before Bloomberg reported no immediate plan. (9) The June setback is the reminder that the credit overhang has not cleared: record investment-banking revenue was not enough to offset an earnings miss and a further 35% fall in asset-management revenue. Today’s price therefore discounts a cyclical peak plus a bid option, not a distressed franchise.

Price moves are facts drawn from the AZI daily price history; the attributed causes are interpretation, cross-referenced to earnings releases, 8-K filings and the litigation disclosed in the Q2 FY2026 10-Q. No recommendation or price target is expressed or implied in this section.

1. Executive Summary

Jefferies Financial Group is the last independent full-service investment bank of scale in the United States: an advisory and underwriting franchise bolted to a sales-and-trading operation and a $76–80bn balance sheet, with a residual Leucadia-era merchant-banking book still being liquidated. FY2025 net revenues were $7,343.8m, up 4.4%, of which Investment Banking contributed $3,790.3m (51.5%), Capital Markets $2,817.7m (38.4%) and Asset Management the $735.7m remainder. The firm employs 7,787 people and is run by Richard Handler (CEO since 2001) and Brian Friedman (President), an unusually long-tenured and unusually well-paid pair.

The share-gain story is real and is not the issue. Jefferies has taken global advisory fee share to 4.5% and moved from eighth to sixth in the league tables, displacing retreating European universal banks. (Two caveats belong on that claim: the rank is on a Dealogic basis that excludes China and Japan, and it is a fee-share rank — on deal value Mergermarket places Jefferies tenth globally, with an average US mandate of $1,180m against roughly $3,441m at Goldman Sachs.) In FY2024 alone it added roughly 110 basis points of share. It has done this while the independent advisory cohort took US M&A fee share from under 15% to over 27%. Management executed the Leucadia transformation competently and returned roughly $6bn of capital in the process.

The issue is that the economics do not reach the owners. In FY2025 Jefferies generated a pre-compensation profit pool of $4,731.3m. Compensation and benefits consumed $3,860.3m, or 81.6% of it. After tax and the participating preferred, common shareholders were left with $635.2m — 13.4%. Employees took 6.1x what shareholders took, and the split worsened from 78.4% the prior year in a year when revenue grew. The consequence is a return on tangible common equity of 8.5% in FY2025 and roughly 10.6% on a trailing-twelve-month basis at a cyclical peak, against an estimated cost of equity of 11–12% for a business with a 1.61 beta, no deposit funding and no access to a central-bank backstop. Ex-FY2021, the five-year average ROTE is approximately 8.2%. Jefferies has not earned its cost of equity in any year since 2021.

FY2025 was worse than the headline. Net revenues rose 4.4% but non-interest expenses rose 7.4%, so pre-tax earnings fell 13.4%. Diluted EPS fell only 5.7%, from $2.99 to $2.83, because the effective tax rate dropped from 29.2% to 21.2%. Held at the prior-year tax rate, FY2025 EPS would have been approximately $2.41 — a 19% decline. Separately, “Other investment banking” net revenues, which carry Jefferies’ 50% share of Jefferies Finance and 45% share of Berkadia, collapsed from $144.1m to $3.0m, accounting for close to the entire pre-tax decline on its own.

A share-count illusion has flattered every per-share metric. Between FY2023 and FY2024, SMBC exchanged 27,562,500 shares of voting common into 55,125 shares of Series B non-voting convertible preferred at a 500:1 ratio. Those shares kept participating in every dividend and automatically converted back into non-voting common in June 2026. They were never repurchased and SMBC’s economic ownership never changed. Adding them back, the FY2019–FY2025 diluted share-count reduction is 21.0%, not 29.7%, and — the more consequential point — as-converted tangible book value per share has moved from $36.96 at FY2022 to $37.87 today. Four years of essentially zero book-value compounding. Reported book value per share, which divides by the voting count alone, appears to have grown 14.8% over the same period. It did not.

What the market is paying. At $55.86 the shares trade at 1.22x as-converted book and 1.47x as-converted tangible book — 1.62x on management’s own “adjusted tangible book value per fully diluted share” of $34.55. Despite sitting 28.7% below the December-2024 high, the stock is in the 87.7th percentile of its own ten-year price-to-book range, because the multiple re-rated while the book did not compound. Jefferies traded between 0.73x and 1.00x tangible book from FY2019 through FY2023. On a standard P/TBV = (ROTE − g)/(COE − g) framework at an 11% cost of equity and 3% growth, today’s multiple embeds a sustainable ROTE near 15% — a level Jefferies has achieved once, in the FY2021 SPAC boom.

The balance sheet is sound, which is why this is not a short. Level 3 assets are $737.8m, just 1.0% of total assets and 8.6% of tangible equity. Average daily firm-wide VaR was $11.2m in FY2025, 0.13% of tangible equity. The liquidity buffer is $17.7bn, 23.9% of assets excluding goodwill. Ratings are Baa2/BBB/BBB+. The genuine risks are not in the trading book — they are in the principal-investment and credit-fiduciary businesses that VaR does not measure, which is precisely where First Brands hit.

First Brands is the live wound. Point Bonita Capital, a Leucadia Asset Management division, managed a roughly $3bn third-party trade-finance portfolio holding approximately $715m of purported First Brands receivables when First Brands filed for Chapter 11 on 2025-09-24, in what the bankruptcy estate alleges was a massive fraud. Jefferies’ own booked loss was modest — $30.0m pre-tax, since it had only $113m of the $1.9bn of invested equity — but the fiduciary damage is larger than the P&L damage. Asset management fees fell 21.1% in Q1 FY2026 and 27.0% in Q2 FY2026, “largely in respect of Point Bonita.” Litigation has followed: the Eugenia funds sued for $18.4m on 2026-02-26 and Western Alliance sued for $126m on 2026-03-06. No reserve or range of loss has been disclosed.

Governance deserves attention. Say-on-pay support has been 55%, 53%, 59%, 72%, 89% and 88% over six years. The single formulaic incentive metric is a ROTE computed on an equity base deflated 27.7% by stripping goodwill, the deferred tax asset and “the weighted average impact of cash dividends and share repurchases” — so the 10% target that pays 100% of the award corresponds to roughly a 6.5–7% GAAP ROE. In FY2023 Jefferies earned a 3.9% adjusted ROTE, below the plan’s own threshold, and the FY2023–25 cycle still paid about 81% of target. Meanwhile SMBC is permitted to accumulate to 20% of the economics while holding under 5% of the votes, with SMFG’s chief executive on the board and a Japan joint venture launching in January 2027.

No recommendation and no price target appear anywhere in this memo outside the clearly-labelled Claude’s Take block above. The body evaluates embedded expectations and scenarios only.

2. Business Overview

2.1 What the company actually is

Jefferies reports two segments, but the disclosure understates how different the pieces are.

Investment Banking and Capital Markets produced $6,608.0m of FY2025 net revenues, 89.9% of the total, and $982.1m of the $845.5m consolidated pre-tax profit — that is, more than 100% of group earnings, because the other segment loses money. It comprises four revenue lines:

Advisory is M&A, restructuring, debt advisory and private capital advisory — fee-for-outcome work with no contractual recurrence. Underwriting is equity and debt origination including leveraged loans and high yield. Other investment banking carries Jefferies’ 50% equity interest in Jefferies Finance (the leveraged-loan arranging and credit-management joint venture with MassMutual) and its 45% interest in Berkadia (the commercial real-estate finance joint venture with Berkshire Hathaway). These two joint ventures are among the best assets Jefferies owns, and their earnings are the most volatile line in the P&L — witness the fall from $144.1m to $3.0m in a single year.

Equities spans cash equities, electronic trading, derivatives, convertibles and prime services. Fixed income covers rates, credit, munis, securitised products and emerging markets. Both are intermediation businesses: they earn spread and commission for providing liquidity, and they require balance sheet and technology to compete.

Asset Management is, on the evidence, not really an asset-management business. It produced $735.7m of FY2025 net revenues but a pre-tax loss of $136.6m — the third consecutive annual loss, following −$44.1m in FY2024 and −$162.7m in FY2023, for a cumulative three-year pre-tax loss of $343.4m. Of the $735.7m, actual management fees were only $140.9m — 1.9% of group net revenues. The rest is investment return on the firm’s own capital ($177.8m) and “other investments,” the Leucadia legacy book ($467.5m), which houses Stratos, HomeFed, Tessellis and assorted principal positions. Reported “AUM” of $30.8bn is misleading: roughly 82% sits at affiliated managers where Jefferies takes only a revenue share, and directly-managed third-party AUM is approximately $2.46bn. The blended economics are correspondingly thin — roughly $57m of pure management fees on ~$31bn of AUM, about 18 basis points, with $62m of the $139m of total fees coming from revenue-share or profit-share participation rather than owned economics. This segment is a proprietary investment book wearing an asset-management label.

The unfinished conglomerate also still carries real people: 1,797 of Jefferies’ 7,787 employees — 23% of the firm — sit in the Stratos, Tessellis, HomeFed and M Science subsidiaries rather than in the investment bank. Notably, the 2024 investor deck disclosed non-core investments falling from $2.6bn to $0.7bn; the 2025 deck dropped the non-core disclosure slide altogether.

2.2 How it makes money, and what is recurring

Almost nothing is recurring. Advisory fees are contingent on transaction completion. Underwriting fees are episodic. Trading revenues depend on volumes and volatility. Management fees — the only genuinely recurring line — are 1.9% of net revenues. Deferred revenue at 2026-05-31 was $62.6m against $7.3bn of annual net revenue. This is a business that must be re-won every quarter, and that fact should govern how any single year’s earnings are capitalised.

2.3 The Leucadia unwind

Jefferies is the surviving entity of Leucadia National Corporation, a diversified holding company run by Ian Cumming and Joseph Steinberg, which acquired Jefferies Group in 2013 and took its name in 2018. The subsequent decade has been a methodical liquidation of the conglomerate: National Beef sold to Marfrig (2019, $890.8m), Spectrum Brands distributed (2019), Idaho Timber sold (2022, $239.3m, a $138.7m pre-tax gain), Vitesse Energy spun off (January 2023), Golden Queen sold at a $57.5m loss (2023), Foursight sold (April 2024, $24.2m gain), OpNet’s wholesale business sold to Wind Tre (2024, $322.8m cash but only a $3.5m gain), and Tessellis now under a binding offer with a $58.2m goodwill impairment taken in Q1 FY2026 and closing expected in Q1 2027. Stratos — acquired involuntarily in 2023 by foreclosing on Global Brokerage’s pledged equity — remains on the books with an explicit impairment warning.

The direction of travel is correct and mostly complete. But the tail is long, low-return and still producing writedowns seven years after the rebranding, and management has just begun re-diversifying: on 2025-12-09 Jefferies agreed to acquire 50% of Hildene Holding Company, an $18bn credit-focused asset manager, contributing its existing revenue share, part of its interest in a Hildene-managed fund and $340.0m in cash, with closing targeted for Q3 2026. No multiple was disclosed.

2.4 The SMBC relationship — increasingly the most important fact about the company

Since a 2021 strategic alliance, Sumitomo Mitsui Financial Group has become progressively more central. At 2025-11-30 SMBC owned 15.7% of the common on an as-converted basis and 14.3% fully diluted, accumulated partly through roughly $913m of open-market purchases. SMFG’s chief executive sits on the Jefferies board. SMBC has provided approximately $2.5bn of credit facilities. In September 2025 the two agreed a memorandum of understanding to form a joint venture in Japan taking over the principal parts of Jefferies’ wholesale Japanese equity research, sales and trading and ECM business, expected to begin January 2027. Simultaneously, an amended exchange agreement permits SMBC to increase its economic ownership to 20% while maintaining under 5% of the voting interest.

This is a slow-motion partial acquisition of the economics without the votes. It supplies Jefferies with distribution, balance sheet and credibility it could not build alone — a genuine benefit. It also entrenches incumbent management, dilutes minority economics, caps takeover optionality, and materially weakens the “last independent full-service bank” scarcity claim on which part of the bull case rests.

Verdict. A focused, competently-run, top-six investment bank attached to a money-losing principal-investment book and an unfinished conglomerate liquidation, becoming progressively more dependent on a single strategic shareholder. The business is far better than the one Leucadia owned in 2018. It is not, on this evidence, a business with recurring revenue or structural earnings power.

3. Industry Dynamics

3.1 The fee pool has round-tripped to a record

Peak-to-trough was −36%. The H1 2026 mix was DCM $25.2bn (+8%), M&A $24.0bn (+18%), loans $17.3bn (+5%) and ECM $13.5bn (+60%). This is a cyclical high, not early innings.

Critically, the M&A boom is the narrowest on record. H1 2026 announced volume was $2.8tn, up 48% and the highest first half since LSEG records began in 1980 — but deal count fell 9% to roughly 24,000, the weakest count in six years, and 47 deals above $10bn accounted for 46% of all value. This is a megadeal barbell funded by record investment-grade issuance. The middle market and the sponsor-exit engine — Jefferies’ natural hunting ground — remain comparatively soft.

3.2 Concentration: the middle is squeezed from both ends

Top-five fee share rose to roughly 35% in the first five months of 2026 from about 32% a year earlier. At the same time, independent advisory firms took US M&A advisory fee share from under 15% in 2018 to over 27% in 2024, and Evercore and RBC recently entered the global top ten by displacing Deutsche Bank and UBS. These facts reconcile: the pool is barbelling. The scale players and the boutiques both gain; the casualty is the second-tier universal bank.

Jefferies is the most successful occupant of that squeezed middle. But the distinction matters for the thesis: Jefferies is taking share from retreating Europeans, not from Goldman Sachs, JPMorgan or Morgan Stanley. Its gains and the top five’s gains are funded from the same casualty pool, and that pool is finite.

3.3 Private credit — a permanent partial loss, intelligently hedged

Private-credit AUM has roughly tripled to about $1.7tn and is forecast near $2.6tn by 2029. Direct lending now matches the broadly-syndicated loan market at $1.5–2tn and finances roughly 85% of leveraged buyouts by count. The structural consequence for Jefferies is that a meaningful share of the leveraged-finance arrangement fee that once flowed to underwriters now accrues to direct lenders. The tell is in Jefferies’ own numbers: in H1 FY2026, against record investment-grade issuance and a booming fee pool, debt underwriting revenue fell year on year ($342m vs $405m) while advisory and ECM surged.

The league-table evidence points the same way and is more damning. Leveraged finance is the one product where Jefferies’ share went backwards — its global single-B levfin share fell from 3.3% in FY2015 to 2.4% in FY2023, with a rank of #8 rather than the #6 it carries elsewhere, and that is on the ex-China/Japan basis. The 2025 investor deck deleted the product-share slide that disclosed it, and no global leveraged-finance rank has been disclosed since. The widely-quoted “#3 in leveraged finance” is triple-qualified in the endnotes: US only, sponsor-backed left-lead only, and deals under $1.75bn only. A firm withdrawing disclosure in a franchise it describes as core, in the same period that private credit is taking that franchise’s economics, is telling you something.

Jefferies’ answer is structurally intelligent and is the single best feature of the model: through Jefferies Finance and Jefferies Credit Partners it converts a one-time arrangement fee into a recurring management fee on the same loan, with proprietary sourcing — the 10-K states that “direct lending investments are primarily sourced through Jefferies.” If a durable advantage exists anywhere in this company, it is here.

Three caveats bound the optimism. First, scale: Ares runs $274bn of direct lending and Blue Owl $158bn of credit; Jefferies’ platform is an order of magnitude smaller in a business with real scale economics. Second, the asset class is in a Marathon-style bust right now — Fitch’s US private-credit default rate hit a record 6.0% in April 2026, the Proskauer index rose to 2.73% in Q1 2026 from 1.84% two quarters earlier, and sector-wide redemption gating has appeared. Third, Jefferies has already taken a live fraud loss in exactly this area.

There is a genuine offsetting positive: direct lenders are now capital-constrained just as banks return to underwrite-and-distribute, which hands share back to arranging desks. We treat that as a cyclical swing-back rather than a structural reversal.

3.4 Regulation — and why the non-bank charter is a shrinking advantage

Jefferies is not a bank holding company. Its FY2025 10-K contains zero occurrences of “Federal Reserve” and zero of “discount window.” It is supervised by the SEC, FINRA and the CFTC, with capital governed by Rule 15c3-1’s alternative method. There is no CCAR, no stress capital buffer, no G-SIB surcharge, no Basel risk-weighted-asset regime and no supplementary leverage ratio. Funding is $12.5bn of unsecured long-term debt plus $10.6bn of equity, a 1.17:1 long-term capital structure with a 7.4-year weighted-average maturity, and zero deposits — against $501bn at Goldman Sachs and $419bn at Morgan Stanley.

For years this was a real advantage: competitors were capital-constrained and Jefferies was not. That advantage is closing fast. The March 2026 reproposal formally rescinded the 2023 Basel III Endgame framework; where the 2023 draft sought roughly a 19% capital increase, the 2026 reproposal delivers approximately $87.7bn of net capital relief, with finalisation expected late 2026. Goldman has already drawn CET1 from 14.3% to 12.5% on a record $5bn quarterly buyback.

When the constraint on your competitor is removed, your freedom from it stops being worth anything — while your funding disadvantage and your missing lender of last resort persist unchanged. Jefferies pays a standing return drag to self-insure through a large liquidity buffer, and carries a permanent pro-cyclical ratings and spread gap; its own 10-K models a two-notch-downgrade collateral call. Do not underwrite the charter as a moat. Separately, the Treasury clearing mandate (cash 2026-12-31, repo 2027-06-30) is a mild negative for a non-deposit-funded repo book — a cost of business, not a thesis driver.

3.5 Cyclicality — Jefferies has roughly 1.4x the industry’s amplitude

Jefferies’ peak-to-trough decline of 50.9% exceeded the pool’s 36%. The variance sits entirely in underwriting: in FY2022 advisory fell just 4.8% while underwriting collapsed 58.7%. This inverts the common assumption — leveraged finance and ECM are the volatile core, advisory is the ballast — and leveraged finance is precisely what private credit attacks. In FY2023 pre-tax earnings fell 66.4% on a 21% revenue decline: roughly 3x downside operating leverage. A single year’s earnings therefore carry very little information about through-cycle earning power, and the mid-2026 trap is the double-count of near-record earnings on a re-rated multiple.

3.6 Marathon capital-cycle read: two cycles, out of phase

On the advisory and underwriting side, supply is unusually disciplined. The 2022–23 downturn forced deep front-office cuts, and the 2026 rebuild has been deliberately narrow — selective senior hiring rather than team lifts, flat base salaries, negotiation displaced into sign-on bonuses and rising deferrals. A +17% fee pool against a headcount base still below 2021 is a favourable supply configuration: high returns have not yet attracted proportionate capacity, and comp ratios could in principle fall further. That is the bull case, and it is legitimate.

Against it: Morgan Stanley, Bank of America, RBC and UBS each roughly doubled external managing-director hiring in 2025 versus 2024, and JPMorgan, Citi and Barclays were up sharply — capacity is entering at the point of maximum expected return, which is the classic Marathon signal. Wells Fargo was the largest US net hirer of 2025 and has told the Financial Times it will add 25–30 managing directors in each of 2026 and 2027, on top of 125 investment-banking MDs hired since 2019, explicitly targeting healthcare, technology, industrials and financial sponsors — Jefferies’ own franchise. The boutiques are building hardest of all: Evercore took investment-banking senior managing directors from 144 to 171 in a single year, up 19%; Moelis is at 178 MDs, up 95 over five years; PJT at 133 partners, up 12%.

And Jefferies has stood down precisely as this happens. Its net managing-director additions ran +50, +45, +20 and then just +5 in FY2025, with a stated 2026 ambition of roughly +37. Jefferies bought talent cheaply in 2021–23 when rivals were cutting — correct counter-cyclical behaviour, and to management’s credit. It is now adding almost nothing while everyone else bids. Whether that is discipline or franchise cession is the live question, and the FY2025 10-K offers an unsettling hint: the sentence “we have increased the number of our Investment Banking Managing Directors and related staff,” present in both the FY2023 and FY2024 filings, was dropped from the FY2025 10-K. Jefferies’ own balance sheet grew 18.1% against 4.4% revenue growth in FY2025, with leverage rising from 6.3x to 7.1x: the asset-growth anomaly in plain sight.

The adjacent pool, private credit, is a textbook Marathon bust already in progress: capital tripled, spreads compressed roughly 100bp from 2021, time-in-market at a post-2008 high, the late-cycle retail-distribution signal fired, and reversion arriving on schedule. For an integrated arranger-and-lender this cuts both ways, and on balance we expect the arranging franchise to benefit over the next four to six quarters. The signal to watch is private-credit fundraising, not investment-banking headcount.

3.7 Verdict: structurally bad

Not marginally — decisively. Greenwald’s market-share-stability test fails: boutiques moved from under 15% to over 27% of US advisory fees in six years, Evercore and RBC displaced Deutsche Bank and UBS, and Jefferies itself moved from eighth to sixth. Share that mobile is definitional evidence of absent barriers to entry; in advisory the entry cost is two managing directors and a Bloomberg terminal. The ROIC test fails too: through a cycle, a balance-sheet investment bank earns roughly its cost of equity at best.

The most damning structural feature is labour capture. Where the scarce asset walks out of the building every evening and can be hired by a competitor tomorrow, the rent accrues to the rainmaker rather than to capital. Evercore pays 64.9% of revenue to bankers before shareholders see a dollar; Lazard 67.3%; Jefferies 52.6% on a bank-like cost base that leaves it taking 81.6% of the pre-comp pool. Add the absence of recurring revenue, roughly 3x downside operating leverage and a 36–48% peak-to-trough fee pool, and earnings in this industry deserve a permanent capitalisation discount.

Two narrow exceptions exist, neither fully available to Jefferies. Scale capital markets — jumbo underwriting, prime brokerage, securities financing, clearing — has genuine economies of scale, and that is exactly where the pool is consolidating. And the current cycle position is favourable, which is the thing most often mistaken for a moat at precisely the wrong moment. The industry favours the extremes and punishes the middle, which is where Jefferies lives.

4. Competitive Position

4.1 The share gain is real — start by conceding it

Jefferies’ own FY2025 Annual Report reports global advisory fee share of 4.5%, ranking sixth, behind Goldman Sachs (10.9%), JPMorgan (8.6%), Morgan Stanley (5.9%), Bank of America (4.7%) and Citi (4.7%). In global ECM it ranks sixth at 4.2%. Dealogic ranks it seventh by total 2025 fees. In 2024 alone it added roughly 110 basis points of share and moved from eighth to sixth. In equity fees it ranks seventh, against twenty-third in 2010. On Jefferies’ own disclosed series, global investment-banking fee share went from 2.7% in 2019 to 3.8% in 2024 to roughly 4.1% annualised in 2025, and the league-table rank from ninth to sixth; global equities cash market share rose from 2.9% to 4.9% over the same period.

Two qualifications materially soften these numbers, and both come from Jefferies’ own footnotes. First, every “top-six global” rank Jefferies publishes is on a Dealogic basis that excludes China and Japan — stated in the endnotes to both the 2024 and 2025 investor decks and in the FY2025 shareholder letter (“Per Dealogic… excludes China and Japan”). On a genuinely global basis the ranks are worse. Second, these are fee-share ranks. On deal value, Mergermarket’s independent FY2025 tables place Jefferies tenth globally and tenth in the US, and its global rank actually slipped from ninth to tenth in 2025 even as its advised value rose 15.1% — because 2025 was a megadeal year (LSEG counted 70 megadeals, the most since records began in 1980) and Jefferies structurally under-participates in that cohort.

The clearest single number is average mandate size. In US M&A in FY2025, Jefferies advised on $284.4bn across 241 deals — an average deal of $1,180m, against roughly $3,441m at Goldman Sachs, $3,498m at JPMorgan and $3,708m at Morgan Stanley. Jefferies’ average US mandate is about a third the size of the bulge bracket’s, though four to six times the true mid-market shops (Houlihan Lokey $197m, Piper Sandler $271m). It occupies a real and distinct band — but it is not competing for the same transactions as the firms it is ranked alongside on fee share. A firm that has quadrupled its position in the league tables over fifteen years has done something real, and any analysis that begins by denying it is not serious.

4.2 But name the moat, or concede there isn’t one

Testing Jefferies against Greenwald’s three genuine barrier types:

Customer captivity — absent, and the 10-K says so. The risk factors state that “if we were to lose the services of certain of our professionals… some of our clients could choose to use the services of a competitor.” That is a written admission that the client relationship attaches to the banker, not to the firm. Advisory mandates are awarded per transaction. Recurring revenue is 1.9% of net revenues. There are no switching costs because there is nothing to switch from.

Cost advantage — inverted. The Investment Banking and Capital Markets segment compensation ratio is 54.7%, against Goldman Sachs at 31.8%. Jefferies is not a low-cost producer; it is a high-cost one. Revenue per employee is roughly $943k.

Economies of scale with captivity — second-tier and increasingly borrowed. Genuine scale economics exist in capital markets — jumbo underwriting, prime brokerage, securities financing, clearing — and Jefferies has real capability there. But it is a fraction of the scale of the top five, and the scale it is adding is increasingly SMBC’s: 15.7% of the equity heading to 20%, a board seat plus a second nominee, roughly $2.5bn of credit facilities, and a Japan joint venture that transfers the Japanese equities business to a jointly-controlled vehicle from January 2027.

The “talent is the asset” bull case collapses on inspection. It is true that Jefferies is a destination for senior bankers leaving bulge-bracket platforms. But revenue per core banker is roughly $1.10m against Goldman’s ~$1.21m — productivity is comparable, not superior. What differs is the split: Jefferies pays out 54.7% and Goldman 31.8%. The franchise generates economic rent; the rent goes to the employees. Talent that must be re-purchased at market price every year is a cost line, not a barrier to entry. It is a treadmill.

4.3 The comparison that settles it

Jefferies’ FY2025 advisory revenue was $2,145m. Evercore’s FY2025 advisory revenue was $3,267m. A 7,787-person global investment bank with a $76bn balance sheet did 34% less advisory revenue than a roughly 2,400-person boutique with net cash — and Evercore earned about 31.7% on equity doing it, against Jefferies’ 6.1–7.3%.

*Lazard figure is the pre-comp pool. Peer figures are drawn from each company’s FY2025 filings and from EDGAR XBRL compensation lines; Jefferies’ figures are computed from its FY2025 10-K.

Jefferies has the worst return on equity in the set by a factor of more than two, and the worst pre-tax margin. The structural reason is visible in the table: Jefferies has the cost structure of an advisory boutique and the capital intensity of a balance-sheet bank, and earns the returns of neither. The boutiques pay out more — Evercore 64.9%, Lazard 67.3% — yet earn 19–32% on equity because they require almost no capital; Houlihan Lokey’s entire tangible equity is roughly $742m. Jefferies pays out 52.6% of revenue and carries $8.5bn of tangible equity against $76–80bn of assets. It absorbs the boutique’s labour economics and the bank’s capital burden simultaneously.

4.4 The Greenwald ROIC test, applied

A moat must show up as sustained excess returns. Jefferies’ return on equity by year: FY2020 11.8%, FY2021 22.3%, FY2022 9.2%, FY2023 3.1%, FY2024 8.1%, FY2025 6.9–7.3%. Six-year average roughly 10.3%; excluding the FY2021 SPAC boom, 7.9%. On tangible equity, ex-FY2021, roughly 8.2%. Against a cost of equity of 11.4–12.2% implied by a 1.608 beta with negative alpha, Jefferies has destroyed on the order of $200–350m of value per annum during the very period in which it was gaining market share.

That is the central paradox of this company and the spine of the memo: a business that gains share for a decade and still cannot earn its cost of capital does not have a moat; it has a growth habit financed by its shareholders for the benefit of its employees.

Verdict. No durable competitive advantage. What exists is a genuine, valuable, firm-level reputational intangible — the Jefferies brand opens doors that a two-person boutique cannot — but it is not a barrier to entry, it does not create captivity, and it does not convert into excess returns on capital. The one place a real advantage could form is the proprietary-sourcing loop between Jefferies Finance’s arranging desk and Jefferies Credit Partners’ recurring-fee credit funds. That is genuinely clever, and it is sub-scale, and it has just taken a fraud loss.

5. Growth History and Forward Opportunities

5.1 The growth record, honestly stated

Jefferies has not regained its FY2021 peak of $8.01bn six years later. More pointedly, core investment banking revenue — advisory plus ECM plus DCM — was $3,787m in FY2025 against $4,366m in FY2021, a 13% decline, achieved on roughly 20% more core headcount. Total headcount rose from 4,800 in FY2019 to 7,787 in FY2025, a 62% increase. Revenue per head has gone backwards.

That said, H1 FY2026 is genuinely strong: net revenues of $4,224m, up 30.9%, with Q2 investment banking revenue of $1,207m up 57% and record equities of $601m. Annualised, FY2026 investment banking would exceed $4.4bn and approach the FY2021 record. This is a real cyclical recovery and the memo should not minimise it.

5.2 Organic versus acquired, and the quality question

Growth has been overwhelmingly organic — hiring bankers — with the notable exception of the pending Hildene transaction. That is the higher-quality path in principle. But it is expensive: the compensation ratio rose from 52.0% to 52.6% to 53.8% in H1 FY2026 even as revenue grew 31%, so the incremental revenue is arriving at a worse margin than the average. In a period of 31% revenue growth, a rising compensation ratio is the definition of low-quality growth.

The per-head evidence says the same thing. Investment-banking managing directors went from 212 in 2019 to 369 in FY2025, up 74%, while investment-banking net revenue rose 124% — so revenue per managing director went from $8.0m to $10.3m, up 29% over six years, which is roughly cumulative US inflation across the same window. In real terms, six years of senior hiring has produced approximately no productivity gain per head. And the hiring engine is stalling: net managing-director additions ran +50, +45, +20, +5 across FY2022–FY2025. All of the H1 FY2026 margin improvement came from the non-compensation line (39.4% to 31.8%), which is operating leverage on fixed costs, not on the variable cost that matters.

5.3 Forward opportunities — and what they are worth

The M&A upcycle. Real, and Jefferies is levered to it. But the H1 2026 boom is a megadeal barbell — deal count is at a six-year low — and Jefferies’ natural franchise is the middle market and sponsor exits, which remain soft. Jefferies’ own deck concedes the point: the 2025 fee wallet rose 11%, but the sponsor wallet rose only 4%, under the deck’s own heading “Sponsor Activity Yet to Return to Growth.”

The sponsor franchise is genuinely strong and it deteriorated in 2025 — a distinction the headline rank conceals. Jefferies reports a sponsor M&A rank of #2, improved from #3; but its sponsor share fell from 8.4% to 7.8% over the same period. The rank improved because a rival fell back, not because Jefferies gained. Independently, Mergermarket shows Jefferies #2 in US buyouts by deal count — a real position, behind only Houlihan Lokey on count — but its sponsor exit business went backwards in a strong year: global exits fell 18.1% by value and eleven deals, dropping from #3 to #9 by value and #2 to #6 by count. Sponsor exits are the single line most levered to the M&A recovery thesis, and it is the line that weakened.

The SMBC alliance. Genuinely valuable: distribution, balance sheet, joint origination in Europe, the Middle East, Canada, Asia and Australia, and revenue-sharing on investment banking transactions. It is also the mechanism by which an outside party acquires 20% of the economics.

Private credit via Jefferies Credit Partners. The single most structurally intelligent element of the model — converting one-time arrangement fees into recurring management fees on loans “primarily sourced through Jefferies,” with a BDC anchored by ADIA at roughly $1.7bn. This is where a durable advantage could form. It is sub-scale against Ares’ $274bn and Blue Owl’s $158bn, and its credibility has just been damaged.

Hildene. On 2025-12-09 Jefferies agreed to acquire roughly 50% of Hildene for approximately $340m of cash plus ~$75m of contributed equity, with up to $100m more in convertible preferred, partly to fund Hildene’s acquisition of SILAC, an insurance business. This is the first large cash acquisition of the post-Leucadia era and it moves the firm back toward diversification — into insurance and annuity reinsurance — immediately after a credit blow-up. No multiple was disclosed.

Verdict: low-quality growth. The share gains are real, the cyclical recovery is real, and the strategic direction of the credit platform is sound. But revenue per head has fallen, the compensation ratio is rising into a boom, the FY2021 peak remains unbeaten six years on, and the newest growth initiative is a re-diversification into insurance. Growth that does not improve returns on capital is not creating value; on the evidence of the last six years, Jefferies’ growth has not improved returns on capital.

6. Financial Quality

6.1 The compensation ratio and the profit split — the single most important economics

The FY2021 and FY2022 ratios in the low 40s are artefacts: those years carried large merchant-banking revenues (Idaho Timber, Vitesse) that arrive with cost of sales rather than compensation. The clean read is the Investment Banking and Capital Markets segment ratio: 53.3% in FY2023, 55.1% in FY2024, 54.7% in FY2025.

The economically correct way to see this is the split of the pre-compensation profit pool:

FY2025 net revenues $7,343.8m − non-compensation expense $2,612.5m = pre-compensation profit pool $4,731.3m Of which: employees $3,860.3m (81.6%) · tax $184.6m (3.9%) · SMBC participating preferred $79.7m (1.7%) · common shareholders $635.2m (13.4%)

Employees took 6.1x what shareholders took, and the ratio worsened from 78.4% in FY2024 in a year when revenue grew 4.4%. Goldman Sachs pays 47.7% of its pre-comp pool to employees; Morgan Stanley 56.7%. Jefferies pays 81.6% — a boutique payout ratio out of a bank cost structure.

One under-appreciated detail: FY2025 compensation included $621.5m of share- and cash-award amortisation, but only $88.2m ran through additional paid-in capital as stock-based compensation. Roughly 86% of Jefferies’ deferred compensation is cash-settled — a real, funded future cash claim, not a non-cash accounting charge.

6.2 Returns — the heart of the matter

ROTE is computed as net earnings attributable to Jefferies — that is, to all common-equivalent holders including the participating preferred — over average tangible common equity. This is the internally consistent construction: because SMBC’s Series B preferred sits inside shareholders’ equity, its earnings must sit in the numerator. Computing earnings excluding the preferred over equity including the preferred’s capital understates ROTE by roughly a point, and should be avoided.

Ex-FY2021, the five-year average ROTE is approximately 8.2%. The cost of equity for a business with a 1.608 beta, a Baa2/BBB rating, no deposits, no discount-window access and 7.5x balance-sheet leverage is on any reasonable construction 11–12%; even assuming a beta of 1.0 it is roughly 9.2%. Jefferies has not earned its cost of equity in any year since FY2021, and it does not earn it today at a cyclical peak with revenues up 31%.

Management’s own preferred metric does not rescue this. The reported “return on adjusted tangible shareholders’ equity” of 10.1% is computed on a denominator deflated roughly 27.7% — GAAP equity of $10,157m less goodwill and intangibles of $2,054m, less the deferred tax asset of $498m, less $259m for “the weighted average impact of cash dividends and share repurchases” — giving $7,346m. Even on that self-written measure, 10.1% is still below the cost of capital. Jefferies cannot clear its hurdle rate even when it writes the hurdle. The metric is also not peer-comparable and must be restated before being set against Goldman’s or Morgan Stanley’s ROTE.

An analyst put the point directly at the October 2025 investor meeting. Brian Grefe of Raymond James: “since 2013, ROTE has only exceeded 11% in three of the last 12 years… can this business sustainably earn a double-digit return on equity?” Friedman’s answer was that the period was “aberrational” and that “I won’t be able to maybe prove it for 10 more years.” James Yaro of Goldman Sachs noted he had asked about the “low teens ROTE” target in five consecutive years.

6.3 Book value — the illusion and the reality

Reported book value per share has grown 14.8% since FY2022. As-converted, it has grown 1.1%. Reported tangible book per share has grown 14.3%; as-converted, 0.7%. Essentially all apparent per-share book growth since FY2022 is the mechanical effect of SMBC exchanging 27,562,500 common shares into non-voting preferred at a 500:1 ratio — removing shares from the denominator without removing a dollar of capital, while those shares continued to participate in every dividend and then converted straight back into non-voting common in June 2026. The company’s own “adjusted tangible book value per fully diluted share” — $32.36, $33.69, $34.55 — is the honest series, and it compounds at roughly 3–4% a year.

Note that widely-used data services report book value per share on the voting count alone. That produces roughly $47.64 and a price-to-book near 1.17x. The correct denominator is 230,552,271 common-equivalent shares — 193,742,690 voting plus 36,809,581 non-voting at 2026-06-30 — giving book value per share of $45.83 and tangible book of $37.87.

6.4 Quality of earnings — four distortions to normalise

(a) FY2025’s decline was masked by tax. Net revenues rose $308.9m; compensation rose $200.7m and non-compensation expense rose $242.8m, so pre-tax earnings fell $134.6m and the pre-tax margin fell from 14.3% to 11.9%. Income tax then added back $108.6m as the effective rate fell from 29.2% to 21.2%. At the prior-year tax rate, FY2025 diluted EPS would have been approximately $2.41 rather than $2.83 — a 19% decline, not 5.7%.

(b) The joint-venture earnings vanished without explanation. “Other investment banking” net revenues fell from $144.1m to $3.0m — roughly the entire pre-tax decline. Jefferies attributes this to the prior year’s inclusion of Foursight and the gain on its April 2024 sale, mark-to-market losses on investment positions, and lower Jefferies Finance performance partly offset by better Berkadia. It does not decompose the figure. Given that this line carries the 50% Jefferies Finance and 45% Berkadia interests, its opacity in a year of leveraged-finance stress is unsatisfactory.

© FY2019’s reported $3.03 EPS is not a real number. Income tax was a benefit of $484m against pre-tax income of $479m, driven by a one-off lodged-tax-benefit realisation of roughly $545m. Normalised FY2019 EPS was approximately $1.30–1.40. Any multi-year growth series anchored on FY2019 EPS is meaningless.

(d) Presentation-basis breaks. FY2020 and FY2021 were restated in the FY2022 10-K for the Jefferies Group merger presentation; FY2019 was not. FY2023 was restated for discontinued operations (Vitesse, Foursight, OpNet), reducing reported net revenues to $4,700.4m. Growth series must state their basis.

6.5 Balance sheet, leverage and risk

Total assets rose 48% in two years and leverage rose 50% in four. In FY2025 the balance sheet grew 18.1% against 4.4% revenue growth. That is Marathon’s asset-growth anomaly in plain sight, and it means the return recovery has been bought with balance sheet rather than earned with margin.

The decisive proof is that return on tangible assets fell while return on tangible equity rose. Return on tangible assets declined from 1.07% in FY2024 to 0.85% in FY2025. On Jefferies’ own disclosed tangible gross leverage measure, the ratio has gone 6.2x (FY2019) → 6.8x (FY2021) → 7.7x (FY2024) → 8.7x (FY2025) → 9.0x (Q2 FY2026). The asset base is earning less per dollar; the equity return is being held up by gearing it harder.

This directly answers the question Jefferies declined to answer. At the October 2025 investor meeting Brian Grefe of Raymond James asked, on the record: “Without a commensurate increase in leverage, can this business sustainably earn a double-digit return on equity?” Friedman replied that the period had been “aberrational” and offered no driver and no number. The balance sheet answers it: the leverage increase has already happened. Note too that Friedman described 7–8x as “the range” at that same meeting, with above-8x reserved for low-risk exceptions — the ratio was already roughly 8.4x when he said it, reached 8.7x six weeks later and 9.0x two quarters after that. Handler said Jefferies ran “like 20% below” the agencies’ ~10x comfort level; at 8.7x the margin was 13%, and at 9.0x it is 10%.

The reassuring side of the ledger is genuine and should be stated plainly, because it is why this is not a solvency story:

• Level 3 assets are $737.8m — 1.0% of total assets and 8.6% of tangible equity. Any bear case built on mark-to-model opacity in the trading book is simply wrong.
• Average daily firm-wide VaR was $11.23m in FY2025 (high $16.03m, low $7.60m; $8.96m at year-end), down from $13.13m in FY2024, and equity-dominated. That is 0.13% of tangible equity. The trading book is not where the risk is.
• Liquidity buffer $17.7bn, 23.9% of assets excluding goodwill. Long-term debt $15.9bn with a 7.4-year weighted-average maturity; long-term capital $23.1bn against $12.5bn of unsecured long-term debt, a 1.17:1 structure. Repo $12.2bn, other secured financings $2.9bn. No deposits.
• Ratings: Moody’s Baa2 (stable, affirmed 2025-11-14), S&P BBB (stable, affirmed 2024-04-23), Fitch BBB+ (stable, affirmed 2026-01-23). No agency downgraded or moved to negative outlook after First Brands or MFS — and, contrary to a common assumption, none has upgraded in two years either.

The important inference is that Jefferies’ risk is not measured by its risk systems. VaR of $11m tells you nothing about a $715m concentration in purported receivables from a single fraudulent servicer, or a £103m warehouse facility secured on double-pledged property. The losses came from the principal-investment and credit-fiduciary businesses, which sit outside the trading VaR framework entirely.

One further item: operating cash flow is structurally negative (−$1,495m in FY2025) because a growing dealer funds inventory and receivables. That is normal for the model and must not be presented as a free-cash-flow failure; conventional FCF metrics are meaningless here and should not be quoted.

Verdict: economics do not improve with scale. Over six years Jefferies added 62% more employees, 48% more assets in two years, and 110 basis points of fee share — and its compensation ratio rose, its pre-tax margin fell from 14.3% to 11.9%, its as-converted tangible book per share went nowhere, and its return on tangible equity stayed below its cost of capital throughout. The balance sheet is sound and the liquidity is ample. The returns are not there.

7. Capital Allocation

7.1 The buyback: two eras, opposite grades

Cumulative FY2018–FY2023: roughly 157.7m shares retired at an average $23.91 — approximately 0.6–0.7x book. That is excellent capital allocation and is the single largest source of per-share value created at Jefferies in the last decade. Full credit is due.

The recent record is the opposite. No open-market repurchase at all in FY2024 or FY2025 despite a live $250m authorisation. The only stock retired at the all-time high was 0.7m shares at $79.57, bought to fund employees’ tax withholding on vesting equity awards. Shares outstanding actually rose in FY2025, from 205,504,272 to 206,296,167.

The month-by-month record around the low, from the 10-Q Item 5 tables, is the sharpest available test:

Dec-2025 $61.16 (nil under programme) · Jan-2026 $62.05 (450k) · Feb-2026 $56.78 (2,048k) · Mar-2026 $37.52 — ZERO under programme · Apr-2026 $47.99 (2,166k) · May-2026 $51.86 (334k)

Jefferies bought at $62 and $57, bought nothing in the month the stock bottomed at $35.74, then resumed at $48–52. A fair caveat: March overlapped the Q1 earnings blackout. But Jefferies operated no Rule 10b5-1 repurchase plan, which would have permitted buying through the blackout. That was a choice. And the authorisation is reset at $250m each quarter rather than raised, which caps the firm’s ability to act decisively on weakness.

7.2 The share-count decomposition: roughly a third of the shrinkage is accounting

Weighted diluted shares fell from 317.0m in FY2019 to 222.7m in FY2025, apparently −29.7%. The components:

• Real cash buyback: $2,218.1m of treasury purchases FY2020–FY2025, of which roughly $1,069m was open-market.
• SMBC preferred reclassification — not a buyback: 27,562,500 common shares exchanged into 55,125 Series B preferred (21.0m in FY2023, 6.56m in FY2024) at 500:1, with SMBC paying $1.50 per share (about $41m). Those shares participated in every dividend throughout and converted back into non-voting common in June 2026.
• Vitesse did not reduce the share count. The January 2023 spin-off was an in-kind distribution charged $526,964k to retained earnings — it distributed value, it did not retire stock.
• Employee issuance offset: roughly +9.79m shares over FY2023–FY2025. Stock-based compensation rose from $45.4m to $63.1m to $88.2m, up 94% in two years.

Adding SMBC’s shares back, the true economic diluted count in FY2025 is roughly 250.3m and the real reduction is 21.0%, not 29.7%. Nearly a third of the apparent shrinkage never happened.

7.3 The preferred is a control instrument, not capital

It bears repeating because it is widely misread: the $79.7m of “preferred stock dividends” in FY2025 is not a cost of capital. It comprises $44.1m of actual cash paid to SMBC — exactly $1.60 per as-converted share, identical to the common dividend — plus roughly $35.6m of allocated undistributed earnings under the two-class method. Earnings per share are denominator-neutral: $630,791k ÷ 215,096k = $2.93, and adding back $79,684k over 27,563k additional shares gives $710,475k ÷ 242,659k = $2.93, identical.

Its only financial consequence was to conceal roughly 12% of the share count for three years. Its real purpose is governance: it lets SMBC hold up to 20% of the economics with under 5% of the votes, with SMFG’s chief executive on the board. This entrenches incumbent management, dilutes minority economics, and caps takeover optionality — the classic structure by which a strategic partner acquires the upside without paying a control premium.

7.4 Executive compensation, and a performance metric set below the cost of capital

“Compensation actually paid” to Handler under Item 402(v) across FY2021–FY2025 totals $282.3m. The FY2022 spike reflects the December 2021 “Leadership Continuity Grant” of $25.0m each (1,743,984 RSUs, five-year vest plus three-year hold) landing alongside the FY2021 annual grant; the board committed to no further one-time awards while it is outstanding and has honoured that.

Say-on-pay support, computed from the Item 5.07 vote-result 8-Ks:

Five consecutive years below 90% and three below 60% is a sustained shareholder objection, not a blip. (ISS and Glass Lewis recommendations could not be sourced and are not asserted.)

The most important governance finding is that the incentive metric is set below the cost of capital. The only formulaic metric is return on adjusted tangible equity: a 7.5% threshold pays 75% of the award, a 10% target pays 100%, and 15% pays the 150% maximum. But the denominator is the deflated one described in the relevant section — 27.7% below GAAP equity. A 10% “target” on that base corresponds to roughly a 6.5–7% GAAP return on equity; the 7.5% threshold that still pays 75% corresponds to roughly 5%. Only at maximum does management earn approximately its cost of capital. The softness is demonstrable: FY2023 adjusted ROTE was 3.9%, below the plan’s own threshold, and the FY2023–25 cycle nonetheless paid about 81% of target — Handler earned 102,547 performance share units and forfeited only 23,742, worth $1.468m.

This grid is also, in practice, Jefferies’ only stated return target — and the company concedes it has been missing it. Management refuses to give public return guidance; Handler told the 2024 shareholder day “we don’t give ROE expectations, because quite honestly, every three to four years the world seems to blow up.” The nearest thing to a public aspiration is Friedman’s negative formulation in October 2025 — that if the firm is right, “you will see it and you’ll see it not at low double digits.” Against that, the 2026 proxy states plainly: “the three-year ROTE in the fiscal 2022–2024 performance period and fiscal 2023–2025 period fell short of the target so that the PSUs tied to those performance periods were earned at a level below target.” On Jefferies’ own reported measure the FY2023–25 three-year average ROTE is 8.3% — below the 10% target and barely above the 7.5% floor. Rebuilt on unadjusted tangible common equity it is roughly 7.3%, below the level at which the board forfeits all performance stock.

7.5 Insider behaviour: $246m of selling, $1.9m of buying, none planned

Across the trailing 60 months there were seven code-P open-market purchases, only four by an individual:

Against that, $246,441,905 of selling across 25 transactions: Friedman $103,998,673 (12 sales), Handler $94,153,260 (2 sales — 1.5m shares at $43.50 on 2024-04-24 and 400,000 at $72.26 on 2024-11-06), Steinberg $44,902,560, Beyer $2,960,200, others $427,212.

Two features make this material. First, the aff10b5One XML element is false on every one of the 25 sales — none was made under a Rule 10b5-1 plan, so every sale was discretionary. Both FY2026 10-Qs confirm no officer or director has adopted, modified or terminated such a plan. Second, roughly $89m was sold by Handler, Friedman and a director between 2024-10-10 and 2024-11-06 — six weeks before the all-time high of 2024-12-16.

Handler and Friedman have never bought a single share on the open market in five years. Not during the October 2025 drawdown, when Handler called the stock “kind of cheap” at $47.72; not during the collapse to $35.74 in March 2026. The three other code-P filings are SMBC’s ($1.18bn across three tranches), and those are not conviction purchases either — each footnote discloses a price “subject to adjustment following an approximately two-month reference period” under a pre-existing agreement with a third party, i.e. a pre-agreed VWAP-referenced programme executing the ~13m-share commitment announced on 2026-02-11.

Ownership: Handler holds 17,398,269 shares (7.8%) and Friedman 5,073,875 (2.4%) as reported — but excluding vested-but-unsettleable RSUs, Handler holds 8,330,535 (4.0%) and Friedman 4,847,185 (2.3%). 2,152,508 of Handler’s and 1,103,996 of Friedman’s shares are pledged in brokerage margin accounts. The ownership guideline is a non-binding 10x salary ($10m).

7.6 The M&A scorecard

Also of note: an $80.0m SEC/CFTC recordkeeping settlement in FY2022.

7.7 The total-return test

Book value per share went from $32.72 at FY2018 to a stated $51.26 at FY2025 — but $45.22 on a common-equivalent basis. Distributions over FY2019–FY2025 were roughly $7.30 of cash dividends plus ~$1.50 of Spectrum Brands and ~$2.20–2.40 of Vitesse in kind, approximately $11.00 per share. Common-equivalent accretion is therefore about $12.50 of book plus $11.00 of distributions on a $32.72 starting base: +71.8% over seven years, roughly an 8.0% compound annual rate (which flatters to 9.6% if one uses the SMBC-reduced share count). For a business running 7–9x tangible leverage on a Baa2 balance sheet, 8% is at or below the cost of equity — and most of it came from arbitraging a depressed multiple through the FY2018–23 buyback, not from operating returns.

The proxy’s own pay-versus-performance table shows $100 invested at 2020-11-30 growing to $307.63 for Jefferies against $209.79 for the S&P 500 Financials index — genuine outperformance. But it was $411.62 a year earlier: Jefferies fell 25.3% in FY2025 while the financials index rose 5.5%. The proxy’s own relative disclosure is blunter still: FY2025 one-year total shareholder return of −25.3% ranked 13th of 13 in Jefferies’ selected peer group — last place — which management attributed to “the Point Bonita news near year-end.”

One further signal that the capital-return story has stalled: the quarterly dividend has been held at $0.40 for five consecutive quarters through Q2 FY2026. The 2024 investor deck’s boast of “Dividend Increases in 6 of 7 Years” has quietly lapsed, and total capital returned fell from $2.61bn across FY2020–22 to $1.23bn across FY2023–25 — a 53% cut — while stock-based compensation doubled.

Verdict: intelligent when the job was liquidating Leucadia; self-serving now that the job is splitting a recurring profit pool. The FY2018–23 buyback at $23.91 was outstanding work and the legacy portfolio was unwound in sensible sequence. But since FY2024: no open-market buyback through two years and a 54% drawdown; the only stock retired at the high was to pay employees’ taxes; shares outstanding rose; a third of the headline share-count reduction is accounting; the dividend was raised 23% and $340m committed to an acquisition in the same period the firm declined to buy its own stock at 0.9x tangible book; and the incentive plan pays full freight for a sub-cost-of-capital return. Discipline that permits a dividend increase and a $340m acquisition but forbids buying stock near book value is a preference, not a constraint.

8. Changes and Headwinds — Last Two Years

8.1 First Brands: the sequence, and why the conduct matters more than the loss

The quantum is modest and should be stated first. Point Bonita Capital, a Leucadia Asset Management division, managed a roughly $3.0bn trade-finance portfolio on $1.9bn of invested equity, of which Jefferies’ own money was $113m (5.9%). At the Chapter 11 filing the portfolio held approximately $715m of purported First Brands receivables — about 24% of the portfolio, concentrated in a single issuer — nominally due from Walmart, AutoZone, NAPA, O’Reilly and Advance Auto Parts. Jefferies’ own economic exposure was roughly $43m, plus about $2m through Apex CLOs. Recognised losses: $30.0m pre-tax ($22.6m after tax, $0.09 per diluted share) in Q4 FY2025, and approximately $10m more in Q1 FY2026, fully writing off the direct exposure. Apex CLOs still hold $52m of First Brands term loans and $10m of debtor-in-possession paper against $4.5bn of CLO assets.

The conduct is the issue. First Brands stopped remitting on 2025-09-15. Jefferies’ Q3 FY2025 earnings release of 2025-09-29 contains zero mentions of First Brands or Point Bonita, and instead states that management is “increasingly optimistic about the near and long-term outlook.” The first disclosure came on *2025-10-08 — 23 days after the default — and opens with the words “In response to inquiries.” It was reactive, not volunteered.

Several disclosure changes deserve to be on the record:

• The bankruptcy date moved. The 2025-10-08 release and the Q3 10-Q state that First Brands filed “on September 29, 2025.” The FY2025 10-K and every filing since state “beginning on September 24, 2025.” The revised date places the first petitions before the earnings release that did not mention them.
• “Receivables” became “purported receivables.” The October wording was that the portfolio “has approximately $715 million invested in receivables.” From the FY2025 10-K onward it is “purported accounts receivable” — an unannounced concession that the assets may never have existed.
• The headline number was re-anchored. The 8 October disclosure led with $715m of client money; the 12 October letter reframed to Jefferies’ own “$43 million, or 5.9%.” Both are true and they measure different things. The fund-level loss to third-party investors has never been quantified in any filing.
• The FY2025 10-K contains no First Brands risk factor and no First Brands legal proceeding. The disclosure sits in “Other Developments” within the MD&A liquidity discussion, between the Russia/Ukraine and tariffs boilerplate. Internal control over financial reporting was reported as unchanged, with no material weakness.
• The Q2 FY2026 10-Q deleted the litigation. Both the Eugenia and Western Alliance suits were removed from Legal Proceedings on the stated basis that they “will have no material adverse effect” — a management judgement, not a resolution.

Management’s forecasts have not held. On 2025-10-12 Handler and Friedman wrote that the impact on equity value was “meaningfully overdone, and we expect this to correct soon”; the stock then fell to $35.74, 31% below the price at which the correction was predicted. Friedman said on 2025-10-16 that “our comp ratio will come down with scale slowly in time, there’s no question”; it went from 52.9% to 53.9%. Handler suggested multiplying Q3 by four; the actual H1 FY2026 run-rate came in roughly 15% short of that.

And it is not idiosyncratic. Gabelli’s Ian Lapey pressed the point directly at the October 2025 investor meeting, calling the 24% single-issuer concentration “a Risk Management 101 failure” and noting that Jefferies had used the word “idiosyncratic” for George Weiss and the Water Station / 3|5|2 Capital Ponzi matter as well: “I think it’s getting harder to say that today.” Friedman conceded “It troubles us the coincidence of several of these” and, asked whether Jefferies should be in asset management at all, answered: “Sadly, I would have said that three weeks ago, so I don’t have the conviction I had three weeks ago. We got work to do.” A fourth event has since emerged.

No specific remedy has been disclosed. There is no published concentration limit, no policy change, no personnel change and no organisational remedy. Friedman’s defence of the concentration was that the obligors were investment grade — which is precisely the protection that failed, since First Brands was the servicer collecting on those obligors’ behalf and the receivables were allegedly fabricated, inflated or sold multiple times. Notably, the 24% concentration figure came from an analyst; Jefferies has never published it.

Three further items belong on the record. First, the disclosure was forced. On or about 2025-10-06 the Financial Times reported that Jefferies had earned undisclosed fees on “side letter” financing provided to First Brands; the Wall Street Journal broke the $715m figure on the morning of 2025-10-08. Jefferies’ first disclosure was published that same day and opens with the words “In response to inquiries.” Second, the fund’s marketing is now an exhibit. Bloomberg reported on 2025-10-21 that Point Bonita’s April 2025 investor letter advertised “% of Positive Months: 100%” and annual gains between 7.56% and 9.38% — a zero-drawdown record marketed while the fund ran a 24% single-obligor concentration. Third, Jefferies presented an adjusted EPS that excluded the Point Bonita markdown in its Q4 FY2025 release ($0.96 adjusted versus $0.87 GAAP, and an “adjusted return on adjusted tangible shareholders’ equity of 12.9%”). Backing a loss on one’s own advised fund out of headline earnings is aggressive non-GAAP presentation.

The dispute has since escalated rather than settled. On or about 2026-07-01 Jefferies sued Western Alliance, alleging it unlawfully froze a $25m Point Bonita deposit account. Separately, First Brands’ indicted founder moved in SDNY on 2026-01-07 to compel Jefferies to comply with a subpoena for internal communications and due-diligence records on the receivables — which, if produced and unsealed, is the largest unquantified risk to the “we were defrauded, full stop” account. First Brands’ former chief executive and a former senior executive were indicted on 2026-01-29 on a nine-count indictment including continuing financial crimes enterprise, wire and bank fraud and money-laundering conspiracy (United States v. James, 1:26-cr-00029, S.D.N.Y.); a third executive pleaded guilty on 2026-01-26. First Brands declared $12m of cash against more than $9bn of liabilities.

One thing this was not: a short-seller attack. No named short seller published on Jefferies, and short interest was approximately 1.17% of float as at 2025-10-09 and falling. The drawdown was long liquidation by real holders, not a raid — which removes the most convenient explanation available to management and makes the loss of confidence harder to dismiss.

8.2 Market Financial Solutions: the same failure mode, six months later

In late February 2026 Market Financial Solutions, a UK bridging lender, entered administration amid allegations of double-pledged property collateral — roughly £1.16bn of loans against perhaps £230m of “true value.” Jefferies was a named lender alongside Barclays, Elliott, Santander and Wells Fargo, and first disclosed its £103m warehouse facility in the 9 March 2026 letter, guiding that the net earnings impact was “likely to be less than $20 million.” The Q2 FY2026 10-Q booked “a gross mark-to-market loss of $58.7 million associated with Market Financial Solutions” in Fixed Income.

Two frauds in six months, both turning on the same defect — collateral or receivables pledged or sold more than once — is a pattern, and it bears directly on whether the credit-management franchise can be trusted with third-party money.

8.3 The redemptions nobody narrated

Third-party net asset value in Jefferies-managed funds: $2,596m at 2024-11-30 → $2,462m at 2025-11-30 → $1,618m at 2026-02-28 → $1,261m at 2026-05-31. A 49% decline in two quarters. Headline “aggregated AUM” nonetheless rose from $30,842m to $32,490m, because affiliate-manager assets grew. The concealment is structural rather than deliberate — the aggregate metric mixes managed and affiliated assets — but the effect is that a halving of the firm’s own third-party fund capital is invisible in the headline. Asset management fees fell 21.1% in Q1 FY2026 and 27.0% in Q2 FY2026, which the 10-Q attributes “primarily [to] Point Bonita.”

8.4 Regulatory and legal overhang

The SEC was first reported by the Financial Times on 2025-11-27 to be investigating whether Point Bonita investors received adequate disclosure of the First Brands exposure, along with internal controls and conflicts between Jefferies units — the reported issue being that fund documents named the retailers as the exposure rather than First Brands. The DOJ was reported in July 2026 to be reviewing Jefferies’ conduct. Neither has been acknowledged in any Jefferies SEC filing; both are press-sourced only, and should be treated as such. First Brands’ former chief executive and former executive vice-president were indicted on 2026-01-29. Eight plaintiffs’ firms have announced securities-fraud investigations, though as of this writing no consolidated class action appears to have been filed.

8.5 The other material changes

A reported takeover approach — the single most important event missing from most accounts. On 2026-03-24 the Financial Times reported that Sumitomo Mitsui Financial Group was working on plans for a possible takeover of Jefferies. The tape corroborates it: the shares gapped open at $41.02 from a $39.25 close, touched $41.43 intraday and closed +2.53% at $40.24 on 6.33m shares — more than double surrounding volume — after being up as much as 14% pre-market. Bloomberg then reported the same day that SMFG had no immediate plan to take over Jefferies, and the move faded. This matters in two directions. It confirms that a strategic acquirer was examining the asset twelve days after the five-year low, which puts a partial floor under the equity; and it means part of what the market now pays for Jefferies is a bid option rather than the underlying return on equity. It also bears on the reading of SMFG’s stated willingness to act “if the falling share price presents an opportunity” — a sophisticated strategic buyer concluding the equity is cheap because returns are impaired is not a bullish signal about the operating business.

The SMBC relationship deepened decisively on 2025-09-19: a path to 20% economic ownership, roughly $2.5bn of new and incremental credit facilities, and a new non-voting share class. Nakashima, SMBC’s chief executive, joined the board on 2024-08-08 and Hyakutome was nominated on 2026-02-11. The Japan wholesale-equities joint venture begins January 2027.

Funding was termed out aggressively: $1.5bn of 5.500% notes due 2036 (2026-01-13), $1.1bn of 5.125% notes due 2031 (2026-04-23) and €850m of 4.500% notes due 2033 (2026-07-08), following €1.25bn in April 2024. Long-term debt rose from $9.1bn to $15.9bn, up 74%, while ratings stayed flat.

A structural governance gap. Jefferies holds no quarterly earnings call. Its releases contain no dial-in, webcast or replay, and the word “conference call” does not appear. The pre-merger Jefferies Group did hold them; the practice ended with the Leucadia combination. Management’s stated venues are the annual meeting and an October investor meeting — two forums a year. There is therefore no regular mechanism by which analysts can question management on any of the above, which is why the October 2025 investor meeting transcript is the only live adversarial exchange available.

Verdict: these developments weaken the thesis. The direct financial damage is absorbable and the balance sheet was never threatened. But the episode has cost Jefferies roughly half its third-party managed capital, produced a demonstrable pattern of counterparty-fraud losses, exposed a concentration-risk process that failed and has not been publicly remedied, drawn regulatory attention that the company has not disclosed, and revealed a disclosure posture that narrowed as the losses grew. For a firm whose principal asset is its reputation with clients and counterparties, that is a more serious cost than the $40m written off.

9. Risk Analysis

The shape of the risk is unusual and worth stating plainly: the risks that would normally dominate a broker-dealer analysis — trading VaR, Level 3 marks, liquidity — are demonstrably small here. The risks that matter are returns, compensation, counterparty selection in the credit businesses, and governance.

10. Valuation Discussion — Embedded Expectations

10.1 Where the shares trade

At $55.86 (2026-07-24) on 230,552,271 common-equivalent shares, the equity is worth approximately $12.9bn.

Own-history context matters more than cross-sectional comparison for a business this cyclical. Jefferies traded at 0.73x–1.00x tangible book from FY2019 through FY2023, peaked near 2.0x at the December-2024 high, and sits at roughly 1.47x today. The AZI valuation index places the current price-to-book in the 87.7th percentile of its own ten-year range, price-to-earnings in the 68.4th and price-to-sales in the 70.4th, for a composite of 75.5. Note that this percentile is computed on the voting share count and therefore, if anything, understates how richly the stock is priced on a true common-equivalent basis.

The essential observation: the stock is 28.7% below its high and simultaneously in the eighth decile of its own valuation range. Both are true because tangible book value per share has not compounded — as-converted, it has gone from $36.96 in FY2022 to $37.87 today. The 2023–24 re-rating was multiple expansion, and only part of it has come back.

10.2 What must be true — the embedded ROTE

For a financial, the cleanest statement of embedded expectations is the relationship between price-to-tangible-book, sustainable return on tangible equity, cost of equity and growth:

P/TBV = (ROTE − g) ÷ (COE − g)

Solving at the current 1.47x, with g = 3%:

At an 11% cost of equity, today’s price embeds a sustainable ROTE of roughly 14.8%. Jefferies’ actual record: 20.5% in FY2021, then 9.1%, 3.4%, 9.4%, 8.5%, and 10.6% on a trailing basis at a cyclical peak. It has printed something near 15% once in the last twelve years, in the SPAC boom — and an analyst noted at the October 2025 meeting that ROTE has exceeded 11% in only three of the last twelve years.

The gap is the thesis. The market is underwriting roughly double Jefferies’ demonstrated through-cycle return, and roughly 4–6 points above even management’s own adjusted figure.

10.3 Scenario analysis

Anchored on as-converted tangible book of $37.87 and a normalised mid-cycle earnings power. Mid-cycle net revenues are taken at roughly $6.8–7.0bn — between the FY2023 trough of $4.7bn and the FY2026 annualised run-rate of ~$8.4bn — with the compensation ratio the principal swing factor.

Fair P/TBV derived from (ROTE − g)/(COE − g) at COE 11% and g 3%. These are scenario outputs illustrating embedded expectations, not price targets, and no recommendation is expressed.

The distribution is the finding: today’s $55.86 sits at the top of the bull case. The market is not pricing a mid-cycle outcome with a possibility of upside; it is pricing the bull case as the base case. That asymmetry — modest upside if everything goes right, 30–50% downside if the cycle merely normalises — is what makes the current price unattractive rather than the business unattractive.

10.4 What the market is pricing correctly, and what it may not be

Priced correctly: the cyclical recovery is real and the H1 FY2026 numbers are excellent; the balance sheet is sound and the liquidity ample; the direct First Brands loss is small; the share gains are genuine and durable; the SMBC relationship adds real capability.

Possibly mispriced: that the compensation ratio will fall — it is rising; that book value compounds — as-converted it has not, for four years; that the First Brands episode is a closed, idiosyncratic incident — it is the fourth of a series, and third-party managed capital has halved; that reported book value per share is the right denominator — roughly 12% of the share count was concealed for three years and reappeared in June 2026; and that management’s 10.1% “adjusted ROTE” is comparable to Goldman’s 16.0% ROTE — it is not, being computed on a base deflated 27.7%.

10.5 The factor and positioning read

The FactorsToday model (Base + Sector + Industry, R² 0.634, 2026-07-24) gives Jefferies a Market beta of +1.44, DividendYield +1.00, Sector Financials +0.70, BetaFactor +0.66, CreditRisk +0.50 and Broker-Dealers +0.49, against Growth −0.28, Value −0.24, Momentum −0.18 and Quality approximately zero. Realised beta is 1.61 with negative alpha (−0.099).

Three inferences. First, the +0.50 CreditRisk loading is the quantitative signature of the leveraged-finance and private-credit exposure that First Brands made visible — this stock is a credit-spread instrument, and it will behave like one in a credit widening. Second, the near-zero Quality and negative Value loadings mean the stock does not trade as either a quality compounder or a cheap asset — the low headline price-to-book does not make it a value stock in factor terms. Third, the risk-adjusted record is poor: over the trailing year the return was +0.9% annualised against a −48.0% maximum drawdown; the three-, five- and ten-year annualised returns of 18.1%, 16.6% and 15.3% carry drawdowns of −54.4%, and the lifetime maximum drawdown is −80.7%. Factor-similar peers are Stifel (0.94 similarity), Evercore (0.93), Piper Sandler (0.91) and Morgan Stanley (0.90), which corroborates the comp set used above.

This is a high-beta, credit-sensitive, negative-alpha cyclical that has just rallied 56% off its low. It is not a falling knife today — the three-month move is roughly +18.5% — but neither is it an abandoned value name. The tape is consistent with a cyclical rebound into a peak, not with a re-rating of quality. Note also that short interest was only about 1.17% of float through the episode: the drawdown was long liquidation by real holders, not short pressure, so there is no crowded-short unwind to provide mechanical support.

10.6 The sell-side has already made this mistake once

The sell-side round trip is instructive. In December 2025 Morgan Stanley upgraded Jefferies to Overweight with a $78 target, explicitly citing “limited First Brands exposure” and a 2026 M&A rebound. On 2026-03-09 it downgraded to Equal-weight and cut the target 37% to $49, with Ryan Kenny writing that “the ongoing fallout from recent credit-related events at Jefferies may persist in the market for some time” and — directly relevant to this memo’s construction — that “the market is currently pricing the stock more against tangible book value per share than earnings given ongoing credit uncertainties,” because “balance sheet driven models typically trade against price-to-tangible book value versus return on tangible common equity when credit fears are elevated.” Oppenheimer’s Chris Kotowski had upgraded on 2025-10-17 calling the exposure “in reality very limited,” with the stock at $51.70; it traded at $36.02 five months later. On 2025-10-16 the median sell-side target was still $74 against a $49 tape.

The lesson is not that analysts are foolish. It is that the street underwrote management’s framing of the exposure, and the framing was accurate about the loss and wrong about the tail. That is precisely the error this memo is constructed to avoid: the $40m written off was never the point.

Note also what the current bull case actually models. Goldman Sachs raised its Jefferies target to $66 on 2026-06-05 with a Buy rating — and its published model carries return on tangible common equity of 10.1% in FY2026, 12.1% in FY2027 and 13.2% in FY2028. Even the constructive case therefore has Jefferies reaching, three years out, roughly half the return Goldman Sachs and Morgan Stanley are generating today — and only just arriving at the 13% level this memo treats as the threshold that would change the verdict. The bull case and the bear case do not disagree much about the returns; they disagree about what those returns are worth.

11. Variant Perception

11.1 Consensus

The consensus view is that Jefferies is a share-gaining, mid-cap investment bank levered to an M&A supercycle, trading at a discount to the bulge bracket on book value, with the First Brands episode a contained and idiosyncratic setback now behind it. On that view, ~1.2x book against Goldman at ~3x and Morgan Stanley at ~4x looks like an obvious relative bargain, and the H1 FY2026 revenue growth of 31% validates the operating story.

11.2 The strongest bull case, stated fairly

Fee share gains are real and documented — sixth globally in advisory at 4.5%, up 110 basis points in a single year, and seventh by total fees against twenty-third in equity fees in 2010. The fee pool is annualising at ~$160bn, matching the 2021 record, and industry front-office capacity remains below 2021 levels, which is the favourable supply configuration in which comp ratios can fall. The SMBC alliance supplies distribution, balance sheet and roughly $2.5bn of facilities that Jefferies could not build alone. The Jefferies Credit Partners model — converting arrangement fees into recurring management fees on proprietarily-sourced loans — is genuinely differentiated and is exactly the right structural response to private-credit disintermediation. Direct lenders are currently capital-constrained, handing share back to arranging desks. The balance sheet is sound, Level 3 is 1% of assets, and the FY2018–23 buyback at $23.91 demonstrates that this management can allocate capital well. If the compensation ratio falls to 48–50% on a sustained record fee pool, ROTE reaches the low-to-mid teens and the stock is worth meaningfully more.

11.3 The strongest bear case

Jefferies is a structurally disadvantaged competitor in a structurally bad industry, and the last two years have exposed the risk-management culture. It carries an advisory boutique’s compensation economics on a bank’s capital base and earns neither’s returns. It has not earned its cost of equity in any year since 2021, at any point in the cycle. As-converted tangible book per share has not compounded in four years. Employees take 81.6% of the pre-comp pool and the ratio is rising into a boom. Four counterparty-fraud events in three years, two sharing the same double-pledging failure mode, with no disclosed remedy. Third-party managed capital halved in two quarters. Undisclosed SEC and DOJ attention. No earnings call, six years of sub-90% say-on-pay, an incentive metric calibrated below the cost of capital, $246m of unplanned insider selling against $1.9m of buying, and no insider purchase through a 54% drawdown. Meanwhile the shares sit in the 88th percentile of their own price-to-book history, discounting a ~15% sustainable ROTE the firm has achieved once in twelve years.

11.4 The three to five assumptions that actually matter

1. Does the compensation ratio fall in an upcycle? This is the swing variable for the entire thesis. Every point of compensation ratio is roughly $73m of pre-tax profit at current revenue — about 0.85 points of ROTE. Moving from 53.8% to 48% would add roughly five points of ROTE and would substantially validate the bull case.
2. Is the current fee pool the cycle peak or a new plateau? At ~$160bn annualised the pool is at the 2021 record. If it holds for three years, normalised earnings are much higher than our base case; if it reverts, Jefferies’ 3x downside operating leverage does severe damage.
3. Is First Brands contained, or is the credit-fiduciary business structurally flawed? Four events and a halving of third-party capital argue the latter; management argues coincidence. This determines whether Jefferies Credit Partners — the only plausible moat — can scale.
4. What is the true share count and book value? Investors anchoring on reported book value per share of ~$52 and a ~194m voting share count are working with the wrong numbers. The correct figures are $45.83 and 230.6m.
5. Does SMBC eventually bid for the whole company? This is no longer hypothetical. On 2026-03-24 the Financial Times reported SMFG was working on plans for a possible takeover; the shares gapped up as much as 14% pre-market before Bloomberg reported no immediate plan and the move faded to a 2.5% close. A creeping 20% stake with board representation and an operating joint venture is the classic prelude, and this is the principal upside optionality not captured in the returns analysis. But the structure cuts the other way too: 20% of the economics with under 5% of the votes removes a fifth of the float from any activist arithmetic while giving SMBC no governance obligation — capital that cannot discipline management, accumulated without paying a control premium.

11.5 What would falsify each side

Falsifying the bear case: four consecutive quarters of ROTE above 13% on a GAAP-comparable basis with the compensation ratio sustained below 50%; as-converted tangible book per share compounding above 8% annually; third-party managed AUM stabilising and recovering; a published concentration-limit framework and disclosed remediation; and insiders buying stock with their own money.

Falsifying the bull case: a fifth counterparty-fraud or credit event; the compensation ratio remaining above 53% through a full year of record revenue; an adverse SEC or DOJ outcome or a securities class action surviving a motion to dismiss; third-party managed AUM falling below $1bn; or the fee pool rolling over with FY2027 net revenues falling below $6bn.

12. Fact vs. Interpretation

13. Open Questions

1. What was the fund-level loss to Point Bonita’s third-party investors? Jefferies has quantified only its own $30m + ~$10m. The $715m client-money figure has never been reconciled to a realised loss in any filing.
2. How much of the $1.2bn decline in third-party managed NAV is Point Bonita redemptions versus other funds? Never disclosed.
3. What concentration limits, policy changes or personnel changes have actually been implemented? Nine months on, no specific remedy has been published, and the FY2025 10-K reported no change in internal control over financial reporting.
4. What is the status and scope of the SEC and DOJ matters? Neither appears in any filing. Absent disclosure there is no reserve and no range of loss.
5. How does “Other investment banking” decompose between Jefferies Finance, Berkadia, Foursight and marks? A fall from $144.1m to $3.0m accounted for most of the FY2025 pre-tax decline and was not broken out. Jefferies Finance and Jefferies Credit Partners standalone financials and AUM are not in the 10-K.
6. What did Jefferies pay for Hildene? ~$340m cash plus ~$75m of contributed equity for ~50% of an $18bn-AUM manager, with no disclosed multiple, EBITDA or fee-rate detail — and it funds an insurance acquisition (SILAC) that moves the firm back toward diversification.
7. What is the succession plan? Handler and Friedman are 25 years in post with no disclosed successor, in a business the 10-K concedes is relationship-dependent.
8. Will SMBC bid for the whole company, and on what terms? The structure lets it reach 20% of the economics with under 5% of the votes.
9. What is the investment-banking managing-director headcount trend since FY2021? Not disclosed; required to complete the supply-side capital-cycle read.
10. Why is there no quarterly earnings call, and does the board intend to reinstate one?

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

The single most informative disclosure in the next twelve months is the FY2026 full-year compensation ratio, due in the January 2027 earnings release. It resolves bull assumption 1 and bear assumption 1 simultaneously, and it is the variable on which roughly five points of ROTE — and therefore most of the valuation gap — depends.

15. Source Appendix

The full source appendix — the 60-month SEC corpus, company communications on First Brands and Point Bonita, quantitative data sources with their caveats, third-party press, the analytical frameworks applied, and an explicit list of items that could not be sourced — is presented as Appendix B to this report.

One disclosure about the evidence base belongs in the body rather than the appendix: Jefferies holds no quarterly earnings call, so there is no quarterly transcript corpus for this issuer. The October 2025 investor meeting transcript, filed as an 8-K exhibit, is the only live adversarial exchange available for the period under review, and it is quoted extensively above.

Sections 1–15 contain no investment recommendation and no price target. The only view expressed in this document is the clearly-labelled opinion block at the front, which is the author’s own and is general information, not investment advice. The author may hold positions in securities mentioned. Always do your own research.

APPENDIX A — Standard Diligence Questionnaire

Report date: 2026-07-26 · Supplemental to the main analysis. Labels: [F] Fact · [I] Interpretation · [A] Assumption. Where a question does not map to a broker-dealer business model, the correct sector analogue is given.

General

What thoughtful questions have other investors asked about this company? The most penetrating questions on the public record come from the October 2025 investor meeting — the only live adversarial forum Jefferies holds, since it runs no quarterly earnings call [F].

• Brian Grefe (Raymond James) asked the central question: “since 2013, ROTE has only exceeded 11% in three of the last 12 years… can this business sustainably earn a double-digit return on equity?” Friedman’s answer — that the period was “aberrational” and “I won’t be able to maybe prove it for 10 more years” — is functionally unfalsifiable [I].
• James Yaro (Goldman Sachs) noted he had asked about the “low teens ROTE” target in five consecutive years [F].
• Ian Lapey (Gabelli) asked the risk question: 24% of Point Bonita’s assets in one issuer was “a Risk Management 101 failure,” and observed that Jefferies had already called George Weiss and the Water Station / 3|5|2 Ponzi matter “idiosyncratic” — “I think it’s getting harder to say that today.” [F]
• Perennial buy-side questions: when does the compensation ratio fall; is the Leucadia legacy book finally finished; and what is SMBC’s endgame.

Cyclicality and the nature of earnings

Are earnings at a cyclical high or low? A high [I]. The global IB fee pool annualises at roughly $160bn in H1 2026 — matching the 2021 record [F]. Jefferies’ H1 FY2026 net revenues of $4,223.6m annualise to ~$8.4bn against the FY2021 peak of $8,014m [F]. Any valuation capitalising current earnings is capitalising peak-cycle earnings.

Driven by the external environment or internal actions? Both, in roughly that order [I]. The fee-pool recovery is external. But the move from eighth to sixth in global advisory, and 4.5% fee share, is genuine internal execution [F].

How stable are revenues? Extremely unstable. Net revenues went $8,014m → $5,979m → $4,700m → $7,035m → $7,344m across FY2021–25 [F]. Peak-to-trough −41%, versus the industry pool’s −36%. Pre-tax fell 66.4% on a 21% revenue decline in FY2023 — roughly 3x downside operating leverage [F]. The variance sits almost entirely in underwriting: in FY2022 advisory fell 4.8% while underwriting fell 58.7% [F].

Outlook for products/services? Advisory and ECM are cyclically strong. Debt underwriting is structurally challenged: it fell year on year in H1 FY2026 despite record investment-grade issuance, because direct lenders now finance ~85% of LBOs by count [F/I]. Asset management is shrinking — third-party managed NAV −49% in two quarters [F].

How big will this market be — growing, shrinking, domestic or international? The fee pool is cyclical around a slowly-growing trend, currently at a record. Growth is not the constraint; the profit split is. Jefferies is genuinely global — the alliance with SMBC extends across the US, Europe, the Middle East, Canada, Asia and Australia [F] — with a Japan joint venture from January 2027 [F].

Business quality and competitive moat

Is the industry getting more or less competitive? More, and structurally so [I]. Independent advisory firms took US M&A fee share from under 15% (2018) to over 27% (2024); top-five share simultaneously rose from ~32% to ~35% [F]. The pool is barbelling and the squeezed casualty is the second-tier universal bank — where Jefferies sits.

How profitable is the business (ROIC, ROE)? Poorly, and this is the memo’s spine. ROE: 22.3% (FY21), 9.2%, 3.1%, 8.1%, 7.3% (FY25) [F]. ROTE: 20.5%, 9.1%, 3.4%, 9.4%, 8.5%, ~10.6% TTM [F]. Ex-FY2021 the five-year ROTE average is ~8.2% [F], against an estimated 11–12% cost of equity [A]. Jefferies has not earned its cost of equity in any year since FY2021 [I]. ROIC in the industrial sense is not meaningful for a levered financial; ROTE is the correct analogue.

How profitable is the industry — how many competitors, what barriers to entry? The industry earns roughly its cost of capital through the cycle [I]. Barriers are low at the individual-banker level (a boutique needs two managing directors and a Bloomberg terminal) and high only at scaled-platform level (jumbo underwriting, prime brokerage, clearing) [I]. Greenwald’s market-share-stability test fails decisively: share this mobile is definitional evidence of absent barriers [I].

Can the business be easily understood? The investment bank, yes. The asset-management segment, no [I] — $735.7m of FY2025 “net revenues” of which only $140.9m are actual fees, the rest being principal marks; “aggregated AUM” of $32.5bn that mixes directly-managed with affiliate assets and rose while directly-managed third-party capital halved [F]. That opacity is a real analytical cost.

Can it be undermined by foreign low-cost labour? No. The constraint is senior relationship talent in New York and London, which is the opposite of a low-cost-labour-exposed input [I].

Do brands matter? Yes, but as a door-opener rather than a barrier [I]. The Jefferies name gets a mandate considered; it does not get it awarded, and it does not create captivity. The FY2025 10-K concedes the point: if Jefferies lost certain professionals, “some of our clients could choose to use the services of a competitor” [F].

What is the nature of competition? Competition for bankers as much as for clients [I]. That is why the economics accrue to labour: Jefferies pays 81.6% of its pre-compensation profit pool to employees, versus Goldman’s 47.7% and Morgan Stanley’s 56.7% [F].

Customers’ switching costs? Effectively zero [F/I]. Mandates are per transaction; recurring revenue is 1.9% of net revenues [F]; deferred revenue is $62.6m against $7.3bn of annual net revenue [F].

Financial condition and balance sheet

Assets not fully recognised on the balance sheet? The 50% Jefferies Finance and 45% Berkadia joint-venture interests are equity-method and are plausibly carried below economic value — the best assets Jefferies owns [I]. Neither publishes standalone financials in the 10-K, which is a genuine gap [F].

Off-balance-sheet liabilities? No material off-balance-sheet leverage was identified. The relevant unrecognised exposures are contingent and legal: the Eugenia ($18.4m) and Western Alliance ($126m) suits, both removed from Legal Proceedings in the Q2 FY2026 10-Q [F], with no reserve or range of loss disclosed [F], plus undisclosed SEC and DOJ matters known only from press reporting [F]. Also note ~86% of deferred compensation is cash-settled — $621.5m of FY2025 amortisation against only $88.2m through APIC — a real funded future cash claim [F].

How conservative is the accounting? Mixed, tending unconservative in presentation rather than in measurement [I]. Conservative: Level 3 assets are only $737.8m, 1.0% of assets [F]; the First Brands position was written to zero promptly [F]. Unconservative: the FY2023 OpNet consolidation booked a +$115.8m “gain” by marking up a $201.6m carrying value [F]; the headline “adjusted ROTE” is computed on a base deflated 27.7% [F]; “aggregated AUM” obscures a halving of directly-managed third-party capital [F]; and First Brands appears nowhere in the FY2025 10-K’s risk factors or legal proceedings [F].

How CapEx-hungry is the business? Not capital-expenditure hungry — depreciation and amortisation was $201.9m in FY2025 [F] — but very balance-sheet hungry, which is the correct analogue. Total assets rose 48% in two years to $79.5bn and leverage went from 5.0x to 7.5x [F]. The sector analogue to “capital intensity” is regulatory and funding capital, and Jefferies consumes a great deal of it for the returns it produces [I].

Capital allocation and management

How much free cash flow does the business generate? Free cash flow is not a meaningful metric for a broker-dealer and should not be quoted. Operating cash flow is structurally negative (−$1,495m in FY2025) because a growing dealer funds inventory and receivables [F]. The correct analogue is earnings available for distribution after regulatory capital retention, approximated by net earnings to common of $630.8m in FY2025 [F].

How does management use it, and what is the philosophy? Historically well, recently poorly [I]. FY2018–FY2023: roughly 157.7m shares retired at an average $23.91, about 0.6–0.7x book — outstanding [F]. FY2024–FY2025: zero open-market repurchase despite a live $250m authorisation; the only stock retired at the all-time high was 0.7m shares at $79.57 to fund employees’ tax withholding [F]; shares outstanding rose in FY2025 [F]. Around the low: bought at $62 and $57, zero in March 2026 when the stock bottomed at $35.74, then resumed at $48–52 [F]. Jefferies operated no Rule 10b5-1 repurchase plan that would have permitted buying through the blackout [F].

Significant acquisitions recently? Yes — a strategic change. On 2025-12-09 Jefferies agreed to acquire ~50% of Hildene ($18bn AUM credit manager) for ~$340m cash plus ~$75m of contributed equity, with up to $100m further in convertible preferred, partly funding Hildene’s acquisition of SILAC, an insurance business; closing targeted Q3 2026 [F]. No multiple was disclosed [F]. This is the first large cash acquisition of the post-Leucadia era and moves the firm back toward diversification — into insurance and annuity reinsurance — immediately after a credit blow-up [I].

Buying back shares? Only intermittently, and not at the lows — see above. H1 FY2026: $371.7m total, of which $264.9m genuine open-market at a $53.42 blended price [F]. Note also that roughly a third of the headline FY2019–25 share-count reduction never happened: 27,562,500 shares were exchanged by SMBC into non-voting preferred and converted straight back in June 2026, so the real reduction is 21.0%, not 29.7% [F].

Issuing large amounts of new shares to insiders? Yes, and rising. Stock-based compensation went $45.4m → $63.1m → $88.2m, up 94% in two years, with roughly +9.79m shares issued to employees over FY2023–25 [F].

Compensation policy of directors and management? The central governance finding. Handler and Friedman received $162.98m and $160.08m respectively across FY2021–25 (summary compensation table); “compensation actually paid” to Handler was $282.3m over five years [F]. Say-on-pay support: 55.3%, 53.2%, 58.5%, 71.8%, 89.0%, 87.9% [F] — five straight years under 90%. The only formulaic metric is a return on adjusted tangible equity computed on a base deflated 27.7% (stripping goodwill, the deferred tax asset, and “the weighted average impact of cash dividends and share repurchases”), so the 10% target that pays 100% of the award corresponds to roughly a 6.5–7% GAAP ROE — below the cost of capital [F/I]. FY2023 adjusted ROTE of 3.9% was below the plan’s own threshold, yet the FY2023–25 cycle paid ~81% of target [F].

Motivations of management? Aligned with growth in firm size and with employee outcomes, not demonstrably with per-share value [I]. Supporting evidence: the incentive metric is set below the cost of capital; the compensation ratio rose in a 31%-growth half-year; and the insider record is $246.4m of sales against $1.86m of purchases across five years, with the aff10b5One flag false on all 25 sales — none under a Rule 10b5-1 plan [F]. Roughly $89m was sold by Handler, Friedman and a director in the six weeks before the December-2024 all-time high [F]. Neither Handler nor Friedman has bought a single share on the open market in five years [F], including through a 54% drawdown after Handler publicly called the stock “kind of cheap” at $47.72 [F]. Additionally, 2,152,508 of Handler’s and 1,103,996 of Friedman’s shares are pledged in brokerage margin accounts [F], and excluding vested-but-unsettleable RSUs their true stakes are 4.0% and 2.3% rather than the headline 7.8% and 2.4% [F].

Valuation and market data

Is the stock an ADR, MLP or K-1 issuer? No. Jefferies Financial Group Inc. is a Delaware C-corporation listed on the NYSE issuing a Form 1099, not a K-1 [F]. Note the two-class structure: 193,742,690 voting and 36,809,581 non-voting common shares at 2026-06-30 [F] — the non-voting class arose from the June-2026 conversion of SMBC’s Series B preferred, and screens reading only the voting line understate the share count by roughly 19% [F/I].

Dividend policy? A quarterly common dividend of $0.40 per share declared January 2026, up 23% over the period from $1.30 to $1.60 annualised [F]. FY2025 common dividends totalled $361.7m; the payout ratio was ~51% of net earnings to common [F]. Note the dividend was raised in the same period the firm declined to repurchase stock near book value [I].

How profitable is the business? See above: FY2025 pre-tax margin 11.9%, net margin on net revenues 8.6%, ROTE 8.5% [F]. Peer pre-tax margins: Morgan Stanley ~31%, Evercore 20.5%, Houlihan Lokey 20.1%, Raymond James 19.6% [F]. Jefferies has the worst margin and the worst return on equity in its peer set [F].

Is net income diverging from cash from operations? Yes, structurally and benignly: net earnings of $710.5m against operating cash flow of −$1,495m in FY2025 [F]. For a growing dealer this reflects inventory and receivable funding, not an earnings-quality failure [I]. The genuine earnings-quality issues lie elsewhere — the tax-rate benefit that masked a 19% underlying FY2025 EPS decline, and the undecomposed collapse of joint-venture earnings from $144.1m to $3.0m [F/I].

Risks and downside

What factors would cause the stock to decline? A fee-pool reversion (3x downside operating leverage); the compensation ratio failing to fall; a fifth counterparty-fraud or credit event; an adverse SEC or DOJ outcome; further erosion of third-party managed AUM; multiple compression from today’s 87.7th-percentile price-to-book; and a credit-spread widening, to which the stock has a +0.50 CreditRisk factor loading [F/I].

Risk of a catastrophic loss? Low, and this is why the memo does not argue a short [I]. Level 3 assets are 1.0% of total assets; average daily VaR is $11.23m, 0.13% of tangible equity; the liquidity buffer is $17.7bn, 23.9% of assets ex-goodwill; long-term debt has a 7.4-year weighted-average maturity; and all three rating agencies affirmed stable outlooks after First Brands [F]. The genuine tail risk is the funding model — no deposits, no discount-window access, and the 10-K models a two-notch-downgrade collateral call [F] — which would matter only in a severe wholesale-funding freeze.

Chance of a total loss? Very low [I]. Jefferies is a Baa2/BBB/BBB+ rated, $10.6bn-equity institution with a large liquidity buffer and a modest trading book, and holds ~$8.7bn of tangible common equity. The realistic bear case is a de-rating toward 0.6–0.7x tangible book on a cyclical reversion, not insolvency.

Recent news and events

Has the business environment changed recently? Materially, in three ways [F/I]. (1) The fee pool has recovered to a record while deal count sits at a six-year low — a megadeal barbell. (2) The March 2026 Basel reproposal delivers ~$87.7bn of net capital relief to bank competitors, eroding the value of Jefferies’ non-bank status. (3) Private credit is in visible stress — a record 6.0% Fitch default rate in April 2026, widespread redemption gating — which cuts both ways for an integrated arranger-and-lender.

Significant acquisitions? Hildene (see above). Also continuing divestment: Tessellis under binding offer with a $58.2m goodwill impairment taken in Q1 FY2026, closing expected Q1 2027 [F].

Change in accounting policies? No material change. ASU 2023-07 (segment disclosures) was adopted for FY2025, affecting disclosure only [F]. The FY2025 10-K reports no material weakness and no change in internal control over financial reporting — notable given the year’s events [F/I]. Note two non-accounting disclosure changes: the First Brands bankruptcy date was revised from “September 29, 2025” to “beginning on September 24, 2025,” placing the petitions before the Q3 earnings release that did not mention them; and “receivables” became “purported receivables” from the FY2025 10-K onward [F].

Recent changes — new markets, facilities, management? The SMBC relationship deepened decisively (2025-09-19): a contractual path to 20% economic ownership with under 5% of the votes, ~$2.5bn of new and incremental credit facilities, a new non-voting share class, SMFG’s chief executive on the board since 2024-08-08 and a second nominee in February 2026, and a Japan wholesale-equities joint venture from January 2027 [F]. Funding was termed out aggressively: $1.5bn of 5.500% 2036 notes (January 2026), $1.1bn of 5.125% 2031 notes (April 2026) and €850m of 4.500% 2033 notes (July 2026), taking long-term debt from $9.1bn to $15.9bn, up 74%, while ratings stayed flat [F]. No change in senior management: Handler and Friedman remain in post after 25 years, with no disclosed succession plan [F].

APPENDIX B — Source Appendix

Report date: 2026-07-26 · All sources accessed 2026-07-26 unless otherwise stated. Sources are ordered by evidentiary weight: primary regulatory filings first, then company communications, then third-party data, then press. Every non-obvious claim in this article traces to an entry below.

A. Primary SEC filings (the 60-month corpus)

The trailing five-year EDGAR corpus was downloaded and read in full: 5 Form 10-K, 16 Form 10-Q, 82 Form 8-K, 6 DEF 14A, 10 DEFA14A, 2 PRE 14A, 4 ARS, 2 S-3ASR, 4 11-K, plus the full Form 4 corpus (271 filings).

EDGAR entry point: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000096223&type=10-K

B. Company communications on First Brands / Point Bonita

Structural note: Jefferies holds no quarterly earnings call. Its earnings releases contain no dial-in, webcast or replay details and zero instances of “conference call.” The pre-merger Jefferies Group did hold them (e.g. Q3-2011). Management’s stated venues are the annual meeting and the October investor meeting. There is consequently no quarterly transcript corpus for this issuer; the ROIC.ai transcript tools returned nothing for every quarter probed, which reflects the absence of the calls rather than a data gap.

C. Quantitative data sources

E. Third-party and press sources

Used only where primary sources are silent, and flagged as such in the memo.

F. Analytical frameworks

• Greenwald & Kahn, Competition Demystified (barriers to entry; the three genuine advantage types; the market-share-stability and ROIC tests) applied in Sections 3.7 and 4.2–4.4; Marathon Asset Management / Chancellor, Capital Returns (supply-side capital-cycle analysis; the asset-growth anomaly) applied in Sections 3.6 and 6.5.

G. Items that could not be sourced — stated as gaps, not inferred

1. Fund-level loss to Point Bonita’s third-party investors. Never quantified in any filing.
2. Dollar value of Point Bonita redemption requests. Never disclosed; the $1.2bn decline in third-party managed NAV is observable but is not attributed by the company.
3. Any Jefferies exposure to Tricolor Holdings. Zero mentions across the FY2025 10-K and both FY2026 10-Qs. Third-party press names Jefferies among lenders; Jefferies has never quantified or acknowledged it, and no exposure is asserted in this memo.
4. ISS and Glass Lewis say-on-pay recommendations. Not sourceable; not asserted.
5. Decomposition of “Other investment banking” between Jefferies Finance, Berkadia, Foursight and marks; and Jefferies Finance / Jefferies Credit Partners standalone financials and AUM.
6. Investment-banking managing-director headcount by year.
7. Hildene transaction multiple, target EBITDA or fee rates.
8. Any reserve or range of loss for the First Brands litigation, or any filing acknowledgement of the SEC or DOJ matters.
9. Verbatim Matt Levine / Money Stuff commentary on Jefferies–Point Bonita. Both relevant columns (2025-10-09, 2026-01-29) are paywalled; no quotation is asserted.
10. Jefferies bond spreads or 5-year CDS levels through the episode. Not sourceable. The only evidence is management’s own concession of an impact on “credit perception,” plus the fact that market access was never impaired — €850m of 4.500% 2033 notes priced at a 4.544% yield on 2026-07-08.
11. Published KBW, Wolfe Research, Autonomous or Bank of America notes on JEF/First Brands. Not located; no views attributed to them.
12. Point Bonita’s formal legal status (dissolved versus in run-off). Never stated by Jefferies. The defensible formulation, used in the memo, is that the fund has been in staggered wind-down since 2025-12-31 with the final scheduled redemption payment due October 2026, and that no closure has been announced.
13. Any DOJ investigation of Jefferies itself (as distinct from of First Brands) is sourced only to secondary aggregators and is treated as unverified in the memo. The DOJ inquiry into First Brands is well sourced (Reuters, 2025-10-09).

One caution recorded for future engagements: a widely-surfacing item headlined “Jefferies CEO Issues Letter Saying Short-Sellers Spreading Malicious Lies About Company” dates from the 2011 MF Global episode, not this one, and a Money Stuff column titled “Jefferies Funded Some Fake Water” dates from 2024-07-11 and concerns the Water Station / 3|5|2 Capital matter. Neither relates to First Brands, and neither is used as evidence of it here.