ONIT — Onity Group Inc. · 2026-09-10 · Verdict: WATCH · Conviction 3
Price $34.86 (screen close, 2026-09-09) · Mkt cap $291M · Tangible book $72.33/share · P/TBV 0.48x · P/normalised earnings to common 8.4x · Corporate net debt $497M (Senior Notes 2029 less unrestricted cash) · ADV $2.3M Sources read: 10-K 2026-02-17 (Items 1, 1A, 7), 10-Q 2026-08-06 (MD&A). No DEF 14A, no 8-Ks, no Form 4s and no transcript are in the bundle for this ticker, so ownership, insider activity and compensation alignment are unverified here and section 4 says so.
Desk stats - Revenue trend. FY2025 revenue $1,066.7M vs $976.0M, up 9.3%: Servicing up 5% and Originations up 44% (10-K 2025, Item 7). Q2 2026 revenue $282.9M vs $246.6M a year earlier, up 14.7%, but down 4% from Q1's $294.3M; H1 2026 $577.2M vs $496.4M, up 16.3% (10-Q Q2 2026). The driver is volume, not price: funded forward loan production rose 50% in H1 and average serviced UPB rose 10% to $337.6B, while the Consumer Direct gain-on-sale margin fell from 2.92% to 2.01% and Q2 lock volume fell 30% sequentially (10-Q Q2 2026, Originations). - Normalised after-tax profit. GAAP FY2025 net income was $189.5M on pre-tax income of only $62.7M, because income tax was a $126.8M benefit (10-K 2025, Item 7). Trailing twelve months to 30 June 2026, pre-tax income was $23.6M (Q3'25 $23.1M, Q4'25 $7.7M, Q1'26 $7.9M, Q2'26 minus $15.1M) against GAAP net income of $141.6M. Adjustments: add back $37.4M of MSR fair-value change net of hedging, negative in each of the last four quarters (10-Q Q2 2026, quarterly table); add back $3.7M of Rithm and FAR severance (10-Q Q2 2026, Compensation and benefits); then charge a normal hedge-slippage cost of $12.7M a year, the six-quarter average of that same line, rather than assume perfect hedging. Normalised pre-tax $52.0M; at 25% tax, normalised after-tax $39.0M, or $34.8M to common after $4.2M of Series B preferred dividends, against GAAP TTM net income of $141.6M. - EV / normalised after-tax profit. All but one of Onity's borrowings are asset-backed and their interest is already charged above the line, so the only corporate claim is the $693.2M of 9.875% Senior Notes due 2029 net of $196.6M unrestricted cash (10-Q Q2 2026). EV $787.8M; adding back after-tax senior-note interest of about $51.8M gives EV / normalised after-tax profit of 8.7x, equivalent to price / normalised after-tax profit to common of 8.4x, not the 1.5x P/E on the screen. - Financial-lane equivalents (no EV/EBIT). P/E 2.1x on GAAP TTM, 8.4x normalised. Price to tangible book 0.48x ($72.33 a share at 30 June 2026). Return on tangible equity 22.9% GAAP TTM, 6.4% normalised. Tangible book per share went from $73.69 to $72.55 in six months, down 1.5%, after rising about 34% in 2025 of which the $120.1M tax release was roughly $14.40 a share. Stockholders' equity is 4.9% of assets, which the company itself calls "a high leverage ratio" (10-Q Q2 2026, Capital Adequacy). Corporate debt is 1.14x common equity; ratings are B-/B3/B- with the notes at Caa1 (Moody's). - Growth sustainable? Volume growth is organic and real but it is bought, not cash-backed: H1 2026 operating cash flow was negative $2,106M, MSR additions were $598M, and the funding came from a $1.2B increase in warehouse capacity, three OLIT securitizations and a $200M add-on of 9.875% notes (10-Q Q2 2026). Management gives no numeric guidance, only that it expects "to continue to generate net income and increase our equity in 2026 and 2027." - What the screen got wrong. Almost everything, all in the seller's favour. (1) The $189.5M of 2025 net income is $126.8M of tax benefit, of which $120.1M is a non-cash release of the valuation allowance on deferred tax assets and $13.3M the resolution of a prior-year uncertain tax position (10-K 2025, Item 7); pre-tax income was $62.7M, so "1.5x P/E, ROTE 30.3%, net income growth 459%" are one accounting entry. (2) The row is CY2025; the company has since reported a $4.2M H1 2026 net loss and a $15.1M Q2 pre-tax loss, its first loss in the six quarters disclosed. (3) $123.8M of the $626M tangible book, 20%, is that new deferred tax asset (10-K 2025, Item 1A). (4) Net debt of $1,369M counts only the $1,566M tagged LineOfCredit, missing the $693M of senior notes and all warehouse, HMBS and securitization debt; total liabilities are $11.69B. (5) The desk's stale-numbers warning holds: $49.9M of Series B preferred sits in mezzanine equity, ahead of the common.
1. What the business actually does
Onity is the old Ocwen, renamed, servicing and originating residential mortgages through PHH Mortgage and Liberty Reverse Mortgage. At 30 June 2026 it serviced or subserviced 1.3 million loans with $341.4B of UPB for more than 3,900 investors and 119 subservicing clients (10-Q Q2 2026). It earns a servicing fee on MSRs it owns (about 0.29% of UPB), a smaller per-loan fee on portfolios it subservices for others, float earnings on escrow, and gain on sale in Originations, which exists mainly to replenish the servicing book as it runs off. About 75% of the 4,300 staff sit in India and the Philippines (10-K 2025, Item 1), the cost advantage the whole model rests on.
2. Why it is mispriced — the edge case
There is no edge case, and that decides the verdict. What exists is a screen artifact any reader of the filing removes in five minutes, plus a stock at 0.48x tangible book. No spin-off, no index event, no identifiable forced seller: Oaktree's overhang cleared in 2025 when it sold its shares and exercised its last warrants (10-K 2025, Item 1). The 35% fall from the 52-week high has dated causes rather than a dislocation. Rithm, the largest subservicing client at 12% of 2025 servicing fees, gave notice on 31 October 2025 and had $22.4B of UPB deboarded by June (10-Q Q2 2026), and Q2 printed a loss. The board has been buying, $10M at an average $41.19 then $2M in June at $37.71 and $2M in July at $38.82, all above today's price, but the authorization is capped at $2M a month, under 7% of the market value in a year.
3. Unit economics and credit quality
The segments have swapped roles. Originations earned $60.5M of pre-tax income in H1 2026 against $18.6M a year earlier on 50% higher funded volume, beating the industry's 26%. Servicing went from $73.5M of pre-tax income to a $12.7M loss, and Corporate absorbed $54.9M. The Servicing swing is mostly the MSR line: fair-value change net of hedging was plus $18.4M in H1 2025 and minus $32.4M in H1 2026, while runoff rose from $84.6M to $107.1M with the portfolio (10-Q Q2 2026). Underlying servicing fees are fine, up 13% on 14% higher UPB, but the fee base loses Rithm, and Q2's sequential fade in Originations (margin 1.95% from 2.05%, lock volume down 30%) says the refinance window that produced H1 is closing. The risk is not loan credit but MSR mark and advance risk: a 10% adverse discount-rate move takes $118.5M off MSR fair value, 19% of common equity (10-Q Q2 2026, sensitivity table).
4. Balance sheet and capital allocation
$693.2M of 9.875% senior notes due 2029 is the only recourse corporate debt, against $610M of common equity and $49.9M of Series B preferred callable from 15 September 2028. About $3.4B of debt comes due, amortises or requires partial repayment within twelve months, almost all warehouse and MSR facilities. Total liquidity is $230.0M, of which only $33.4M is available borrowing capacity against eligible collateral; the tightest covenants are $65.0M of consolidated liquidity and $275.0M of consolidated tangible net worth, with compliance stated at 30 June 2026 (10-Q Q2 2026). The January 2026 add-on of $200M at 9.875% costs roughly $20M a year pre-tax and partly repaid cheaper secured MSR debt, so it buys flexibility rather than earnings. Share count fell about 3.5% in seven months. Insider ownership, Form 4 activity and compensation alignment cannot be checked: the bundle holds no proxy and no Form 4s here.
5. Management: what they said versus what they did
Management flagged in the 10-K that 2026 income tax expense would "track income before income taxes at an effective tax rate moderately higher than the U.S. combined Federal and state statutory tax rate," an honest warning that the 2025 headline was not repeatable, and in the 10-Q that it expects "a reduction of our fee revenue in 2026" from Rithm. Both proved true within two quarters. Against that, Servicing headcount was cut 11% as promised, Fitch upgraded the servicer ratings in May 2025 and Moody's the second-lien assessment in June 2025 (10-K 2025, Item 1), and the FAR reverse sale closed on 30 June 2026 at book value for about $77M net cash. This team describes its business accurately, but has given no numeric earnings target anyone can hold it to.
6. Valuation
Base (40%): the Rithm cost base comes out, hedging stops leaking, normalised return on tangible equity recovers to about 9% ($56M after tax on $620M of equity) and the market pays 0.65x book, $47. Bear (35%): the post-Rithm fee hole is not filled, hedge slippage persists, 2026 finishes near breakeven and an MSR markdown or partial re-provision against the deferred tax asset takes book toward $65 at 0.40x, $26. Bull (25%): Originations holds its H1 run-rate of roughly $120M of annual pre-tax income while Servicing returns to 2025 form, normalised after-tax reaches $80M and a 13% return on tangible equity, and the market pays 0.9x book of $78, giving $70. Probability-weighted $45.40, about 30% above price. Reverse DCF: at $34.86 against $72.33 of tangible book, using a 13% cost of equity and 2% growth, the price implies a sustainable return on tangible equity of about 7.3% in perpetuity, barely above the 6.4% just earned on normalised profit and far below the 22.9% GAAP figure the screen priced.
7. Catalysts and timeline
The Q3 2026 10-Q in early November is the decisive print: the first full quarter with Rithm deboarded, the downsizing done and FAR subservicing fees replacing reverse gain-on-loans. The $20M buyback runs at $2M a month. MAV's sale lockout steps down in 25% increments through 30 September 2027, and the Series B preferred becomes callable 15 September 2028.
8. Risks and pre-registered kill criteria
- Pre-tax loss in both Q3 and Q4 2026, which would make Q2 the trend rather than hedge noise.
- Tangible book per share below $68 at any quarter end, or any re-establishment of a valuation allowance against the deferred tax asset.
- Total liquidity below $150M, or available borrowing capacity on eligible collateral below $25M (it was $33.4M at 30 June 2026), or headroom lost against the $275M consolidated tangible net worth covenant.
- Loss or non-renewal of a second major subservicing client. MAV is 11% of UPB and 8% of servicing fees, and its sale lockout expires in stages to September 2027.
9. Verdict and one-paragraph summary
WATCH, conviction 3. Onity's 1.5x P/E and 459% earnings growth are one accounting entry, a $120.1M non-cash release of the valuation allowance against deferred tax assets in Q4 2025 plus $13.3M from an old tax dispute; strip it and the company earned $62.7M pre-tax in 2025 and lost $15.1M pre-tax in Q2 2026, so the real multiple is about 8.4x normalised after-tax profit and the real return on tangible equity is about 6%, not 30%. What is genuinely interesting is the 0.48x tangible book: the price implies a sustainable 7.3% return on tangible equity forever, only a point above what the company earns today, so almost no recovery is being paid for, and management is buying stock above today's price while Originations pre-tax income tripled to $60.5M in the first half. What stops it being an idea is that there is no edge case to underwrite, no spin, no index event and no identifiable forced seller, just a servicer that lost its largest subservicing client and printed a loss, and that the margin for error is thin: equity is 4.9% of assets on the company's own description of "high leverage," 20% of tangible book is a deferred tax asset that only exists if the profits continue, available borrowing capacity against eligible collateral is $33.4M, and a 10% adverse discount-rate move would take 19% of the equity. Wait for the Q3 print in November to show whether the post-Rithm servicing hole closes, and this becomes a real idea. Research for discussion, not investment advice. Positions and sizing are the reader's decision.