FirstService Corporation (NASDAQ / TSX: FSV) — Buying Its Own De-Rating on Borrowed Money While the Weather Waits

Prepared by: Independent fundamental research Date: August 15, 2026 (UPDATE — supersedes the June 7, 2026 report) Subject company: FirstService Corporation — Canadian foreign private issuer (FPI), reports in US dollars under US GAAP, files Form 40-F (annual) + 6-K current reports (no 10-K/10-Q, no Forms 3/4/5 — insiders remain Section 16-exempt; insider data is Canadian SEDI-only), fiscal year ends December 31 Price reference: ~US$140.81 (August 14, 2026 close) · 52-week range $119.15–$208.07 (dividend-adjusted; the prior report’s $209.66 was the unadjusted print of the same September 11, 2025 event) · 44,154,011 shares (June 30, 2026, post-buyback) · market cap ~$6.22B · enterprise value ~$7.3B simple (net debt only) / ~$8.1B lease- and RNCI-inclusive CIK: 0001637810

⚡ Claude’s Take

This block is Claude’s own independent, subjective opinion, deliberately placed outside the body of this memo. It is general information, not investment advice. The body of this memo (Sections 1–15) deliberately carries no recommendation and no price target; only this fenced block takes a position.

Verdict: HOLD / accumulate-on-weakness — the June call stands, re-based. The bear case (~$120–145) still sits roughly on top of today’s ~$141 price; base-case value re-derives to ~$195–220; the bull case (~$280–315) is now explicitly a 2027 story. Accumulate on weakness below ~$135 — a zone where the most aggressive bidder this quarter was the company itself, on borrowed money.

Tag: “The contrarian bidder is the company itself — and it brought leverage.”

The call did not change versus the June 7, 2026 report, but its composition did. Ten weeks, one earnings print, a maxed-out NCIB and $248M of repurchases later, the stock is flat at ~$141 — and the asymmetry statement survives intact, with a debt-financed floor now under it. The evidence split cleanly since June. The Residential half confirmed the bull case again (+5% organic in Q2, H1 segment margin +30bp, captivity actively deepened via the Resilience First cross-sell launch). The restoration snapback was deferred a second year — 2026 is tracking quieter than 2025 (H1 insured catastrophe losses of $42–47B, the lowest H1 since 2020; NOAA’s August 6 update went to 75% probability of a below-normal season) — but the character of the recovery improved: it is arriving share-led through non-weather large-loss wins converting over 12–18 months, which is more durable than a storm bounce, just later. Roofing got worse (−10% organic, backlog down year-over-year, the Q4’26 goodwill test still live on a <5% cushion). And management answered the June memo’s central criticism — an unused buyback at a decade-low valuation — by spending $248.4M at an average of $135.91 in a single quarter, debt-funded, deliberately moving leverage from ~1.5× to ~1.9× with stated comfort to ~2.5×.

Why this stays a HOLD rather than a pound-the-table buy: the buyback manufactures essentially no EPS (at ~24× Adjusted EPS the ~4.2% earnings yield ≈ the ~4.0% after-tax cost of debt — the zero-accretion zone), so it is an intrinsic-arbitrage bet that pays only if management’s claim — that the enterprise trades at a “meaningful discount to smaller private market businesses” — is right. On conservative owner earnings (~$4.71 TTM) the stock is still ~30×, fair-to-full on trough earnings, cheap only against its own history (~4th percentile) and only if Brands normalizes. The CEO selling ~$10M of stock at ~$213–215 in February, four months before the company bought at ~$136, is a factual tension worth staring at. The framing is mild-contrarian / abandoned de-rated compounder, post-falling-knife, pre-recovery-confirmation: the factor evidence (Momentum loading −0.12 on ~zero Quality, 65–73% of variance idiosyncratic, growth funds exiting, no crowded contrarian bid) says this is not a momentum trade and not a factor-rotation victim — only company-level evidence re-prices it. Conviction: medium. Bull flip: two consecutive quarters of Brands/restoration organic re-acceleration — share-led counts, no storm required. Bear flip: a roofing impairment at the Q4’26 annual test plus Residential organic falling below mid-single digits.

Changes since 2026-06-07

This update summarizes what moved the thesis, what confirmed, and what falsified against the June 7, 2026 baseline.

What moved the thesis (new facts):

1. The buyback regime change. FSV repurchased 1,827,750 shares for $248.4M at an average of $135.91 in Q2 2026 (zero in Q1, zero in all of 2025), all cancelled — ~4% of the share base retired in one quarter, debt-funded (net debt $928.3M → $1,077.0M; net leverage ~1.5× → ~1.8–1.9× TTM Adjusted EBITDA, top of the historical range). The NCIB was amended June 2, 2026 to the TSX maximum — 4,118,199 shares, 10% of public float — with an automatic share purchase plan; 2,290,449 shares of capacity remained at June 30; the bid expires August 25, 2026 and no renewal was announced as of this report date. The June baseline’s “fortress balance sheet” margin of safety has been deliberately converted into equity-price convexity.
2. The restoration snapback slipped to 2027. Q2 restoration revenue was down slightly; the H2 guide is only ~+5%, explicitly built on non-weather large-loss wins (factories, warehouses, government, big-box retail, multifamily) converting over 12–18 months — revenue skewed to Q4’26/2027. NOAA’s August 6 update maintained a below-normal season; H1’26 insured catastrophe losses were the lowest since 2020. June’s “no snapback priced” is now more true: the market is underwriting two trough years, not one.
3. Roofing deteriorated past its own guidance. Q2 roofing organic was −10% (Q1 guide had implied ~flat Q2 with sequential H2 improvement), with backlog down year-over-year and Q3 guided to a mid-single-digit organic decline. The drag inside Brands has migrated from weather-led to roofing-led.
4. Orbis emerged as an 8.8% holder. 13G/A filed August 14, 2026: 4,044,289 shares (8.8%) at June 30, up from 8.6% at March 31 — a value-oriented manager accumulating through the drawdown, the only institutional echo of the company’s own bid.
5. New quality-of-earnings lever. An uncommitted $300M receivables-sale facility (CIBC, April 2026) sold $35.3M of AR in Q2, derecognized — flattering both H1 operating cash flow (headline +7% becomes ~−10% underlying ex the facility) and reported leverage optics. Small so far; a watch item.

What confirmed:

• Residential. The bear-side test (“organic below mid-single digits”) did not fire; the bull side confirmed for a second consecutive quarter (+5% organic Q2 after +4% Q1; H1 +4% all-organic; segment EBITDA margin +30bp to 9.9%). Resilience First (launched July 6) actively deepens captivity.
• Owner-earnings anchor. TTM owner EPS ≈ $4.71 — inside June’s $4.50–$5.00 range; the QoE thesis is unchanged (recurring add-backs — RNCI increment + SBC ~$0.54/sh per half, intangible amortization annualizing >$1.30 — are drifting up, not down).
• Valuation percentile. AZI own-history composite percentile 3.6 (verified fresh, 2026-08-14) — the same bottom ~3–4% zone as June.
• M&A discipline. H1 spend $48.4M across 5 tuck-unders, on pace for the ~$100M/yr floor; earn-out reversals of −$17.2M confirm honest booking (and RCA underperformance).

What falsified (or half-fired):

• The prior bull flip (two consecutive quarters of restoration organic re-acceleration) was not met — but the test has been restated (see scorecard) to count share-led recovery, because management disclosed a non-storm recovery path the original wording would have misread as failure.
• The roofing bear leg half-fired: the decline continued (worse than guided), but no impairment — the Q4’26 annual test is the decision point.
• The combined bear trigger (roofing impairment + Residential below mid-single organic) did not fire.

Falsification scorecard vs the June report’s valuation and risk-section tests:

Corrections to the June report (carried through this memo): (i) the claim that FSV insiders “became subject to Section 16 in March 2026” is unsupported — the EDGAR full-history record contains zero Forms 3/4/5 for CIK 0001637810; FSV’s FPI exemption is intact and insider-trade data is Canadian SEDI-only (a transparency gap that is arguably larger than the baseline stated); (ii) the 52-week high is $208.07 on the dividend-adjusted series (the prior $209.66 was the unadjusted print of the same event); (iii) share count is 44,154,011 at June 30, 2026.

📈 Stock Price Action — Five-Year Event Map

The arc: FSV compounded from ~$183 (August 2021) through the 2022 rate shock ($109.53 low, September 2022), re-rated to an all-time-area high of $208.07 (September 11, 2025, adjusted) on the 2024 hurricane cycle and acquisition-led growth, then de-rated ~43% peak-to-trough into the 2025–26 weather trough and roofing-goodwill scare ($119.15, May 18, 2026). It now sits at $140.81 (August 14, 2026) — 32% below the high, 18% above the low, and flat (+0.1%) versus the June 7 report. 52-week range: $119.15–$208.07. Price moves below are FACT; attributed drivers are INTERPRETATION unless a specific dated print or event is cited.

Cycle narrative. (1–2) The 2022 drawdown was macro, not company — long-duration quality de-rated on rates; the recovery tracked resumed earnings growth. (3) The RCA platform deal (December 2023) was the market’s last act of trust in the roll-up algorithm. (4) The 2024 hurricane year (Helene/Milton) showed what restoration operating leverage does to the P&L. (5–6) 2025 whipsawed on tariffs and weather evidence before the Q2’25 print carried the stock to its high. (7) The decisive repricing was three prints long: Q3’25 (−9.8% on organic/guidance concern despite an EPS beat), Q4’25 (a beat and an 11% dividend raise could not overcome the <5% roofing goodwill cushion disclosure), and Q1’26 (Brands margin pressure). (8) The May–August round trip is the current standoff: a −7.4% earnings-day flush on July 23, 2026 was fully recovered in three sessions — the market sold roofing and consumer weakness, then bought the maxed-out NCIB and the restoration inflection. FSV’s drawdowns are idiosyncratic (specific vol ~23%/yr; factor-model R² only 0.27–0.37; beta 0.66 yet a −44% five-year max drawdown) — this stock reprices on company evidence, not factor rotation. This block is price history, not a call.

1. Executive Summary

FirstService Corporation is a North American provider of essential, recurring, outsourced property services in two segments. FirstService Residential (FY2025 revenue $2.29B, 42% of total) is the largest manager of residential communities in North America — over 9,500 communities and 4.7M+ residents under multi-year contracts with mid-90s% retention. FirstService Brands (FY2025 $3.21B, 58%) is a federation of company-owned and franchised property-services businesses: First Onsite (the #2 commercial/large-loss restoration provider), Century Fire Protection, Roofing Corp of America (the December 2023 platform), and capital-light franchise brands (California Closets, CertaPro, Paul Davis, Floor Coverings International, Pillar to Post). Spun out of the Colliers predecessor in June 2015, FSV eliminated its dual-class structure in May 2019; today it is a one-vote-per-share company with no ≥10% holder, founder Jay Hennick as non-executive Chairman (~6%), and CEO Scott Patterson and CFO Jeremy Rakusin running the business.

The June report framed the investment question as cyclical-versus-structural: a quality Residential compounder hidden behind a Brands segment at a weather-and-construction trough. The Q2 2026 print (July 23) sharpened rather than resolved that question. H1 2026: revenue $2,766.3M (+3.7%), Adjusted EBITDA $267.4M (+2.7%, 9.7% margin), Adjusted EPS $2.69 (+2%), GAAP diluted EPS $1.43. Residential delivered the bull case — +5% organic in Q2 with margin expansion — while Brands went backwards organically (−3% in Q2, −1% H1), with the driver now explicitly roofing (−10% organic) rather than restoration weather. Management’s full-year framing — revenue growth “similar to or modestly better” than H1’s +4%, mid-single-digit EBITDA growth — implies FY2026 lands at roughly half the company’s own 10% long-term algorithm, and pushes the restoration normalization story into 2027.

The defining event of the period is capital allocation, not operations: a $248.4M buyback of 1.83M shares at $135.91, executed in one quarter, debt-funded, after the NCIB was upsized to the 10%-of-float TSX maximum. Leverage moved from ~1.5× to ~1.8–1.9× with stated comfort to ~2.5×; the chronic ~1%/yr SBC dilution reversed (share count −3.4% net in H1); and management articulated the rationale in the same public-private arbitrage language that governs its RNCI partnership model. The EPS math is essentially neutral (earnings yield ≈ after-tax debt cost), so the buyback is a statement of intrinsic value, not financial engineering for reported EPS — and it arrived in the same year the CEO sold ~$10M of stock at ~$214 and shareholders approved +2.0M of new option-plan headroom.

Valuation is unchanged in structure and slightly changed in level. At $140.81 the market cap is ~$6.22B and EV is ~$7.3B simple / ~$8.1B lease- and RNCI-inclusive: ~24.2× TTM Adjusted EPS of $5.81, ~30× owner earnings of ~$4.71, 12.8–14.2× TTM Adjusted EBITDA, ~5.4% headline FCF yield (~4.8% ex the new AR-sale benefit), and a ~3.6th-percentile reading against its own ten-year history. Embedded expectations work out to roughly high-single-digit Adjusted-EPS growth with no weather snapback through 2026 and continued roofing decline — a hurdle the buyback mechanically lowers by ~2–4pp/yr. The bear scenario (~$120–145) sits on today’s price; the base (~$195–220) and bull (~$280–315) scenarios require normalization that is now dated 2027. The body that follows argues each verdict from the evidence; it takes no position and sets no price target (the only view is the fenced Claude’s Take above).

2. Business Overview

2.1 What FirstService does

FirstService is an outsourced, branded, essential-property-services company. Its unifying logic is that property — residential communities and commercial/residential buildings — requires a continuous stream of management and maintenance services that owners increasingly outsource to professional, scaled, brand-name providers. The company captures that demand through two go-to-market models, organized as two reportable segments. The model description below carries forward from the June baseline (the structure has not changed); the numbers are updated to Q2 2026.

2.2 Segment 1 — FirstService Residential (FY2025 revenue $2,287M; 42% of total)

FirstService Residential is the largest manager of residential communities in North America, serving the boards of homeowner associations, condominium and co-op corporations, master-planned communities, active-adult communities, and high-rise buildings — over 9,500 communities representing more than 4.7 million residents across 25 U.S. states and 3 Canadian provinces, with particular density in Florida, Texas, California, and the Northeast.

The core service is professional property management under contract: financial administration, board governance support, vendor management, maintenance coordination, compliance with the dense and rising body of state condominium/HOA law, and on-site staffing (“sited labour” billed through to associations, often near cost). Revenue is contracted and recurring, with a growing higher-margin ancillary layer — FirstService Financial (banking and insurance brokerage), FirstService Energy, and amenity management. Two 2026 developments update the picture. First, at the start of Q2 the segment divested its non-core single-family residential aquatic operations (keeping commercial pools), which depresses reported growth relative to organic for roughly the next three quarters (Q2: +4% reported vs +5% organic). Second, on July 6, 2026 the company launched “Resilience First” — a cross-sell program bundling FirstService Residential with the restoration and roofing brands (loss prevention, proprietary leak detection, insurance-cost reduction for managed communities), explicitly framed by management as differentiating the Residential franchise. That is the ancillary-attach flywheel extended into the Brands portfolio — a captivity-deepening move worth more attention than it has received.

This remains the higher-quality half: non-cyclical demand, multi-year contracts with mid-90s% retention, and a secular tailwind — the Community Associations Institute projects U.S. community associations growing from ~373,000 at end-2025 to as many as ~377,000 in 2026, roughly one-third of U.S. housing stock. H1 2026 evidence: revenue $1,162.5M (+4%, all organic), Adjusted EBITDA $115.3M (+7.6%), margin 9.9% vs 9.6% — “continued operating efficiencies in our service delivery model” (MD&A).

2.3 Segment 2 — FirstService Brands (FY2025 revenue $3,211M; 58% of total)

FirstService Brands is a portfolio of essential property-services businesses, delivered through company-owned operations (larger, more capital- and labour-intensive) and franchise systems (capital-light royalty models):

• First Onsite Property Restoration — commercial and large-loss restoration; the #2 player in North America behind BELFOR, and the segment’s most weather-sensitive line. Q2 2026 revenue was “down slightly” year-over-year on a weak pipeline inherited from mild Q4 2025 weather; late Q2/July brought “a number of large-loss projects” (factories, warehouses, government buildings, big-box retail, multifamily — regional weather/fire/water events, not hurricane-driven) that rebuilt the pipeline to “historically healthy levels,” converting over the next 12–18 months. Management is also pulling a new adjacency — specialty construction/retrofits out of its healthcare-sector restoration expertise — with early wins.
• Century Fire Protection — code-mandated fire-sprinkler design, installation, inspection, and service; the segment’s organic-growth standout. Q2 revenue +10%+ year-over-year including high-single-digit organic; backlog up sequentially and “well up over prior year”; data centers only ~15% of backlog. Acquired Titan Fire Protection (Tampa) and GSC Fire and Security (Austin/San Antonio) in Q2; guided to 10%+ growth in Q3/Q4.
• Roofing Corp of America (RCA) — the commercial-roofing platform acquired December 2023, now the segment’s problem child: Q2 revenue −6% reported, −10% organic (“lower than our expectation”), backlog down year-over-year (though up sequentially May→June), and a Q3 guide of a mid-single-digit organic decline. RCA closed its 16th acquisition (Scheffer’s Roofing, Kansas City; 27 branches) in June 2026 and is mid-way through consolidating 14 operating/financial systems into one.
• Franchise / home-services brands — California Closets, CertaPro Painters, Paul Davis Restoration (372 franchises), Floor Coverings International, Pillar to Post. Q2 revenue “up slightly,” with growth driven by close-ratio and average-job-size gains — share capture, not market growth — against a housing/consumer backdrop management describes as “10-year lows.”

The Brands model is a roll-up: acquire regional operators, plug them into a national brand, shared back office, procurement scale, and a partial-ownership incentive structure for operating managers (Section 7). H1 2026: Brands revenue $1,603.8M (+4% total, −1% organic); Adjusted EBITDA $160.0M (−1.9%); margin 10.0% vs 10.5% — “competitive pressures in the roofing industry, as well as margin compression in home services due to promotional activities” (MD&A). Note the H1’25 comparator was flattered by a $14.6M negative earn-out adjustment inside acquisition-related items (excluded from Adjusted EBITDA), so the −50bp Adjusted margin decline is clean.

2.4 How it makes money & revenue quality

TTM (through June 30, 2026) revenue is $5,597M, splitting ~42% Residential / ~58% Brands. Revenue quality remains a tale of two segments: Residential recurring, contracted, counter-cyclical-leaning; Brands a blend of recurring (fire inspection, franchise royalties) and event/cycle-driven revenue (restoration storms, roofing construction, discretionary home services). Gross margin ~33%; capex runs ~2.2% of revenue (service fleet and IT). The company’s preferred metrics remain Adjusted EBITDA ($569.8M TTM, ~10.2% margin) and Adjusted EPS ($5.81 TTM), both of which strip acquisition accounting and stock-based comp — scrutinized in Sections 6 and 10, and this quarter joined by a new presentation lever (the receivables-sale facility, Section 6).

2.5 Segment unit economics — why the two halves report so differently

The baseline’s caveat stands and matters more each quarter: Residential’s operating margin (8.6% of revenue in Q2’26) looks unremarkable, yet Residential is the higher-quality earner. A large share of Residential revenue is pass-through sited labour — on-site staff employed by FirstService and billed to the association near cost — which inflates the revenue base and compresses the percentage margin while adding little economic risk. The economically meaningful margin is on the management-fee and ancillary layer, which is high and growing. Brands books more value-add, more cyclical revenue at a structurally different margin shape. Judge Residential on Adjusted EBITDA growth, organic rate, and retention — not headline margin — and resist comparing the segments’ margins directly.

Verdict (Business Overview): Unchanged and re-confirmed: a coherent two-engine model — a genuinely high-quality recurring Residential franchise (now actively deepening its moat via Resilience First and pruning non-core aquatic ops) paired with a competently assembled, more ordinary, more cyclical Brands roll-up whose internal mix is shifting (Century Fire and franchise share-gains up; roofing and restoration weather down). The blended business is good and durable; it is not a uniformly wide-moat compounder, and conflating the two halves remains the most common analytical error here.

3. Industry Dynamics

3.1 Residential property management — structurally attractive, with two new nuances

The baseline verdict holds: a large, fragmented, growing, recurring industry with a regulatory tailwind. CAI projects the U.S. community-association count growing ~3–4k/yr to ~377,000 in 2026 (~one-third of U.S. housing); most associations remain self-managed or locally managed; the national players (FirstService Residential, Associa, RealManage, Inframark) hold low-single-digit shares of association count. Demand is non-cyclical and contracted.

Nuance one — Florida regulation softened at the margin. The operative 2025 change is HB 913 (effective July 1, 2025): it extended the Structural Integrity Reserve Study deadline to December 31, 2025, allowed associations to fund reserves via special assessments, lines of credit, or loans, permitted pooling of reserve accounts, and created a temporary pause on reserve contributions for associations completing milestone inspections. The SB 4-D structural core — milestone inspections at 30 years (25 coastal), SIRS, no reserve waiving for structural components — is intact. (The 2026 session produced incremental bills — HB 657, SB 1028 — confirmed only at headline level; low materiality.) Interpretation: HB 913 diffuses the cost shock that was driving board turnover and budget stress — the “outsourcing wave” thesis loses some urgency as associations get financing relief — but compliance complexity, the actual driver of professional-management demand, is unchanged. FSR’s own 2026 BENCHMARK High-Rise Report (August 5, 2026; ~1,500 buildings, 22 markets) documents insurance relief and rising reserve contributions as live budget realities — the regime is now a planning fact for boards, which supports the manager’s advisory value-add.

Nuance two — PE is now explicitly eyeing the space. Roland Berger published “Unlocking value in residential property management — Private equity’s next…” in July 2026, noting FSR/Associa/RealManage “remain relatively small compared with the total addressable market”; trade advisors count ~15 PE firms already invested in HOA management. The June claim that “capital is not flooding into Residential” needs softening — PE entry could raise FSR’s future tuck-under costs. Low-severity watch item, not a thesis change; no transformational Associa/RealManage/Inframark move surfaced in 2026 YTD, and FSR’s Q2 Florida Gulf Coast luxury-condo win streak (Viceroy Clearwater, 400 Central, Art House, Marina Pointe, Cricket Club) contradicts any retention-crisis read. Verdict: structurally good industry — reaffirmed, with the Florida tailwind modestly less urgent and a new competitive-watch item.

3.2 Restoration — the snapback is not happening in 2026

The structural description from June carries forward: a ~$40B+ U.S. market growing mid-single digits; demand largely non-discretionary and insurance-funded but highly variable with catastrophe activity; fragmented below a consolidating top (BELFOR #1, First Onsite #2, Servpro now 2,390+ franchises — +12% over three years under Blackstone ownership, with an internal CEO succession in April 2026 signaling continuity; Rytech absorbed into Summit Partners’ Fortify platform in December 2025; BELFOR continuing bolt-ons); and TPA/managed-repair economics (~5%+ of claim value in fees, pre-negotiated insurer-friendly pricing) capping contractor economics.

The update is the demand evidence, and it is decisive. 2026 is tracking quieter than 2025 — itself a quiet year:

• NOAA (August 6, 2026): maintained a below-normal Atlantic season outlook, lowered from May — 7–13 named storms (was 8–14), 2–6 hurricanes, 0–2 major hurricanes; 75% probability of below-normal (up from 55%). Season to date: two named storms (Arthur, Bertha), zero hurricanes; a strengthening El Niño is suppressing activity. CSU’s July forecast: 9 named storms vs a 14.4 average.
• H1 2026 insured catastrophe losses: Swiss Re Institute $42B (lowest H1 since 2020; 16% below the ten-year H1 average); Munich Re $44B (vs $80B in H1 2025); Aon $47B (vs $100B in H1 2025; lowest H1 since 2018 on its basis).

FSV’s Q2 commentary maps onto this: restoration revenue down slightly; the recovery is being built on non-weather large-loss share gains (national accounts, healthcare/government verticals) with an explicitly modest ~5% H2 guide because scoping, permitting, and insurance navigation slow conversion. Interpretation: the weather-normalization thesis is intact but deferred a second year — and the recovery, when it shows in revenue, will be share-led, which is higher quality than a storm bounce. Caveat: peak hurricane season (August–October) is still ahead; a single landfalling major hurricane changes the FY picture. Verdict: decent industry, cyclical and competitive — reaffirmed; normalization now a 2027 event.

3.3 Commercial roofing — the weak link, and the supply side is now correcting

The June verdict — “structurally unattractive; FSV’s worst exposure and the impairment flashpoint” — is reaffirmed and corroborated by two more quarters of evidence. Q2: −10% organic, backlog down year-over-year, “stubbornly weak and ultra-competitive,” new construction ex-data-centers down, new-construction roofers pivoting into reroof and intensifying competition, reroof demand deferrable. Two named weak regions: Las Vegas (>50% new-construction-weighted vs a ~30% portfolio average, lapping strong 2023–24) and Southwest Florida — post-Hurricane-Ian overcapacity, with “operations pulling out and closing their doors.”

The industry structure is the textbook Marathon capital-cycle sequence playing out: PE capital flooded in (roofing platforms 17 in early 2023 → 56 by late 2024; 134 acquisitions in 2024 alone; typical platform pricing 6–10× EBITDA residential, 9–13× commercial; QXO’s $11B Beacon acquisition closed; Tecta America under Altas Partners the largest PE-backed commercial platform) → returns collapsed (2025–26) → marginal capacity is now exiting the most overbuilt market (SW Florida) and deal competition is cooling (FSV: “fewer bidders,” PE sellers unwilling to crystallize losses against 2023–24 earnings). Non-residential leading indicators are K-shaped: the AIA/Deltek ABI printed 49.8 in March 2026 (still <50; newly signed design contracts have declined 25 consecutive months); the Dodge Momentum Index rose to 264.2 in April (+14.1% YoY, but commercial ex-data-centers only +5.8%) then dipped again in June. Any non-res recovery is a 2027, data-center/power-led story. For FSV-as-acquirer this is medium-term positive (cheaper tuck-unders as PE marks wash out) and near-term negative (organic decline and price pressure persist into 2027). Verdict: structurally unattractive sub-industry — reaffirmed; the bust phase has begun, which eventually restores returns for disciplined survivors.

3.4 Fire protection — attractive, code-driven, and the scale race is accelerating

The industry verdict (code-mandated, non-discretionary, partly recurring, mid-to-high-single-digit growth) stands. The competitive update runs against FSV’s relative position. APi Group reported record Q2 2026 results and raised FY guidance (July 30, 2026); its valuation re-rated through 2025–26 to ~20×+ EV/EBITDA (2025: ~20.5× vs ~15.8× in 2024; the stock ended 2025 at $38.26 vs $23.98 at end-2024) — the live benchmark for what the market pays for scaled, recurring fire/life-safety revenue. One caution: APi’s Safety Services organic growth has been decelerating (from 8.7% in Q3 2025); if that slide continues toward mid-single digits, the premium-recurring multiple support weakens sector-wide. Pye-Barker — the PE-backed consolidator — did 57 acquisitions in 2025 and 22 more in H1 2026 (200+ cumulative), now #3 on the SDM 100, 9,000+ employees, 47 states, with a dedicated President of Commercial Services (ex-Baird FIRE-services banker, hired July 2026) running acquisition strategy. Against this, Century Fire is executing well (+10%+ growth, high-single-digit organic, backlog up, two Q2 tuck-unders) but remains sub-scale — and the gap is widening faster than before. Verdict: good industry; FSV is a small-but-growing participant, not a leader — reaffirmed, with the scale gap widening.

3.5 Franchise home services — still at the cyclical floor

Unchanged verdict: economically attractive royalty model, cyclically pressured demand. Existing-home sales ran 4.06M SAAR in July 2026 (−1.7% m/m) and 4.09M in June — still near multi-decade lows; NAR cut its 2026 existing-home-sales forecast from +14% to +4%; mortgage rates projected ~6.5% for 2026. Conference Board Consumer Confidence printed 90.8 in July (down from 92.2). FSV’s home-service brands grew “slightly” via close-ratio and job-size gains — share capture with margin protection in a down market, which mildly supports the “thin moat + good model” framing. Verdict: reaffirmed.

3.6 Capital-cycle (Marathon) lens — modified, not reversed

Structurally, PE capital in FSV’s roll-up arenas keeps compounding (roofing platforms 17→56; Pye-Barker’s 200-deal pace; Rytech/Summit in restoration; Roland Berger teeing up residential management; Patterson on the Q2 call: “the level of private equity capital that we’re competing with increases every year… could be defined as a structural change”). Cyclically, 2026 shows cooling: fewer bidders, some funds pulled back, sellers unwilling to crystallize losses, FSV tuck-under spend tracking low (~$40M in Q2, FY like 2025’s ~$107M). The tell is FSV’s own behavior: with the public-private arbitrage temporarily inverted (public equity trading, in management’s words, at a “meaningful discount to smaller private market businesses”), the capital-allocation edge is expressing itself through a $248M buyback rather than acquisitions. Marathon’s framework says capital exiting roofing and restoration is precisely what eventually restores returns for patient, disciplined survivors; FSV is positioned as the patient balance sheet — now deliberately less patient-looking on its own leverage. Net: less negative than June for M&A-driven returns over a 2–3-year horizon; unchanged-negative near-term.

Verdict (Industry Dynamics): One structurally excellent industry (Residential — reaffirmed, with softer Florida urgency and a PE watch item), one good-but-cyclical industry whose trough just got a year longer (restoration), one good-but-sub-scale (fire — scale gap widening), one structurally poor industry now entering its bust phase (roofing), and a mixed franchise tail at the demand floor. Blended exposure remains above-average but not pristine; the capital cycle is cyclically cooling in FSV’s favor as an acquirer while remaining structurally crowded.

4. Competitive Position

4.1 The moat, segment by segment (Greenwald taxonomy) — all June verdicts reaffirmed, two strengthened by evidence

FirstService Residential — a genuine (if modest and local) moat: local economies of scale + customer captivity. REAFFIRMED, with marginally stronger evidence. The mechanism (route/operational density within metros; switching costs from financial-system migration, resident portals, re-papered vendor relationships, board-election inertia, fiduciary re-bid risk; ancillary banking/insurance/energy embed) is unchanged — and the financial fingerprint improved: +5% organic in Q2 (+4% Q1, H1 +4% all-organic), segment EBITDA margin +30bp to 9.9% in a weak macro, mid-90s% retention, and the Resilience First launch actively extending the ancillary moat into loss prevention. The share-stability and margin tests pass. Watch item: PE interest in HOA management (Roland Berger, July 2026) could raise future tuck-under costs. The moat remains local and modest — Associa competes at similar scale in many markets; it protects margins more than it commands outsized ROIC.

First Onsite (restoration) — weak/operational advantage. REAFFIRMED. Two consecutive quiet catastrophe years (H1’26 insured cat losses the lowest since 2020) mean First Onsite’s recent large-loss wins — across national accounts, healthcare, government — are capability- and relationship-led share gains without any weather tailwind. That is modestly encouraging evidence for the “operational advantage” framing. But it remains re-earnable, not structural: economics are partly insurer/TPA-controlled, the industry is fragmented, and PE consolidation (Rytech/Summit, Servpro’s 2,390+ franchises, BELFOR bolt-ons) keeps well-capitalized rivals in the field.

Roofing Corp of America — no moat. REAFFIRMED — and the Greenwald test is failing in real time. A fragmented market, a price war (FSV walked away from below-cost-priced work in SW Florida), deferrable demand, and −10% organic growth with a declining backlog is what the absence of barriers to entry looks like in practice. The 2023 platform’s earn-outs keep being reversed (−$17.2M of contingent-consideration fair-value reductions in H1 2026 alone; the liability has fallen from $47.0M to $33.7M), and the reporting unit’s goodwill ($363.4M) cleared its Q4 2025 impairment test by less than 5%. This line earns its skeptical treatment.

Century Fire — modest recurring-revenue captivity, sub-scale. REAFFIRMED. Code-mandated inspection/service stickiness is real at the branch level (+10%+ growth, high-single-digit organic, backlog up “well over prior year”), but the structural position is not improving: Pye-Barker’s 200±deal pace and APi’s 20×+ re-rating widen the scale gap. Advantage here is execution, not industry position.

Franchise brands — thin brand/franchisee lock-in. REAFFIRMED. Slight growth at ~10-year demand lows via share capture, with royalty margins protected — the model is doing its job; the moat is still thin.

4.2 The honest synthesis

Unchanged from June, restated with the new wrinkle: FirstService is one genuine local-scale/customer-captivity moat (Residential) attached to a competently run, largely moat-light roll-up (Brands), with company-level advantage resting on (i) operating skill, (ii) a disciplined, incentive-aligned acquisition machine, and (iii) public-versus-private multiple arbitrage. The new wrinkle: with the arbitrage temporarily inverted, that third edge is currently expressing itself through buying its own stock on leverage rather than buying private businesses — an execution advantage being redeployed, still not a fortress. The Greenwald share-stability test passes cleanly only in Residential; in Brands, shares are contestable and the capital cycle, while cooling cyclically, remains structurally crowded.

Verdict (Competitive Position): A durable but narrow advantage — reaffirmed. The Residential moat’s evidence base strengthened this quarter; the Brands moat-light verdict was corroborated by roofing’s real-time failure of the share/price test. Durable, yes; impregnable, no.

5. Growth History and Forward Opportunities

5.1 The historical record, updated through H1 2026

* includes a ~$35M one-time AR-sale benefit in Q2’26 (Section 6.3).

The long-arc story is unchanged: ~14% revenue CAGR 2021→2025, heavily acquisition-powered (RCA in December 2023 drove the +20% in 2024), Adjusted EBITDA compounding ~14.5% — but Adjusted EPS compounding only ~6% over that window, depressed by the 2022 dip, rising interest expense on acquisition debt, intangible amortization, dilution, and the 2025 cyclical trough. H1 2026 extends the mediocre per-share patch: Adjusted EPS +2%. Over a full decade the Adjusted-EPS CAGR has been mid-teens; honesty requires acknowledging the recent per-share compounding has been modest, and the Q2 print did nothing to change that. One mechanical caveat to the forward per-share path: the buyback shrinks the denominator (~3–4%/yr at the stated cadence), so reported per-share growth can now outrun operational growth by a few points — the reverse of the dilution drag of prior years.

5.2 Organic vs. acquired — the trough persists, and its composition changed

FY2025’s problem (all growth from acquisitions; consolidated organic ≈ 0%) persisted into H1 2026 with a composition shift:

• Residential: +4% reported, +5% organic in Q2; H1 +4% all-organic — contract-win-driven, margin-expanding, now aided by the aquatic-ops divestiture optics. This half of the growth engine is confirming.
• Brands: +4% reported, −1% organic in H1 (−3% in Q2) — and the drag is now explicitly roofing (−10% organic Q2), not restoration weather. Restoration was “down slightly” in Q2 with a rebuilt pipeline; Century Fire (+10%+) and the franchise brands (slightly up via share gains) are the offsets.

Q2’s +2% consolidated revenue / +3% Adjusted EBITDA came in below the trajectory the Q1 call implied (Q1 guided Q2 to “mid-single-digit top line growth”), on the roofing miss — a reminder that segment-level guides have been optimistic this year. Management’s FY2026 framing (revenue growth “similar to or modestly better” than H1’s +4%; mid-single-digit EBITDA growth; Q3 low-single-digit) implies the consolidated growth algorithm runs at roughly half speed for a second consecutive year.

5.3 Forward opportunities

1. Restoration normalization — now a 2027 story, share-led. The largest swing factor, deferred. Two consecutive below-average cat years (2025, 2026 YTD) have been logged; NOAA’s August 6 update keeps 2026 below-normal; the recovery is being built through non-weather large-loss wins (national accounts, healthcare/government verticals) converting over 12–18 months, plus a specialty-construction adjacency. A mere reversion to average weather on top of a rebuilt pipeline is the bull engine — and it remains largely unpriced. The August–October peak season is free optionality against a ~5% H2 restoration guide.
2. Residential penetration + ancillary attach — structural association-count growth (~+3–4k/yr per CAI), continued share gains from self-managed/local operators, deeper FirstService Financial/Energy attach, and the Resilience First cross-sell. FY guide: mid-single-digit growth with continued modest margin improvement.
3. Tuck-under M&A — H1’26 deployed $48.4M across 5 deals (Paul Davis Cleveland; California Closets Indianapolis; Scheffer’s Roofing; Titan Fire; GSC Fire); FY2026 expected “similar to last year” (~$107M). The constraint is supply (PE sellers won’t crystallize losses), not discipline (“mid-teens return on any of our capital deployment initiatives”). The cooling bid environment sets up cheaper 2027 deal flow.
4. Century Fire scaling — the clearest organic winner; guided to 10%+ Q3/Q4 growth; but note the sub-scale caveat vs APi/Pye-Barker (Sections 3.4, 4.1).
5. The buyback as a per-share growth lever — new this report: at anything near the stated cadence and ~2.5× leverage comfort, share-count shrinkage contributes ~2–4pp/yr to per-share EPS growth mechanically, lowering the organic bar the market is underwriting (Section 10.4).
6. Margin/operating leverage — Residential is demonstrating it (+30bp H1); Brands is not (roofing price pressure, home-services promotional spend).

Verdict (Growth): High-quality growth in Residential; mixed, cyclically depressed, still acquisition-dependent growth in Brands — with a new, debt-funded per-share lever bolted on. The 10%+ revenue algorithm is intact on paper but running at half speed; the quality of forward growth hinges on (a) the restoration conversion pipeline becoming revenue in 2027, (b) roofing merely stabilizing, and © tuck-under economics surviving a cooling-but-crowded deal market. The burden of proof has not moved: show me the organic re-acceleration. What has changed is that per-share growth can now be partially manufactured — which helps the base case and widens the gap between reported EPS progress and operational progress.

6. Financial Quality

6.1 Revenue, margins, and operating leverage

Q2 2026: revenue $1,449.2M (+2.4%), Adjusted EBITDA $161.7M (+3%, 11.2% margin, +10bp), GAAP operating earnings $99.7M (+2.4%), GAAP diluted EPS $1.00 (vs $1.01), Adjusted EPS $1.75 (+2%). H1: revenue $2,766.3M (+3.7%), Adjusted EBITDA $267.4M (9.7% margin, −10bp), Adjusted EPS $2.69 (+2%), GAAP diluted EPS $1.43 (vs $1.07 — H1’25 GAAP was depressed by $19.9M of earn-out fair-value increases, so the +34% GAAP EPS growth is a base effect, not an earnings-quality improvement).

Segment margins: Residential 11.2% Adjusted EBITDA in Q2 (+20bp), 9.9% H1 (+30bp) — the “continued operating efficiencies” claim is supported. Brands 11.5% Q2 (−10bp), 10.0% H1 (−50bp) on roofing competitive pressure and home-services promotional compression; management noted Q2 Brands margin came in better than Q1 and better than internal expectation. Corporate −$3.6M Adj EBITDA in Q2, flat.

Two profit-tailwind notes with a reversal ahead. Net interest expense fell — Q2 $15.5M vs $19.2M; H1 $30.8M vs $38.4M — on lower average debt. That tailwind ends here: the Q2 buyback was debt-funded (long-term debt +$195.1M net in the quarter), so H2 average debt steps up and the ~$10M/yr incremental after-tax interest cost of the buyback now sits in the forward run-rate. Tax rate 27% (vs 30% prior year), guided ~27% for FY2026. FY2026 capex guided ~$130M (cut from $140M).

6.2 The GAAP-to-Adjusted gap — quality of earnings

The bridge, per the company’s reconciliation (per diluted share, after tax):

Has the gap widened? Mixed, and the composition matters more than the headline. Q2 standalone widened slightly ($0.75 gap = 43% of Adjusted EPS vs $0.70 / 41%), on a higher RNCI redemption increment and rising intangible amortization. H1 narrowed ($1.26 / 47% vs $1.56 / 59%) — mechanically, because H1’25 GAAP was depressed by lumpy earn-out charges; a base effect, not a quality improvement. The structural direction is clear: the recurring wedges keep growing. Intangible amortization is $0.65/sh in H1’26 (annualizing >$1.30 vs $1.16 for FY25); SBC $0.32/sh (annualizing ~$0.60 vs $0.55). The two add-backs the June baseline flagged as most aggressive — the RNCI redemption increment (a real economic cost of the partnership model accruing to non-FSV owners) and SBC (a real, recurring, dilutive cost) — total $0.54/sh in H1’26.

Owner earnings (Adjusted EPS less RNCI increment and SBC): H1’26 $2.15 (H1’25: $1.99); TTM ≈ $4.71 (Adj EPS $5.81 − RNCI increment $0.52 − SBC $0.58). That sits inside the June baseline’s $4.50–$5.00 range, mid-point-ish. The QoE thesis is unchanged: capitalize ~$4.7, not $5.81; and if anything the recurring (non-lumpy) add-backs are drifting up.

6.3 Cash flow and capital intensity — with a new quality flag

Headline cash generation looks fine: Q2 OCF $129.8M (−20% YoY), H1 OCF $218.0M (+6.8%), capex $59.6M H1, H1 FCF $158.4M. TTM: OCF $459.8M, capex $124.4M (~2.2% of revenue — still genuinely asset-light), FCF ≈ $335M.

The new flag: the AR sale facility. In April 2026 FSV entered an uncommitted receivables-purchase facility (maximum capacity $300M; CIBC as purchaser, FirstService as guarantor, FirstOnsite as initial seller). In Q2 the company sold $35.3M of receivables for $34.9M and derecognized them — meaning the proceeds run through operating working capital. H1 OCF of $218.0M therefore includes a ~$35M one-time benefit: underlying H1 OCF ex-AR-sale ≈ $183M, i.e. DOWN ~$21M (−10%) year-over-year, not up 7%. The MD&A states the facility’s purpose plainly: “reduce interest costs and reported financial leverage.” The scale is small so far ($646.1M of revolver availability at June 30 already nets this), and uncommitted means availability is at CIBC’s discretion — but it is a new lever that flatters both OCF and net-debt optics, and its stated purpose is cosmetic as much as economic. Monitor utilization every quarter.

6.4 Balance sheet — the “conservative” label is spent

This is the section whose June characterization must be revised. At June 30, 2026: total debt $1,250.4M ($13.6M current + $1,236.7M non-current), cash $173.4M → net debt $1,077.0M, versus $928.3M at YE2025 and $864.3M at Q1 2026. Against TTM Adjusted EBITDA of $569.8M, net leverage ≈ 1.9× (company-reported basis 1.8×; ROIC’s GAAP-denominator print is 2.00×) — versus ~1.5× at the June baseline and ≤1.65× at YE2024. The balance sheet moved from conservative to the top of FSV’s historical range in one quarter, deliberately, to buy stock at ~$136, with the CFO stating comfort “at least to the mid-2s… 2.5 times would be a strong comfort level.”

The structure beneath remains sound: a $1.75B unsecured revolver (February 2030 maturity, 0.20–2.50% pricing grid, $646.1M undrawn), laddered private-placement notes (NYL/Prudential, coupons 4.53–5.64%, maturities 2029–2032), the NYL $250M uncommitted shelf running to April 2027 (the Prudential $300M shelf expired September 2025, unrenewed), covenant compliance affirmed, liquidity “>$800M” per the CFO. Cross-checks (ROIC, GAAP basis): FY2025 ROIC 7.8%, ROE 11.3%; Q2’26 current ratio 1.67, Altman Z 5.0, EBITDA/interest coverage ~9.9×. Goodwill + intangibles are now $2,229.7M against shareholders’ equity of $1,218.1M (equity fell from $1,376.0M at YE25 on the buyback) — tangible book is deeply negative, and the goodwill-heavy structure is why consolidated ROIC stays in the high single digits. RNCI stands at $507.4M (YE25: $486.2M); the contingent-consideration liability fell to $33.7M from $47.0M — earn-outs being reversed, consistent with Brands/roofing softness.

6.5 The goodwill-impairment flashpoint — still live, pin still in

No new impairment test and no charge in Q2 2026. The disclosure is identical in the interim FS and MD&A: the Brands reporting unit tested in Q4 2025 (roofing, $363.4M goodwill) passed with fair value exceeding carrying value by less than 5%; no additional indicators of impairment were noted during Q2 2026 and no changes to key inputs and assumptions. But the trigger condition persists — roofing organic is still falling (−10% in Q2, guided down mid-single digits in Q3) and management itself cites “competitive pressures in the roofing industry.” Per the FY25 disclosure, a 0.5× reduction in the valuation multiple or a 3% EBITDA decline roughly erases the cushion. The Q4 2026 annual test is the decision point.

Verdict (Financial Quality): Unchanged in substance, with one new wrinkle: the economics remain genuinely good — low capital intensity (~2.2% of revenue), ~$335M TTM FCF, Residential compounding organically at 4–5% with margin expansion — but the reported numbers are still dressed at their most generous (recurring RNCI/SBC add-backs ~$0.54/sh per half and rising, intangible amortization annualizing >$1.30, and now a receivables-sale facility flattering OCF and leverage optics), and management has chosen to lever the balance sheet to ~1.9× to repurchase stock, converting the prior margin of safety into a bet on its own multiple. Adjusted-EPS growth of +2% in H1 against +14% in FY2025 confirms the per-share engine is still idling on the Brands/roofing cycle. Capitalize owner earnings of ≈$4.71, not the $5.81 headline.

7. Capital Allocation

7.1 Governance and control — with a correction to the June report

The June baseline’s core governance facts stand: single class of one-vote common shares since May 2019; no holder ≥10% (the 2026 circular confirms); founder Jay Hennick non-executive, independent Chairman at ~6% (excluded from the option plan by design); 8 directors, 7 independent; PwC audit; say-on-pay 92.4% approval.

The correction: the baseline stated that FSV’s insiders “became subject to U.S. Section 16 reporting in March 2026.” The EDGAR record does not support this. A full-history check of CIK 0001637810 (submissions API and filing index, 2021-08 through 2026-08-14) shows zero Forms 3/4/5 — ever, before or after March 2026 — and the 2026 management circular contains no mention of Section 16 or SEDI status. As a foreign private issuer, FSV insiders remain Section 16-exempt; principal shareholders likewise (they file 13D/G only; Hennick reports via Canadian SEDI). The only insider-adjacent EDGAR filing is a Form 144 (February 11, 2026): director Frederick Reichheld sold 6,000 shares ($971,280) from a February 2025 stock award — routine and immaterial, and Form 144 is not Section 16 reporting. The practical consequence: insider-trade transparency for FSV is weaker than the June report implied — the insider record is Canadian SEDI-only, and “no U.S. insider filings” is a transparency gap, not evidence of no trading. That gap matters this period, because the period’s most consequential insider datum sits on SEDI: CEO Scott Patterson disposed of 46,700 common shares on February 18, 2026 at $213.50–$215.464 (~$10.0M, a −32.2% reduction of his direct holdings), per Canadian Insider’s SEDI-based reporting (February 20, 2026). The sale coincided with large 2025 option exercises (Patterson exercised 150,000 options in 2025 at strikes of $111.36/$140.03, a $7.71M notional gain, per the circular) — likely monetization of exercised awards — but no 10b5-1-equivalent plan disclosure is available for Canadian filers, and no offsetting open-market insider buying was found anywhere in 2026. State the tension plainly and factually: the CEO sold ~$10M of stock at ~$214 in February; four months later the company spent $248M buying stock at ~$136. Both facts are verifiable; the interpretation — personal diversification after exercise vs. a difference of opinion on value — is not resolvable from available disclosure.

The countervailing ownership fact: Orbis Investment Management (Bermuda/US) filed a 13G on May 15, 2026 (3,968,227 shares, 8.6%, as of March 31) and a 13G/A on August 14, 2026 reporting 4,044,289 shares = 8.8% as of June 30 — passive (Rule 13d-1(b)), not activist. A value-oriented fundamental manager built an ~9% position through the drawdown, in the same window the company was buying. Orbis was not disclosed as a >5% holder in the June report; it is now one of FSV’s largest.

7.2 The “partnership model” (RNCI) as capital allocation

The mechanism is unchanged: operating managers hold minority equity in the subsidiaries they run (Redeemable Non-Controlling Interests), with FSV call rights and partner puts priced by a formula — a fixed multiple of the subsidiary’s trailing two-year average earnings, less debt — typically below FSV’s own public multiple, making buy-ins accretive while keeping operators aligned. The 2026 data: RNCI balance $486.2M (YE25) → $507.4M (June 30, 2026) — bridge: share of earnings +$9.4M; redemption increment +$9.9M; distributions −$13.6M; purchases of interests −$10.2M; RNCI recognized on acquisitions +$26.3M; other −$0.6M. The aggregate formula redemption amount is $432.4M — below carrying value because some formula prices sit below inception amounts. Buy-in pace slowed: $10.2M in H1’26 vs $29.3M in H1’25 (FY25: $33.8M) — consistent with cash redirected to the NCIB and fewer partner put exercises. Total H1 cash to minority partners: $23.8M (vs $40.9M H1’25). The redemption increment was $0.22/sh in H1 ($0.21 in Q2 alone — re-accelerating) vs $0.35 in H1’25. Nothing here changes the June assessment: a genuinely clever alignment device with a real, recurring per-share cost that Adjusted EPS strips out.

7.3 M&A track record and discipline

H1 2026: 5 acquisitions, all in Brands, ~$55M total consideration ($48.4M cash net of cash acquired + $6.6M contingent) — a Paul Davis franchisee (Cleveland, Q1), a California Closets franchisee (Indianapolis, April), Scheffer’s Roofing (Kansas City, ~June 10; RCA’s 16th acquisition, 27 branches), Titan Fire Protection (Tampa), and GSC Fire and Security (Austin/San Antonio). No platform-sized deal YTD; no multiples disclosed (FSV never discloses tuck-under terms). The CFO put Q2 spend at “just over $40M,” guided FY2026 M&A “similar to last year” (~$107M), and was explicit that the constraint is supply, not discipline: “we’re not seeing many quality companies come to market… fewer companies come to market” — PE owners of roofing/restoration assets won’t sell at marks that crystallize losses against 2023–24 earnings; “the number of bidders for opportunities is probably lower right now.” The hurdle, verbatim: “we target a mid-teens return on any of our capital deployment initiatives.”

The honesty check from June continues to pass — and to warn. H1 2026 contingent-consideration fair-value adjustments were −$17.2M (earn-out reversals); the liability fell $47.0M → $33.7M. Earn-outs keep being marked down on roofing-linked underperformance: honest booking, and confirmation that the 2023 RCA platform continues to underperform its price. The June verdict stands: cadence and funding discipline exemplary; the returns record blemished specifically by the cyclical, construction-exposed platform bought near the cycle peak. FSV’s edge is in recurring-services tuck-unders, not in buying cyclical platforms at full prices.

7.4 The buyback — the quarter’s defining capital-allocation event

The June report’s central criticism was an unused NCIB at a decade-low valuation. It has been decisively answered:

• Q2 2026: 1,827,750 shares repurchased for $248.4M at an average of $135.91, all cancelled (zero in Q1 2026; zero in all of 2025). Pace: 931,182 shares at $132.38 through May 31; ~896,568 at ~$139.6 implied in June — buying accelerated after the NCIB amendment and was deliberately concentrated pre-blackout at the lows (the CFO confirmed the ASPP’s parameters were not met during the pre-earnings blackout: “we’ll be out of blackout on Monday, and then we can be active without our hands tied”).
• June 2, 2026: the NCIB was amended to the TSX maximum — 4,118,199 shares (10% of public float), up from 1,600,000 (3.9%) — plus an Automatic Share Purchase Plan, effective June 4. Remaining authorization at June 30: 2,290,449 shares. The current NCIB expires August 25, 2026; no renewal had been announced as of August 15 (last year’s was announced ~August 13–19 — the nearest-dated signal event after this report).
• Funding and intent: Q2 long-term debt draw +$195.1M net; leverage 1.5× → 1.8× (company basis); CFO: comfortable “at least to the mid-2s… 2.5 times would be a strong comfort level”; “we have been buying at current levels and you can be sure that we will continue to do so.” Rationale, verbatim: buybacks executed when “our large, diversified enterprise [is] trading at a meaningful discount to smaller private market businesses in our respective industries” — the same public-private arbitrage logic as the RNCI model, applied to the company’s own stock.
• Share-count effect: 45,722,486 (YE25) → 45,981,761 (Q1’26, option exercises) → 44,154,011 (June 30, 2026) = −3.4% net in H1. The chronic ~1%/yr SBC dilution has been reversed, decisively. Q2 diluted average shares fell year-over-year (45.34M vs 45.66M) for the first time in years.

The EPS arithmetic is worth stating precisely, because it disciplines what the buyback is and is not. At a ~5.5% pre-tax blended cost (~4.0% after the 27% tax rate), $248.4M drawn costs ~$10.0M/yr after tax (~$0.23/sh on the remaining share count). Pro forma: TTM Adjusted net income (~$265M) less that interest, over 44.15M shares, gives $5.77 vs $5.76 pre-buyback — ~+0.2%, essentially EPS-neutral on Adjusted EPS; ~−1% on owner earnings; ~+1% on FCF/share. At ~24× Adjusted EPS the earnings yield (~4.2%) ≈ the after-tax cost of debt (~4.0%): the classic zero-accretion zone. The buyback manufactures no EPS; it creates value only if intrinsic value exceeds the ~$136 paid — an intrinsic-arbitrage bet, not an accretion trade. It is rate-sensitive (+100bp on marginal debt cost flips it ~−0.5% dilutive on Adjusted EPS), and it converts balance-sheet slack into equity convexity: in the bear scenario the same buyback that assists base-case EPS has consumed the slack that would have absorbed a roofing impairment.

7.5 Compensation alignment — unchanged architecture, rebuilt dilution capacity

The 2026 circular (FY2025 pay year) shows no change to incentive architecture: annual bonus = 3-year trailing average Adjusted-EPS growth modified by 3-year average organic revenue growth (≤3% → 80% factor; 4–5% → 100%; ≥6% → 120%); CEO bonus capped at 5× salary; LTI remains ~100% stock options; still no ROIC, TSR, or capital-efficiency metric. 2025 inputs: 11% AEPS growth, 5% organic → CEO bonus $1,603,100; CEO total comp $7,801,303; CFO Rakusin $4,638,887; no discretionary bonuses. The circular’s own table concedes the point the June report made: 5-year TSR of 26.1% vs the S&P/TSX Composite’s +110.8% — while AEPS-gated bonuses kept paying.

The new item, and it deserves emphasis: in the same year the buyback retired 1.83M shares, shareholders approved (April 1, 2026 special business) a +2,000,000-share increase in the option-plan reserve — 7,313,500 → 9,313,500 shares, ~20.4% of shares outstanding — plus a new annual grant limit for non-employee directors. Options outstanding at December 31, 2025 were 2,536,190 (5.5% of shares). Interpretation: net anti-dilution progress is real, but it is partly mortgaged forward — dilution capacity is being rebuilt behind the buyback, and compensation still runs ~68% through options against a bonus metric (Adjusted EPS) that adds back the SBC that funds it.

7.6 Dividends and the full capital stack

Quarterly dividend $0.305/share ($1.22/yr) — set with the +11% raise announced February 3, 2026 (more than a decade of ≥10% annual increases); paid again in July. H1 dividends $26.6M; yield ~0.87% at ~$141. The H1 2026 capital stack, in round numbers: FCF ~$158M (headline) → buybacks $248M + dividends $27M + M&A $48M + RNCI buy-ins/distributions $24M — i.e., the company outspent its cash generation on shareholder return plus growth by roughly $190M, funded with debt, by design.

Verdict (Capital Allocation): Upgraded from June’s “intelligent, not brilliant” to “intelligent and, for the first time in years, genuinely opportunistic.” Full marks on the specific NCIB question: $248M concentrated into one quarter at $135.91 near the period low, the bid upsized to the 10% TSX maximum mid-stream, an ASPP put in place, stated leverage headroom to ~2.5×, and a coherent public-private-arbitrage rationale. Held back from “brilliant” by four deductions: (a) the CEO’s ~$10M sale at ~$214 four months before the company bought at ~$136, with FPI-level (SEDI-only) disclosure and no visible plan structure; (b) the +2.0M option-reserve top-up approved behind the buyback; © still no ROIC/TSR incentive metric while the circular itself shows 26% five-year TSR against the index’s 111%; (d) roofing earn-out reversals confirming RCA underperformance. Watch items: the August 25 NCIB expiry/renewal, H2 buyback pace at higher prices, and AR-facility utilization.

8. Changes and Headwinds — Last Two Years

Strategic / portfolio:

• The capital-allocation posture inverted in Q2 2026: from an unused NCIB and ~$100–150M/yr of M&A to a $248M debt-funded buyback against $48M of H1 M&A — with the NCIB maxed to 10% of float (June 2, 2026) and an ASPP attached. The public-private arbitrage that powered the growth algorithm is, for now, being run against the company’s own stock.
• Tuck-under build-out continued (Century Fire: Titan, GSC; RCA: Scheffer’s; franchise buy-ins), with no platform deal since RCA (December 2023).
• Residential divested non-core single-family aquatic operations (start of Q2’26) and launched Resilience First (July 6, 2026) — pruning plus captivity-deepening.
• Dividend raised +11% to $1.22/yr (February 2026); option-plan reserve increased +2.0M shares (April 1, 2026); new $300M uncommitted AR-sale facility (April 2026); the Prudential $300M note shelf expired September 2025 unrenewed.

Operating environment (headwinds):

• Weather, year two. 2025 was quiet; 2026 is tracking quieter (H1 insured cat losses $42–47B, lowest since 2020; NOAA August 6 below-normal outlook at 75% probability; two named storms, zero hurricanes to date). The restoration trough now spans two full years; recovery is being built share-led for 2027.
• Roofing. From soft to worse: −10% organic in Q2’26 (below the Q1-implied trajectory), backlog down year-over-year, SW Florida overcapacity with operators exiting, Las Vegas lapping new construction; Q3 guided to mid-single-digit organic decline. Non-residential leading indicators (ABI <50, design contracts down 25 straight months) say any recovery is 2027 and data-center-led.
• Home services. Housing turnover ~4.0–4.1M SAAR near multi-decade lows; consumer confidence 90.8; growth only via share capture; no market help expected in H2.
• Residential strengthened throughout — 4–5% organic, margin expansion — with the Florida post-Surfside cost shock partially diffused by HB 913’s financing relief (the compliance-complexity tailwind intact).
• Interest expense: the H1 tailwind (lower average debt) reverses in H2 as the buyback-funded draw annualizes.

Leadership/structure: stable — Patterson (CEO), Rakusin (CFO), Hennick (non-executive Chairman). New anchor holder: Orbis at 8.8% (passive), built May–August 2026. One director Form 144 sale (Reichheld, immaterial) and the CEO’s February SEDI-reported sale (Section 7.1).

Verdict (Changes/Headwinds): The two-year picture is still “a pause, not a break” at the company level — but the composition shifted. The identifiable, cyclical headwinds (weather, housing turnover, non-res construction) persisted or deepened; the roofing problem looks incrementally more structural; and management responded not by waiting but by re-levering to retire 4% of the float at the lows. Net: the operating thesis weakened slightly at the margin (snapback deferred, roofing worse) while the capital-allocation thesis strengthened materially — and the two are now linked, because the buyback raises the stakes on H2–2027 organic delivery.

9. Risk Analysis

Catastrophic-loss assessment: the probability of permanent capital impairment remains low, but the margin has thinned at the edges. The June report could point to a fortress balance sheet as the ultimate backstop; at 1.9× leverage with stated comfort to 2.5×, the backstop is now partially spent — deliberately, on the stock itself. The realistic downside is unchanged in kind (multiple compression + an earnings air-pocket + a roofing write-down) and slightly larger in degree.

Verdict (Risk): The risk profile is moderate and predominantly cyclical/identifiable, with one self-inflicted addition: the leveraged buyback raises the cost of being wrong about 2027. The two exogenous risks that matter most (multi-year restoration trough; roofing impairment) are joined by a financial one (leverage into the trough). Tail risk remains low.

10. Valuation Discussion (Embedded Expectations)

No price target and no recommendation are expressed in this section. It frames embedded expectations, comps, scenarios, and a sum-of-the-parts, and identifies what the market appears to be underwriting.

10.1 Where the multiple sits — recomputed from filings at $140.81 / 44,154,011 shares

Market cap = $6,217M. Three EV constructions, all stated so nobody mixes them: simple (net debt only, filing basis): $7,294M — 12.8× TTM Adjusted EBITDA, 1.30× revenue. Plus RNCI at carrying value ($507.4M — the minorities’ share of consolidated EBITDA is inside the $570M, so the claim belongs in EV): $7,802M — 13.7×. ROIC-style (lease-inclusive debt $1,559.3M less cash $173.4M, plus RNCI): $8,111M — 14.2×, 1.45× revenue. ROIC’s own Q2’26 print (EV $8,168M at a slightly stale market cap) ties when recomputed live: 15.0× GAAP EBITDA / 14.2× company-Adjusted.

Against its own ten-year history, the AZI valuation index (pull verified fresh, 2026-08-14) puts FSV at a composite percentile of 3.6 (P/E 4.5 / P/B 1.6 / P/S 4.8) — the same bottom ~3–4% zone as June. (The AZI P/B level uses a stale-hybrid BVPS; the percentiles are used, levels computed independently.) Read: the price is flat since June but the per-share stack improved marginally (fewer shares, higher net debt). The stock remains fair-to-full on trough earnings capitalized conservatively (owner EPS ~30×), cheap only against its own history and only if Brands normalizes — the June verdict survives Q2 intact.

10.2 What de-rated it — and what has not re-rated it

The five June drivers all still bind: (1) the post-2021 rate reset; (2) the restoration weather trough — now two years deep; (3) roofing softness + the goodwill scare — worse, with the Q4’26 test ahead; (4) M&A multiple-creep — cyclically cooling but structurally crowded; (5) ~0% consolidated organic growth — H1’26 consolidated organic is still roughly flat (+4% reported with organic concentrated in Residential). Four of five remain cyclical/identifiable. The new facts — the buyback, Orbis — are what a re-rating’s supply side looks like (float shrinking from both sides); they have not yet produced the re-rating, which awaits company-level evidence. The tape evidence agrees: ~65–73% of FSV’s variance is idiosyncratic, so factor rotation neither caused the drawdown nor will end it (Section 11.4).

10.3 Comparable companies — refreshed (ROIC Q2’26 data, TTM)

† CIGI/RBA/APG GAAP P/Es are distorted (heavy NCI at CIGI, amortization/transaction drags at RBA/APG) — EV/EBITDA is the only honest cross-sectional metric here, same conclusion as June. The forward-P/E column from the June table was not refreshed (no reliable free forward-estimate source was available for this update); flagged, not invented.

FSV at ~14–15× EV/EBITDA still sits above CIGI, below RBA/ROL/APG — the same sandwich as June, but the spread to the quality names widened ~1 turn (APG re-rated to ~22.5×, ROL ~24.5×; FSV de-rated on EV). The market continues to price FSV as “better than a brokerage roll-up, much worse than a recurring-services compounder” — the Residential half is still not being paid for. Two cross-reads discipline the table. First, the CIGI pair (same founder, same RNCI model, lower multiple on transactional-brokerage earnings) validates that the market pays for recurring quality and discounts transactional cyclicality — arguing FSV’s Residential half is under-credited inside the blend. Second, the factor-twin evidence (Section 11.4): the tape does not trade FSV against CIGI/CBRE/JLL at all — its empirical pricing neighbors are dividend-aristocrat ETFs and CPRT/EFX/BN/BAM/BX/SHW. Treat the fundamental comp table as valuation scaffolding, not as the pricing cohort.

10.4 Embedded-expectations (reverse) analysis — what changed since June

Method unchanged for comparability: hold the multiple flat at ~24.2× Adjusted EPS (or ~14× EV/Adjusted EBITDA), 0.87% dividend yield, ~9–10% required equity return → the price is met by ~8–9% forward Adjusted-EPS growth with no multiple re-rate. Three things changed:

1. The snapback got deferred, again. 2026 is tracking quieter than 2025 (H1 insured cat losses lowest since 2020; NOAA below-normal). The H2 restoration guide of ~+5% is explicitly built on non-weather large-loss wins converting over 12–18 months — revenue skewed to Q4’26/2027. Embedded expectations now embed two trough years, not one; June’s “no snapback priced” is more true, not less.
2. The buyback lowers the fundamental bar. If H2–2027 repurchase cadence continues anywhere near the stated intent (leverage comfort to ~2.5×; “we will continue to do so”), share-count shrinkage contributes ~2–4pp/yr to per-share EPS growth mechanically. The organic+M&A EPS growth the price requires drops to roughly ~5–7%/yr — below even the FY26 guidance trajectory of ~4–5% revenue growth plus buyback assist. A genuinely low embedded hurdle.
3. Offsetting new conservative input: the market must now also underwrite ~$10M/yr of incremental interest cost and a 1.9× (top-of-historical-range) balance sheet — the “fortress balance sheet” option embedded in June’s price is spent.

Net read: the market is underwriting ~high-single-digit Adjusted-EPS growth, no restoration snapback through 2026, continued roofing decline, and modest buyback assist — and still only getting to ~9–10% expected return. A return to average weather, or roofing merely stabilizing, remains almost entirely unpriced. June’s asymmetry statement holds verbatim, with the buyback as a new partial floor under the per-share math.

10.5 Scenario analysis (5-year; Adjusted EPS $5.81 TTM × exit multiple)

Deltas versus June (zones ~$115–140 / ~$190–215 / ~$280–310): the value zones are essentially unchanged, with two composition changes worth stating. (a) The base-case CAGR is now ~1–2pp easier to hit organically because the buyback does part of it. (b) The bull case is now explicitly a 2027 story — June’s bull assumed the snapback could begin in 2026 H2. The bear case still ≈ today’s price (~$140 midpoint): two trough years and an impairment are roughly what $141 already pays for. The favourable-but-not-riskless asymmetry framing survives; what has changed is that management has now spent $248M of the company’s own balance sheet expressing the same view.

10.6 Sum-of-the-parts sanity check — re-run on TTM segment Adjusted EBITDA

TTM segment EBITDA (bridged FY25 − H1’25 + H1’26; ties to the $569.8M TTM total): Residential $233.1M / Brands $350.5M / Corporate −$13.9M.

Versus the current EV of $7.3B simple / $7.8B with RNCI / $8.1B full-load: on trough EBITDA the full-load EV sits at or just above the high SOTP case — headline ~14–15× “looks full” because the Brands denominator is trough. Normalize Brands and the SOTP midpoint (~$7.6–8.9B) covers the current EV. Identical structure to June: full on trough, fair-to-cheap on normalized — a cyclical-trough read, not structural overvaluation. The normalization is now a 2027 event, not a 2026 one.

10.7 Which earnings to capitalize

Unchanged discipline, updated numbers: the honest anchor is owner earnings of ≈$4.71 TTM (Adjusted EPS $5.81 less the RNCI redemption increment $0.52 and SBC $0.58) and FCF of ~$300M underlying ($335M headline less the ~$35M AR-sale benefit; ~4.8% yield ex-AR). On those anchors FSV is fairly valued today on trough earnings and attractively valued only if Brands normalizes. The buyback, executed at $135.91, is management’s own mark on that question — worth ~4% of the company — but it is a claim, not a proof, and it was financed with the balance sheet that previously provided the margin of safety.

Verdict (Valuation): The market is pricing FSV near a decade-low relative valuation (3.6th percentile of its own history) that embeds two trough years, continued roofing decline, and modest buyback assist. On conservative anchors it is fair today; on normalized Brands earnings it is cheap. The embedded-expectations bar fell mechanically (buyback assist) even as the operating bar slipped a year (snapback deferred) — favourable but not riskless asymmetry, with the restoration conversion as the un-priced optionality, the Q4’26 roofing test as the priced-in-but-confirmable downside, and the NCIB renewal (due by August 25) as the nearest-dated signal on whether the debt-financed floor holds.

11. Variant Perception

11.1 Consensus view — a show-me reset, unchanged by ten weeks and a quarter-billion-dollar buyback

Post-Q2 consensus (fact on estimates, interpretation on framing): Buy consensus intact across ~10 analysts; mean price target ~$174 (~+24% implied); CIBC cut $204→$175 on July 24; no Sell ratings surfaced. The print (EPS beat, revenue miss, roofing −10% organic versus guide, “stronger H2” on restoration backlog) produced a −7.4% flush on 5× volume that fully recovered within three sessions. The consensus framing: good company, bad tape; roofing is now the named problem; restoration recovery deferred to 2027; pay a trough multiple, collect the buyback, wait for proof. This is a show-me reset, not a thesis break — and it is ~unchanged from June, which is itself information: ten weeks, one earnings print, a maxed-out NCIB, and $248M of repurchases later, the stock has not moved.

11.2 The strongest bull case — updated

The market is pricing two trough years as the trend while every identifiable incremental fact points the other way. Residential keeps confirming (+5% organic, margin +20–30bp, captivity-deepening Resilience First launch). Restoration is inflecting on share, not weather — pipeline rebuilt to “historically healthy,” large-loss wins across national accounts, a higher-quality recovery than a storm bounce, converting into 2027. Roofing’s capital cycle is turning for the acquirer (SW Florida capacity exiting, PE sellers stuck, fewer bidders — cheaper tuck-unders ahead). And the two most informed buyers — the company (1.83M shares at $135.91, NCIB maxed to 10% of float, leverage comfort to 2.5×) and Orbis (8.8% and adding) — are buying exactly what growth funds are abandoning, at a ~4th-percentile own-history valuation, with the bear case already ≈ price.

11.3 The strongest bear case — updated

The drag migrated from exogenous to structural, and management just levered up to deny it. Brands organic has now printed negative for six-plus quarters and the Q2 driver is roofing — a no-moat, PE-crowded, price-war business whose goodwill cushion is <5% with the annual test in Q4’26 — plus home services pinned at 10-year-low consumer/housing. Adjusted EPS grew +2% in H1’26; on owner earnings (~$4.71) the stock is ~30×, and the recurring add-backs (RNCI increment + SBC ~$0.54/sh per half and rising; intangible amortization annualizing >$1.30) keep widening the gap the market must believe. The buyback is debt-funded at the top of the historical leverage range, with the CEO having sold ~$10M at ~$214 four months earlier — financial engineering substituting for organic growth, executed just as OCF ex-AR-sale fell ~10% year-over-year. If 2027 weather is also benign and roofing impairs, the “trough” was the trend and ~18× on ~$7.8 EPS is where this lives.

11.4 The factor-positioning read — what the tape is and is not pricing

Facts (FactorsToday/AZI, pulled 2026-08-14/15): Momentum loading −0.12 with Quality ≈ 0 (zeroed); Market +0.69, LowVol +0.31, Value +0.09; model R² 0.27–0.37 → ~65–73% of variance idiosyncratic (specific vol 23.4% annualized). Trailing 12m return −28.9%, 3m +8.8%; beta 0.66; five-year max drawdown −44% despite the low beta — idiosyncratic, not market, drawdowns. Factor-similar cohort = dividend-aristocrat/quality ETFs plus CPRT/EFX/BN/BAM/BX/SHW — not real-estate-services. Regime: no style factor at an extreme; Value mildly in favor (z63 +0.99); FSV’s home-construction/real-estate tilts neutral-to-soft. Positioning evidence: FSV was a named Q4’25 detractor in at least one growth fund’s letter (Alger Weatherbie, February 3, 2026) and the subject of a full 13F exit (January 2026 coverage); volume unremarkable; price flat since June.

Interpretation: the tape prices FSV as a de-rated, abandoned, idiosyncratic story stock — not a crowded momentum trade and not a factor-rotation victim. The dangerous configuration — momentum carrying a low-quality business — is absent; this is the inverse (negative momentum on a neutral-quality loading), historically where value entries cluster. The drawdown was stock-specific news (weather, roofing), which cuts both ways: factor tailwinds will not rescue it and factor headwinds did not cause it — only company-level evidence (restoration conversion, roofing, buyback follow-through) re-prices it. Positioning reads abandoned, not yet accumulated — with two exceptions accumulating: the company itself ($248M at $135.91) and Orbis (8.8%). Stated as positioning context, not a price call.

11.5 The assumptions that matter most (and their falsification tests — restated where June’s wording misfired)

1. Restoration revenue re-accelerates within the next 1–2 years, weather or no weather. June’s test (“two consecutive quarters of restoration organic re-acceleration on storm activity”) is restated, because management disclosed a non-storm recovery path (large-loss wins, 12–18-month conversion) that the original wording would misread as failure: restoration/Brands-restoration organic ≥ mid-single digits for two consecutive quarters counts regardless of storm contribution; conversely a storm-driven spike alone, without backlog conversion, does not. If H2 prints +5% twice, the restated bull test passes without a single named storm. Falsify (bear): 2027 prints flat/negative restoration organic even as the converted backlog lands.
2. Residential’s moat is intact and compounding. Confirm (bull): continued 4–5% organic + margin expansion — hit twice now (Q1 +4%, Q2 +5%, +30bp H1 margin). Falsify (bear): organic below mid-single digits or retention slipping below the low-90s%.
3. Roofing does not structurally impair value. Falsify (bear), restated: a Q4’26 impairment confirms the leg fully only alongside continued organic decline or disclosed assumption deterioration (0.5× multiple or 3% EBITDA decline erases the cushion). Confirm (bull): roofing organic stabilizes (backlog’s sequential May→June improvement extending).
4. Tuck-under M&A stays accretive despite PE competition. Tracking: H1 $48.4M at undisclosed but hurdle-gated (“mid-teens”) pricing; earn-out reversals corroborate RCA overpayment specifically, not a broken model. Falsify (bear): multiples paid rising materially / incremental ROIC falling.
5. Adjusted EPS is a fair proxy for economics. Ambiguous, drifting bearish: Q2 gap 43% of Adjusted EPS (vs 41%); recurring wedges rising; market has not re-anchored (24× holds; the company itself is buying at that multiple). Falsify (bear): re-anchor to owner earnings → ~30× de-rating.
6. (New) Management’s intrinsic-value claim is right. The buyback only creates value if the enterprise is worth more than ~$136. Confirm: NCIB renewed at scale and buying continues without leverage drifting past ~2.5×, followed by Brands normalization. Falsify: buyback stops at the August 25 expiry and Brands keeps deteriorating — meaning even management’s bid was liquidity, not conviction.

Verdict (Variant Perception): June’s variant — “a weather trough mistaken for structural decline” — has split. The restoration half strengthened (the recovery is arriving share-led, more durable, just later — 2027). The roofing half weakened (a genuinely structural-looking, no-moat problem with a live impairment trigger). The new layer June lacked: the buyer of last resort is now the company itself, levering deliberately toward 2.5× to retire float at $135.91 — which converts the valuation debate from “is it cheap?” into “is management’s intrinsic-value claim right?” Orbis’s 8.8% is the only institutional echo. The honest stance remains constructive-but-demanding: the bear’s owner-earnings (~30×) and roofing-structure points stand unrefuted.

12. Fact vs. Interpretation

13. Open Questions

1. NCIB renewal. The bid expires August 25, 2026 — ten days after this report — with no renewal announced as of August 15 (last year’s pattern: announced ~August 13–19). Renewed at scale, resized, or allowed to lapse? This is the nearest-dated signal on whether the debt-financed floor persists, and on whether H2 buying continues at prices above the Q2 average.
2. Restoration conversion pace. The rebuilt pipeline (“historically healthy”) has a disclosed 12–18-month conversion lag; win sizes are undisclosed. Q3/Q4 prints test the ~5% H2 guide; the 2027 prints test the thesis. A quiet August–October peak season confirms the deferral; a landfalling major hurricane compresses it.
3. The Q4’26 roofing goodwill test. Organic decline continued (−10% Q2, guided down mid-single in Q3) against a <5% cushion. Does the annual test force the first impairment in company history — and does management pair it with disclosed assumption changes (0.5× multiple / 3% EBITDA sensitivities)?
4. AR-facility utilization. $35.3M sold in Q2 against $300M uncommitted capacity. Does usage scale — and with it, the gap between headline and underlying OCF and leverage?
5. H2 interest step-up. H1’s ~$7.6M interest tailwind reverses as the buyback-funded draw annualizes (~$10M/yr headwind). Does H2 Adjusted EPS guidance absorb it cleanly?
6. Orbis and the Canadian holder base. Orbis is visible only because it files on EDGAR; other >5% Canadian institutions report on SEDI/early-warning and are invisible here. Is there broader institutional accumulation behind the company’s bid?
7. PE in HOA management. Roland Berger’s July 2026 paper and ~15 PE firms already invested: does a scaled PE platform emerge in community management (à la Pye-Barker in fire), pressuring FSR’s future tuck-under pricing?
8. APi organic deceleration. If Safety Services organic keeps sliding toward mid-single digits, does the ~20×+ multiple support for scaled fire/life-safety — and thus the SOTP premium logic for Residential-type recurring revenue — hold?

14. What Must Be True

Bull thesis — what must be true (scored at 2026-08-15)

• Restoration revenue re-accelerates by 2027, share-led or storm-led. Status: TRACKING but the timeline is being consumed — 2026 is quieter than 2025; the snapback is now a 2027 event; one of the three years June allowed is spent; the August–October season is pending. The restated test: Brands-restoration organic ≥ mid-single digits for two consecutive quarters, storms optional.
• Residential keeps compounding at 4–5%+ organic with stable-to-rising margins. Status: HIT — two consecutive confirming quarters (+4%, +5%), margin +30bp H1, Resilience First launched.
• Roofing stabilizes (no value-destroying impairment cascade) as non-residential construction recovers. Status: MISS so far — worse than the Q1 guide; Q3 guided down; the Q4’26 test is the decision point.
• Tuck-under M&A stays disciplined and accretive. Status: TRACKING — ~$100M/yr pace, mid-teens hurdle restated, supply-constrained; earn-out reversals confirm honesty and RCA overpayment simultaneously.
• The buyback continues at scale and the market re-rates the blend once organic re-accelerates. Status: NOT YET — NCIB renewal pending (August 25); no re-rate (price flat since June).

Falsification test (bull): Restoration organic flat/negative through 2027 even as the converted backlog lands, and/or Residential organic below mid-single digits, and/or the NCIB lapsing while Brands deteriorates — would prove the trough is the trend and that even management’s bid was liquidity, not conviction.

Bear thesis — what must be true (scored at 2026-08-15)

• The organic stall is structural, not cyclical. Status: PARTIAL — the roofing/home-services leg is corroborated (six-plus quarters negative, price war, 10-year demand lows); the restoration leg is contradicted by non-weather share gains.
• A roofing goodwill impairment confirms capital was destroyed in the 2023 platform deal. Status: PENDING — Q4’26 annual test against a <5% cushion.
• M&A returns compress as the capital cycle turns against the acquirer. Status: PARTIAL/CORROBORATED for RCA specifically (earn-out reversals) — but the cyclically cooling bid now cuts in FSV’s favor.
• The market re-anchors to owner earnings (~$4.71), exposing a ~30× “real” multiple. Status: NOT HIT — 24× Adjusted EPS holds, and the company is buying at that multiple.
• The leveraged buyback backfires — leverage drifts toward/past 2.5× into a third trough year, converting the margin-of-safety spend into a solvency-adjacent problem. Status: UNTESTED — new this report.

Falsification test (bear): Two consecutive quarters of Brands/restoration organic re-acceleration (share-led counts) plus continued Residential margin expansion plus a clean Q4’26 goodwill test plus the NCIB renewed and executed without leverage breaching ~2.5× — would prove the model intact and break the bear case.

15. Source Appendix

A full source appendix is included as Appendix B below. Primary sources relied upon include: FirstService’s Form 40-F for FY2025 (filed 2026-02-20, accession 0001171843-26-000985) with AIF, MD&A, and audited US-GAAP financial statements; the Q2 2026 earnings press release (6-K Ex-99.1, filed 2026-07-23, accession 0001171843-26-004850) and Q2 2026 interim consolidated financial statements + MD&A (6-K Ex-99.1, filed 2026-07-31, accession 0001171843-26-005106); the Q1 2026 interim FS (6-K, 2026-05-01, accession 0001171843-26-002917); the 2026 management information circular (6-K Ex-99.1, 2026-03-10, accession 0001171843-26-001462); the AR Facility agreement (6-K Ex-99.1, 2026-04-13, accession 0001171843-26-002411); the NCIB amendment (6-K, 2026-06-02); Orbis 13G/13G/A (accessions 0000940594-26-000028 / -000046); Form 144 (accession 0001959173-26-000878); the Q1 and Q2 2026 earnings-call transcripts (ROIC.ai, read in full); market/factor data (AZI valuation index and price CSV; FactorsToday factor model; ROIC.ai fundamentals); and industry sources (NOAA, CSU, Swiss Re Institute, Munich Re, Aon, CAI, NAR, Conference Board, AIA/Deltek, Dodge, Roland Berger, company/competitor releases for APi, Pye-Barker, Servpro, BELFOR, Rytech). Full citations, URLs, and access dates are in Appendix B.

This memo expresses no investment recommendation and no price target except within the clearly-labeled “Claude’s Take” block, which is the author’s own independent, subjective opinion and is provided as general information, not investment advice. All facts are sourced; interpretations, assumptions, and open questions are labeled as such.

APPENDIX A — Standard Diligence Questionnaire

FSV — Standard Diligence Questionnaire Appendix (UPDATE, 2026-08-15)

Supplemental diligence questionnaire to the research memo above. Answers are grounded in the primary filings and sources listed in Appendix B; Fact / Interpretation / Assumption labels applied where material. Greenwald (Competition Demystified) and Marathon (Capital Returns) lenses applied where they add insight. Baseline for this update: the June 7, 2026 report.

General

What thoughtful questions have other investors asked about this company?

• Is the two-year organic stall (FY2025 ≈ 0%, H1 2026 ≈ 0% consolidated ex-Residential) a weather/cyclical trough or a structural slowdown? (Still the defining debate — and the Q2 2026 print split the evidence: restoration is recovering share-led while roofing looks incrementally structural.)
• Was the Roofing Corp of America (December 2023) acquisition a capital-allocation mistake? The evidence firmed up since June: −10% organic in Q2’26, a declining backlog, and H1’26 earn-out reversals of −$17.2M — with the reporting unit’s goodwill ($363.4M) one Q4’26 test away from a write-down on a <5% cushion.
• Which earnings number should be capitalized — GAAP ($3.53 TTM), Adjusted ($5.81), or owner earnings (~$4.71)? The Q2’26 GAAP-to-Adjusted gap was 43% of Adjusted EPS, and the recurring add-backs (RNCI increment + SBC ~$0.54/sh per half) are rising.
• Is the $248M debt-funded buyback conviction or financial engineering? New question this period: zero measured EPS accretion at current rates, leverage moved ~1.5× → ~1.9× with stated comfort to ~2.5×, and the CEO sold ~$10M of stock at ~$214 four months before the company bought at ~$136.
• Is tuck-under M&A still accretive as PE crowds the space — and does the cyclically cooling 2026 bid environment (fewer bidders, stuck sellers) restore FSV’s buying advantage in 2027?
• How durable is the FirstService Residential moat against Associa/RealManage — and now against ~15 PE firms eyeing HOA management (Roland Berger, July 2026)?
• Will the NCIB be renewed at/after its August 25, 2026 expiry — i.e., was the Q2 buying a one-quarter event or a regime?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Interpretation: Cyclically depressed, and the trough deepened/widened since June. Brands earnings sit at a two-year trough: restoration named-storm revenue remains far below its >10%-of-restoration-revenue historical average (management’s FY25 disclosure: <2%), H1 2026 global insured catastrophe losses of $42–47B (Swiss Re / Munich Re / Aon) are the lowest H1 since 2020, and roofing organic fell to −10% in Q2’26. Residential earnings are at a normal-to-improving level (+5% organic, margin +30bp H1). Blended, FSV earnings remain below mid-cycle.

Driven by the external environment or internal actions? Both, with a composition shift since June. External: weather/catastrophe activity (restoration), non-residential construction and rates (roofing), housing turnover ~4.0–4.1M SAAR and consumer confidence 90.8 (discretionary home services). Internal and positive: Residential contract wins and operating efficiencies (margin +30bp H1), Century Fire share gains (backlog “well up over prior year,” only ~15% data-center), home-services close-ratio/job-size gains. Internal and negative: nothing new identified — the buyback is an internal financial action, not an operating one.

How stable are revenues? Bifurcated. Residential (~42% of TTM revenue) is recurring, contracted, mid-90s% retention — very stable, and Q2’26 confirmed it again. Brands (~58%) blends recurring (fire inspection, franchise royalties) with event/cycle-driven lines — moderately volatile, and 2025–26 demonstrated the amplitude (Brands organic −3% FY25, −1% H1’26, −3% Q2’26).

Outlook for products/services? Fact (management commentary, treated as hypothesis): FY2026 revenue growth “similar to or modestly better than” H1’s +4%; mid-single-digit EBITDA growth; Q3 low-single-digit; Brands H2 growth skewed to Q4 on restoration backlog conversion. The long-term algorithm (mid-single organic + tuck-unders = 10%+) is unchanged on paper but running at half speed for a second year. Interpretation: the Q1→Q2 roofing guide slip (guided ~flat, printed −10%) means segment-level guides have been optimistic this year.

How big is this market — growing, shrinking, domestic or international? Large and growing, North America-focused. Community associations: ~373,000 end-2025 → ~377,000 projected 2026 (CAI), ~one-third of U.S. housing. Restoration ~$40B+ growing mid-single digits. Roofing and fire protection large and fragmented. Minimal international exposure (25 U.S. states + 3 Canadian provinces; USD reporting).

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Mixed, with movement since June. Residential: competitively benign today, but PE interest in HOA management is now explicit (Roland Berger, July 2026; ~15 PE firms invested) — a future tuck-under-cost watch item. Brands: structurally more competitive (PE capital “increases every year… could be defined as a structural change” — Patterson) but cyclically cooling (fewer bidders, funds pulling back, sellers unwilling to crystallize losses, SW Florida roofing capacity exiting). Marathon lens: the bust phase of the 2022–24 capital flood has begun in roofing; that is what eventually restores returns for disciplined survivors.

How profitable is the business (ROIC, ROE)? Fact (ROIC.ai, GAAP basis, FY2025): ROE 11.3%, ROIC 7.8%, effective tax 28.4%. Interpretation: consolidated ROIC ~8% modestly clears a ~7–8% WACC on a goodwill-heavy base ($2,229.7M goodwill + intangibles vs $1,218.1M equity at 6/30/26); cash-ROIC (adding back acquired-intangible amortization) is meaningfully higher (~12% per the June computation, direction unchanged); return on incremental/tangible capital is far higher (asset-light operating units, capex ~2.2% of revenue). Good, not great — unchanged from June. Greenwald lens: genuine moats should show up as high returns on incremental capital in the moated segment; Residential’s margin expansion (+30bp H1’26 in a weak macro) is exactly that signature, while Brands’ blended returns show what moat-absence plus acquisition goodwill does to reported ROIC.

How profitable is the industry — how many competitors, what barriers to entry? Residential: healthy for scaled players; real local barriers (density, compliance expertise, switching costs). Restoration: fragmented below a consolidating top (BELFOR #1, First Onsite #2, Servpro 2,390+ franchises under Blackstone, Rytech now in Summit Partners’ Fortify platform); TPA/insurer-controlled pricing caps economics. Roofing: no barriers; 56 PE-backed platforms by late 2024; price war in overbuilt regions. Fire: code-mandated demand but a scale race (APi at ~20×+ EV/EBITDA; Pye-Barker 200+ deals). Interpretation: barriers exist at the branch/metro level in Residential and fire service; nowhere structural in roofing or restoration.

Can the business be easily understood? Yes at the model level (outsourced property services, two segments). The complexity lives in (a) RNCI partnership accounting (a $507.4M balance with a $432.4M formula redemption amount), (b) the GAAP-to-Adjusted bridge, and © — new in 2026 — a derecognized receivables-sale facility running through operating cash flow.

Can it be undermined by foreign low-cost labour? No. All lines are inherently local, on-site, and regulation/relationship-bound. The live labour risk is domestic wage inflation and availability (including immigration-policy effects), flagged in the MD&A.

Do brands matter? Yes, moderately. FirstService Residential’s brand supports board trust and retention (mid-90s%); First Onsite’s #2 position carries reputational value in national-account programs; California Closets/CertaPro/Paul Davis carry consumer brand value. Brand contributes to the moat; it is not the moat. Roofing’s experience — a roll-up of competent regional brands losing share to price — is the counter-proof.

What is the nature of competition? Residential: contract bids to HOA/condo boards on service, price, local density, ancillary breadth. Restoration: insurer/TPA program relationships plus large-loss mobilization capability. Roofing: bid/construction work, increasingly price-led. Fire: code-driven service density. Franchise: consumer marketing and franchisee economics.

Customers’ switching costs? High in Residential (financial-system migration, resident portals, board inertia, fiduciary re-bid risk, embedded ancillary banking/insurance/energy — deepened further by the July 2026 Resilience First cross-sell). Lower in Brands (project-based in restoration/roofing; recurring-contract stickiness in fire inspection; franchisee lock-in in the franchise systems).

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Interpretation: The Residential customer-contract franchise (mid-90s% retention, recurring, margin-rising) is worth more than its carrying intangibles; internally generated brand value is understated. Offsetting: $2,229.7M of goodwill/intangibles (vs $1,218.1M equity) is acquisition-derived, and the roofing unit’s $363.4M goodwill carries a <5% impairment cushion. Tangible book is deeply negative — the balance sheet understates the good asset and overstates none of the bad one.

Off-balance-sheet liabilities? Fact: operating leases are on-balance-sheet under ASC 842 (and the ROIC-style EV of $8,111M includes them); contingent acquisition consideration $33.7M at 6/30/26 (down from $47.0M, fair-valued on the balance sheet, undiscounted range $29.8–39.7M, expiries to May 2028); RNCI put obligations are recorded ($507.4M carrying). Insurance self-retention (high deductibles) is a real contingent exposure. The AR sale facility ($35.3M derecognized in Q2, $300M uncommitted capacity, FirstService as guarantor) is the item to watch — derecognition removes receivables from the balance sheet while the guarantee keeps a residual link. No material undisclosed OBS liabilities identified.

How conservative is the accounting? Mixed — and slightly less conservative in presentation than at the June baseline. Positive: audited US GAAP (PwC); earn-outs honestly booked and reversed when missed (−$17.2M H1’26); no restructuring charges or impairments recorded anywhere in the five-year stream. Negative: the headline metric (Adjusted EPS) adds back real recurring costs (SBC, RNCI redemption increment — now ~$0.54/sh per half combined and rising); and the new AR facility’s stated purpose — “reduce interest costs and reported financial leverage” — is an optics admission: headline H1 OCF (+7%) reverses to ~−10% underlying ex the $35M sale. Interpretation: management presents profitability and leverage at their most generous. Capitalize owner earnings (~$4.71 TTM), not Adjusted EPS ($5.81).

How CapEx-hungry is the business? Low. H1’26 capex $59.6M; FY2026 guide ~$130M (cut from $140M); ~2.2% of TTM revenue. The capital sink is M&A (and now buybacks), not capex.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? Fact: TTM OCF $459.8M, capex $124.4M → FCF ≈ $335M headline, ~$300M ex the AR-sale benefit. H1’26 uses: buybacks $248.4M + dividends $26.6M + M&A $48.4M + RNCI buy-ins/distributions $23.8M — outspending headline FCF by roughly $190M, debt-funded, deliberately. Interpretation: the philosophy changed this year — from “reinvest in tuck-unders, grow the dividend, ignore the buyback” to “buy whichever is cheaper: private businesses or our own stock,” with a stated mid-teens return hurdle on either and leverage comfort to ~2.5×. Marathon lens: this is what a disciplined capital-cycle actor does when the public-private arbitrage inverts — buy the cheap asset (its own equity) and wait for the deal market to clear.

Significant acquisitions recently? H1 2026: 5 tuck-unders, all Brands, ~$55M total (Paul Davis Cleveland; California Closets Indianapolis; Scheffer’s Roofing, Kansas City; Titan Fire, Tampa; GSC Fire, Austin/San Antonio). No platform deal since RCA (December 2023, ~$413M). FY2026 spend guided “similar to last year” (~$107M). Interpretation: supply-constrained discipline — PE sellers won’t crystallize losses; FSV refuses to overpay; earn-out reversals confirm the RCA platform specifically underperforms its price.

Buying back shares? Yes — the defining change of 2026. Q2 2026: 1,827,750 shares for $248.4M at $135.91 average (~4% of the share base, all cancelled), after zero in 2025 and zero in Q1 2026. NCIB amended June 2, 2026 to the TSX maximum (4,118,199 shares = 10% of public float) plus an ASPP; 2,290,449 shares remained at June 30; the bid expires August 25, 2026 with no renewal announced as of this report. Share count: 45.72M (YE25) → 44.15M (6/30/26); the ~1%/yr SBC dilution trend reversed (−3.4% net H1).

Issuing large amounts of new shares to insiders? LTI is ~100% stock options (~68% of CEO pay) — the primary dilution channel — and the April 1, 2026 special meeting increased the option-plan reserve by 2,000,000 shares (to 9,313,500, ~20.4% of shares outstanding). Interpretation: the buyback retired 1.83M shares while shareholders approved +2.0M of new option headroom — net anti-dilution progress is real but partly mortgaged forward. 2025 NEO option exercises were large (Patterson 150,000 options, $7.71M notional gain).

Compensation policy of directors/management? Fact (2026 circular, FY2025 pay): CEO Patterson total $7,801,303 (salary ~$860K, options ~$5.34M, bonus $1,603,100 on 11% 3-yr AEPS growth / 5% organic); CFO Rakusin $4,638,887; bonus = 3-yr trailing Adjusted-EPS growth × organic-revenue modifier (80/100/120%), CEO cap 5× salary; say-on-pay 92.4%; no ROIC/TSR/capital-efficiency metric; the circular’s own table shows 5-yr TSR of 26.1% vs the S&P/TSX Composite’s +110.8%. Hennick is paid as an owner (modest director fees; ~6% stake; excluded from the option plan). Interpretation: per-share, multi-year, anti-empire-building metrics — a genuine positive — but the metric (Adjusted EPS) adds back the SBC that funds the options, and capital efficiency is not incentivized.

Motivations of management? Interpretation: long-term, owner-aligned compounders — the RNCI partnership model, EPS-gated multi-year incentives, a decade-plus of ≥10% dividend raises, and now a willingness to lever the balance sheet to buy the stock at a claimed discount to private-market value. The tension to hold alongside that read: the CEO’s February 2026 sale of ~$10M at ~$214 (SEDI-reported, likely exercise-related, no plan disclosure available) against the company’s Q2 buying at ~$136. Facts on both sides; motive not resolvable from disclosure.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? None. FSV is a Canadian foreign private issuer with ordinary common shares dual-listed on NASDAQ and TSX, reporting in USD under US GAAP, filing 40-F/6-K (no 10-K/10-Q; no Forms 3/4/5 — the Section 16 exemption is intact, correcting the June report). No K-1. U.S. holders may face Canadian withholding on dividends.

Dividend policy? Quarterly, $0.305/share ($1.22/yr) after the +11% February 2026 raise — more than a decade of ≥10% annual increases; ~0.87% yield at ~$141; conservative payout.

How profitable is the business? TTM Adjusted EBITDA $569.8M (10.2% margin); GAAP net margin ~3%; ROE 11.3%, ROIC 7.8% (FY25, ROIC.ai GAAP basis); interest coverage ~9.9× EBITDA/interest; Altman Z 5.0; current ratio 1.67 (Q2’26). Profitable and cash-generative, with the higher-quality earnings in Residential.

Is net income diverging from cash from operations? Fact: TTM OCF ($459.8M) comfortably exceeds net earnings — D&A ($185.2M FY25, largely acquired-intangible amortization) drives the favourable gap. New caveat: the divergence narrowed in quality terms in Q2’26 — headline H1 OCF (+7% YoY) includes ~$35M of one-time derecognized AR-sale proceeds; ex-facility, H1 OCF fell ~10% YoY. Watch quarterly utilization of the $300M facility. No adverse accounting divergence identified; the presentation divergence is new.

Risks & Downside

What factors would cause the stock to decline? A third benign weather year (2027) stranding the restoration-recovery thesis; a roofing goodwill impairment at the Q4’26 test; roofing/home-services organic decline proving structural; the NCIB lapsing August 25 with no renewal (removing the debt-financed floor); leverage drifting toward/past 2.5× into a still-troughing Brands; a market re-anchor from Adjusted EPS (24×) to owner earnings (~30×); higher-for-longer rates raising the buyback’s carry cost (accretion flips dilutive at ~+100bp); a recession hitting discretionary home services.

Risk of a catastrophic loss? Interpretation: Low, but the margin thinned. Diversified, recurring, asset-light, no accounting/fraud/solvency red flags; coverage ~9.9× and Altman Z 5.0. The June backstop — a fortress balance sheet — is partially spent: leverage is 1.9× by design, and the same buyback that assists the base case consumed the slack that would absorb the bear case. Realistic downside remains multiple compression + an earnings air-pocket + a non-cash roofing write-down, slightly amplified by financial leverage.

Chance of a total loss? Negligible. No solvency, fraud, single-customer, or single-asset dependency; the Residential franchise alone is durable and cash-generative; the revolver matures 2030 with $646.1M undrawn and liquidity >$800M.

Recent News & Events

Has the business environment changed recently? Yes. Since the June 7 baseline: (i) Q2 2026 printed July 23 (EPS beat, revenue miss, roofing −10% organic; stock −7.4% on the day, fully recovered in three sessions); (ii) the 2026 hurricane season is tracking below normal (NOAA August 6; H1 insured cat losses lowest since 2020) — the snapback deferred to 2027; (iii) the capital cycle in roofing/restoration cooled cyclically (fewer bidders; SW Florida capacity exiting); (iv) Florida’s HB 913 (effective July 2025) diffused part of the condo-reserve cost shock while leaving the SB 4-D compliance core intact; (v) CIBC cut its target $204→$175 (July 24) with the Buy consensus intact (mean PT ~$174).

Significant acquisitions? Five tuck-unders in H1 2026 (~$55M, all Brands): Paul Davis Cleveland, California Closets Indianapolis, Scheffer’s Roofing (Kansas City, RCA’s 16th deal, 27 branches), Titan Fire (Tampa), GSC Fire & Security (Austin/San Antonio). No platform deal.

Change in accounting policies? None material. New in 2026: the April 2026 uncommitted AR-sale facility (CIBC; $300M capacity; $35.3M sold and derecognized in Q2) — a presentation-relevant change in cash-flow/leverage optics rather than an accounting-policy change.

Recent changes — new markets, facilities, management? Resilience First cross-sell launched (July 6, 2026); non-core residential aquatic operations divested (start of Q2’26); RCA consolidating 14 operating systems into one; specialty-construction adjacency opened out of healthcare restoration; NCIB maxed to 10% of float with an ASPP (June 2); option-plan reserve +2.0M shares (April 1); Orbis disclosed an 8.8% passive stake (13G/A August 14); leadership stable (Patterson/Rakusin/Hennick). Insider-transaction reporting remains SEDI-only — the June report’s claim that insiders became Section 16-subject in March 2026 is corrected: zero Forms 3/4/5 exist on EDGAR for FSV, ever.

APPENDIX B — Source Appendix

FSV — Source Appendix (UPDATE run 2026-08-15)

Sources relied upon for the 2026-08-15 memo and diligence appendix. Primary sources first; third-party/convenience data labeled. All web/market data accessed 2026-08-14/15 unless noted.

A. Primary company filings (SEC EDGAR, CIK 0001637810; issuer is a Canadian FPI — Form 40-F + 6-K; SEDAR+ holds the Canadian equivalents)

B. Earnings-call transcripts (ROIC.ai transcript feed = source of record; management commentary treated as hypothesis, validated against filings)

C. Market / quantitative data feeds (third-party; reconciled to filings — not primary)

D. Industry & competitor sources (all URLs accessed 2026-08-15 unless noted)

Weather / catastrophe data:

Restoration / roofing / fire protection:

Residential property management / housing / consumer:

E. Company releases & news articles (publisher, date; ROIC.ai company-news feed 2026-05-01 → 2026-08-15 used as the triage layer, material items validated at source)

F. Analytical frameworks

Data-discrepancy notes: (1) a $290.34 “52-week high” figure surfaced in one aggregator quote (Robinhood/Macrotrends) is erroneous — the AZI adjusted series and FactorsToday both print $208.07 (2025-09-11); (2) AZI’s P/B percentile uses a stale-hybrid BVPS (YE25 equity ÷ post-buyback share count) — percentiles retained, levels recomputed from filings; (3) FactorsToday’s All-Factors model is dated 2026-07-31 vs the Base model’s 2026-08-14 (weekly recalibration lag).