VEL — Velocity Financial, Inc. · 2026-09-10 · Verdict: PASS · Conviction 3

Price $18.12 (screen row, 2026-09-09 run; no live price) · Mkt cap $715M · Financial lane, so no EV/EBIT: P/E 6.6x GAAP LTM and 24.8x normalised · P/TBV 1.00x · Diluted BVPS $18.43 · ROTE 16.3% GAAP, 4.0% normalised · Corporate recourse debt $575M face · ADV $1.9M Sources read: 10-K filed 2026-03-12 (Items 1, 1A, 7), 10-Q filed 2026-08-06 (Q2 2026), DEF 14A filed 2026-04-10, 8-Ks of 2026-05-06, 2026-08-05 and 2026-08-27, Q2 2026 prepared remarks and press release (2026-08-05), Form 4s (12m), screen row in universe_v2.csv.

Desk stats - Revenue trend: interest income +35.4% in FY2025 to $550.8M and +18.8% in Q2 2026 to $161.0M; the company's own rate/volume tables attribute $121.5M of the FY2025 increase to volume and only $22.5M to rate, and in Q2 2026 volume added $31.6M while rate subtracted $6.2M (10-K Item 7; 10-Q Q2 2026). Portfolio UPB is $7.0B, +19.2%, but loan production fell to $672.6M from $725.4M (8-K 2026-08-05). - Normalised after-tax profit: LTM GAAP pre-tax income $151.5M (10-Q quarterly table). Subtract $97.6M of non-cash fair-value marks (FV loans +$77.6M, FV securitized debt +$19.4M, MSRs +$0.6M), $19.3M of gain on the December 2025 sale of $128.9M of nonperforming loans to BPC MC Trust, a Beach Point Capital affiliate and 17.7% shareholder (DEF 14A, Related-Party Transactions; 10-K Item 7), and a $2.4M employee retention credit (10-Q). Add back $1.3M of unamortised issuance costs written off on the January 2026 term-loan payoff, $6.0M of protective advances written off into the 2026-MC2 trust, and $0.7M of M&A diligence (10-Q). Normalised pre-tax $40.2M; at the 28.6% LTM effective rate, normalised after-tax profit $28.9M ($30.2M at the desk's 25% convention) against GAAP net income attributable of $107.7M. - EV / normalised after-tax profit: EV is meaningless here (it would sweep in $6.18B of non-recourse securitized debt). Financial-lane equivalent: market cap $715M / $28.9M = 24.8x, versus 6.6x on GAAP LTM earnings. - Financial-lane metrics: P/E 6.6x GAAP, 24.8x normalised; P/TBV 1.00x ($715M vs $714.8M tangible equity; $18.12 vs $18.43 diluted BVPS is 0.98x); ROTE 16.3% GAAP, 4.0% normalised; tangible BVPS growth +18.0% ($15.62 to $18.43). Assets/equity 11.0x; equity/assets 9.1%. - Is the growth sustainable? Organic in units but not yet in earnings: portfolio UPB grew 19.2% while quarterly net interest income rose just $1.3M (+2.7%) because corporate interest tripled to $14.5M; the next leg is bought, via a cash purchase of Toorak's platform closing Q4 2026 at guided 4-6% book dilution and accretive only in 2027. - What the screen got wrong: the 6.8x P/E, 14.7% ROTE and 17.5% book growth all rest on FY2025 net income of $105.0M, of which $86.4M was non-cash valuation gain by the company's own reckoning and $19.3M a related-party sale. The screen also carries ltd = 0 and net cash of $76.1M (its own debt_note flags this), missing that corporate recourse debt went from $290M to $575M in January 2026 at 9.375%.

1. What the business actually does

Velocity originates business-purpose loans secured by 1-4 unit residential rental and small commercial property, through roughly 1,777 active independent mortgage brokers, then finances them with warehouse lines and securitizes them into REMIC trusts consolidated as secured borrowings (10-K Item 1). It keeps the loans and earns the spread rather than selling them. The average loan is $390 thousand at a 65.2% origination LTV with a 100% personal guarantee and a 700 weighted-average credit score; 91.4% fully amortise over 30 years. It has done 46 securitizations and issued $10.6B of securities since 2011. A small consolidated subsidiary, Century, originates and services FHA/Ginnie Mae multifamily and healthcare loans. Portfolio net interest margin was 3.61% in 2025 (10-K Item 1).

2. Why it is mispriced — the edge case

There is none, which caps the verdict. Three private funds own 83.9% of the shares: Snow Phipps 34.0%, PIMCO 32.2%, Beach Point 17.7% (DEF 14A). Snow Phipps and PIMCO hold board designation rights under a stockholders' agreement and registration rights that let them force a secondary (10-K Item 1A). That is a supply overhang on an 11-15% float trading $1.9M a day, not a forced seller creating a bargain. Nothing dislocated the price: the stock is 12.7% off its 52-week high after a flat six months, and the last twelve months of Form 4s show zero open-market purchases and $286,394 of sales across six officers, including the CFO trimming at $18.00-$18.16 in July and August.

3. Credit quality, margin and funding (financial lane)

Nonperforming loans are $673.3M, or 9.6% of held-for-investment loans, versus $601.8M and 10.3% a year ago and $394.6M in 2023 (10-Q; 10-K Item 7). The company's defence is that NPLs resolve at a profit, and it is largely true: Q2 2026 resolutions of $90.5M UPB recovered 107.7%. But the recovery rate has slipped from 109.7% a year ago and sits below the five-quarter average of 108.5%; charge-offs rose to 0.27% of UPB in 2025 from 0.07%; REO went from 129 properties worth $68.0M at end-2024 to 254 worth $118.3M at end-2025 to $142.1M by June 2026, with REO expense doubling to $6.7M a quarter. Office is the fastest-growing collateral, +43.4% to $658.2M (8-K 2026-08-05). Allowance is 0.28% of amortised-cost UPB.

Portfolio NIM held at 3.66%, but total-company NIM fell from 3.39% to 2.82% because the January 2026 refinancing replaced a $215M term loan at 7.125% with $500M of unsecured notes at 9.375% on top of a $75M loan at 9.875%. Corporate interest went from $6.1M to $14.5M a quarter. The whole point of raising the money was to fund portfolio growth, and so far the incremental spread has not covered the incremental cost.

The decisive number is the company's own: net cash provided by operating activities was $18.2M in 2025, $37.8M in 2024 and $48.8M in 2023, falling every year while net income rose from $52.3M to $105.0M. The 10-K explains the 2025 figure as "$105.0 million in net income ... offset by $86.4 million of net valuation gain" (Item 7, Liquidity). The marks are not invented — under fair-value option elected on all originations since 1 October 2022, they pull forward the value created at origination — but they are non-cash, they require origination growth to keep recurring, and they amortise back against the in-place book's future net interest income.

4. Balance sheet and capital allocation

Total assets $7.96B against $725.2M of equity, financed by $6.18B of securitized debt, $311.7M of warehouse and $559.6M of carrying-value corporate debt. Unrestricted cash is $76.1M; total liquidity $240.0M. Warehouse covenants require minimum liquidity, minimum net worth, a maximum debt-to-net-worth ratio and a minimum EBITDA-to-interest ratio; the company was compliant at June 30, 2026 (10-Q). No dividend, no buyback; share count rose 2.7% on warrant exercises by Snow Phipps and PIMCO affiliates. Executive PSUs vest on three-year average Core Net Income growth of 15-20% (DEF 14A), and Core Net Income adds back stock compensation and M&A costs while leaving every fair-value mark in.

5. Management: what they said versus what they did

Farrar has run this for 22 years and the underwriting record supports him. But two related-party patterns deserve naming. The largest discrete income item of 2025 came from selling NPLs to a 17.7% shareholder's affiliate. And in the Toorak deal, the "third-party investment firm" buying Toorak's $3B portfolio, which Velocity will then manage for fees and sell future production to, is identified in the 8-K's own legal text as affiliates of "TOBI III SPE I LLC ('TOBI')" and a Master Transaction Agreement with Titan RR LLC, "an affiliate of an affiliate of TOBI" (8-K 2026-08-27, Item 1.01) — while the 10-K uses "TOBI" as its defined term for the PIMCO funds that own 32.4% of Velocity and designate two directors (Item 1A). The press release calls it only a third-party investment firm and discloses no affiliation (Ex. 99.1). A reader should resolve that before underwriting the capital-light fee stream. Separately, the deck headlines a "$63 million" purchase price while the agreement says "approximately $62 million, plus estimated tangible book value," so cash out is not yet knowable. No Toorak revenue or profit figure is disclosed anywhere in the filing.

6. Valuation

Base (55%): the $500M is deployed and Toorak adds fee income, lifting normalised after-tax profit toward $55-60M by 2028, a sustainable ROTE near 9%; against a 13% cost of equity that is ~0.75x a post-dilution book of $17.51, or $13. Bear (20%): recovery rates break below 100% as REO and office compound, marks reverse, and a levered lender at 11x assets/equity trades at 0.5x book, or $9. Bull (25%): the marks convert to cash, Toorak makes the model genuinely capital-light, mid-teens ROE holds, book compounds to ~$21 by end-2027 and rerates to 1.2x, or $25. Probability-weighted $15.20, about 16% below price. Reverse DCF: at $18.12 against $18.17 of tangible book per share, the price implies a sustainable ROTE of roughly 13%, essentially the GAAP-with-marks figure, against the 4% the business produced on a cash basis over the last twelve months.

7. Catalysts and timeline

Q3 2026 results in early November, the first quarter carrying the full $14.5M corporate interest run rate. The Toorak close, guided to Q4 2026, with a 120-day outside date of 24 December 2026, and the pro forma financials that must follow it. The first 2027 guidance, which is when management says accretion arrives.

8. Risks and pre-registered kill criteria

1. Cash conversion. FY2026 net cash provided by operating activities above $60M, versus $18.2M in 2025, would mean the marks are converting and this PASS is wrong.
2. Credit. Total NPL recovery rate below 100% in any quarter, or REO carrying value above $175M, confirms the bear case; REO below $110M with recovery above 107% for two consecutive quarters retires it.
3. Deal disclosure. If Velocity discloses Toorak standalone revenue and pre-tax income implying a purchase multiple under 8x pre-tax earnings, upgrade to WATCH.
4. Insiders. Any open-market purchase (code P) by Farrar or the CFO after twelve months of zero buys and $286K of sales.

9. Verdict and one-paragraph summary

PASS, conviction 3. Velocity looks like a 6.8x-earnings, 1.0x-book lender compounding tangible book 18% a year, but the earnings are mostly not cash: of $151.5M of last-twelve-month pre-tax income, $97.6M was non-cash fair-value marks and $19.3M was a gain on selling nonperforming loans to a 17.7% shareholder's affiliate, and the company's own cash flow statement shows operating cash flow falling three years running to $18.2M in 2025 while reported net income doubled to $105.0M. Normalise for that and the business earned about $28.9M after tax, or 25x the market value and a 4% return on tangible equity, right as the January 2026 refinancing tripled the corporate interest bill to $14.5M a quarter and held net interest income growth to 2.7% on a portfolio that grew 19.2%. Meanwhile nonperforming loans are 9.6% of the book, REO has more than doubled in eighteen months to $142.1M with office collateral the fastest-growing line, and management has committed balance-sheet cash to buy KKR's Toorak platform with no target financials disclosed, 4-6% book dilution, no accretion before 2027, and a fee stream that depends on an entity the 8-K's own defined term ties back to Velocity's 32% shareholder. There is no edge case to underwrite: three funds own 84% with registration rights and an exit problem, the float is thin, and insiders sold $286K and bought nothing in twelve months. At 0.98x book you are paying fair value for reported earnings that the cash flow statement does not yet support.

Research for discussion, not investment advice. Positions and sizing are the reader's decision.