Abacus buys existing life insurance policies from people who no longer want them, paying more than the insurer's cash surrender value. It keeps paying the premiums, then either collects the death benefit or, far more often, sells the policy on within about 230 days. Most of those buyers are funds Abacus itself manages: the Luxembourg Carlisle funds, four US LP funds launched in 2025, the new registered interval fund (ABXGX), and its securitisations. Abacus earns three ways: a gain on each policy (Life Solutions, 88% of 1H26 revenue), management and servicing fees on ~$3.2B of fee-paying AUM (Asset Management, 12%), and a small mortality-data business (Technology Services).
The economics in one line: buy policies at a high implied yield from sellers, sell them at a low implied yield to fund investors, and earn fees on the funds that hold them. The size of that spread, and whether fund investors keep supplying the cash, is the whole investment.
Price (10/21 intraday)
$8.64
Base intrinsic value
$6.27
Bear → Bull · prob-wtd
$1.26 – $13.24 · $6.41
What the price assumes
Today's ~24% realised gain held forever. Management's own target is 20%
The Take: don't own it at $8.64. The marks aren't the problem. The buyer is.
First, a correction to my September 8 note. I said 57% of Q2 revenue was an unrealised mark. That ignored the 10-Q lines that reverse prior marks when policies are sold. Net of those reversals, marks were −$6.8M in Q2 and −$42.0M in 1H26, because Abacus sold down a book it had marked up in 2025 (+$49.3M). Across the last 18 months the marks net to about +$7M. Abacus has been turning its marks into cash, and cash-basis earnings were stronger than reported this year. The marks finding is withdrawn. The verdict stands, for a better reason.
The real issue is who pays. In 1H26, 85% of realised gains ($143.9M of $169.6M) came from selling policies to funds Abacus manages. In FY25 it was 828 of 1,059 policies sold. From Q2 the balance-sheet marks are set by the same transactions: under the new “Historical Return Method”, each policy is carried at cost × the return earned on past sales by risk grade. So the gain, the mark and the fee all come from one source: fund investors paying full price for policies Abacus selects, prices and services. That works as long as money keeps coming in ($544M in 1H26, already past the $500M target). The Coventry litigation exhibits unsealed in August go straight at it. They allege that the life-expectancy provider Abacus leaned on, Lapetus, saw only 41.5% of the deaths it predicted, which would mean policies were priced as if the insured would die sooner than they do. If fund investors decide they overpaid, inflows stop, and so do deployment and the spread. That is the bear case, and it has no floor above run-off value.
The spread is a one-time windfall, not a margin. The CIO's own letter shows the implied discount rate on Abacus's sales fell from 22% (2023) to 10% (March 2026). Selling inventory into a market whose required return halved produced FY25's 32.5% realised gain. That is a re-rating of inventory. It does not repeat. Management's long-term target is 20%. Run the business at that target and it is worth $6.27. Working backwards, $8.64 requires a ~33% gain spread on deployed capital, about 24% realised, held forever, while deployment also grows to $950M.
The headline EPS flatters by 2x. FY26's “$1.00–1.05 adjusted EPS” is gross: it is quoted before tax on the add-backs. The company's own net-of-tax figure is about $0.87. Treat stock comp and the recurring deal and litigation costs as the real expenses they are, add back only the acquired-intangible amortisation, and economic EPS is $0.49. At $8.64 you are paying 17.6x real earnings. Strip out the goodwill and the book is $1.10 a share, so about $740M of the $845M market cap is a bet that the spread lasts.
Recommendation: AVOID. Sell existing positions into strength above $8. Accumulate only below ~$6.27, where you pay for management's own target and nothing more. Below $5.00 it becomes a real buy. Not a short: a $100M buyback (12% of the cap), 46% insider ownership, a 34M-share float and a Street at $14 make borrow expensive and squeezes cheap. And the bull case ($13.24) is roughly what the Street is underwriting. Q3 on Nov 5: watch the third-party share of realised gains, the realised gain % (below 22% confirms the base), fund inflows, and how the $400M securitisation is booked (fee or gain on sale).
Risk Profile: Very High
Related-party realisation and fund-flow dependence. 85% of realised gains come from Abacus-managed funds, and those same trades calibrate the marks. If inflows stall, deployment, spread and fees fall together. Abacus is long fund-investor confidence.
SEVERE
Life expectancy, mortality and methodology. The inventory is strangers' remaining lifespans, and longer lives mean lower returns, so Abacus is short longevity. Valuation moved to the Historical Return Method in Q2-26, a year after a short attack, with no impact quantified. Lapetus actual-to-expected deaths of 41.5% (Coventry exhibits) are unrebutted.
SEVERE
Spread compression. The implied discount rate went 22% → 10%. A move back toward 13–15% from rates or competition (Coventry is #1 by volume) compresses the realised gain toward the bear case. Abacus is short rates and credit spreads.
HIGH
Leverage and refinancing. ~$412M debt against $107M adjusted tangible equity. 9.875% notes (callable 2/27, due 11/28), SOFR+525 term loan, 12% PIK sponsor note. FY27E interest cover ≈2.8x on economic EBIT.
HIGH
Governance and insiders. Founders and the sponsor own ~59%. Insiders have only sold since August. Related-party acquisition of NIB from the CEO and founders. Adjusted-NI definition re-cut between 10-Ks. Non-accelerated filer with no auditor ICFR attestation.
MED–HIGH
Litigation. Abacus v. Coventry is advancing, with Coventry counterclaims and discovery under way. Shareholder-firm investigations, no class action filed yet.
MEDIUM
What this really is: a levered originate-to-distribute platform whose distribution channel is its own funds. The widest bars in the tornado below are fund inflows (capital deployed) and the gain spread. Those two, not the discount rate and not the one-time mark haircut, decide the value. Size it as a flow-dependent specialty-finance trade, never as a core hold.
Value per share by scenario and method vs price
Framework: why equity-level, not FCFF
For a financial-services firm, debt is raw material. Abacus finances policy inventory with secured, non-recourse and fund-level borrowing, so “operating cash flow before financing” is not a meaningful number. Policy purchases run through operating cash flow, which was −$209M in FY24 and +$131M in 1H26 purely from inventory timing. Following Damodaran's treatment of financial firms, the model values equity directly:
1. Excess Return Model (60%). Equity = adjusted operating tangible book + PV of (economic NI − cost of equity × book). Reinvestment is the equity needed to carry a larger policy inventory: 45% equity-funded at 2.0x turnover. With clean-surplus accounting this is identical to discounted FCFE, and the model shows both: $7.12 vs $7.11. 2. Sum of the parts (40%). Fee-related earnings × alt-manager multiple, plus Life Solutions earnings × specialty-finance originator multiple. 3. Adjusted NAV is the floor, and with asset-management run-off it sets the bear case.
The result: the equity is almost entirely franchise value. Adjusted tangible book is $1.10 a share, so every dollar of the price above that is a claim on future spreads and inflows.
10-K / 10-Q modifications: reported → economic
Item (10-K / 10-Q)
Company treatment
Our treatment
Impact
Goodwill $252.8M
Kept in equity; tested qualitatively, never impaired
Excluded. Value is earned from the flow, not paid-for acquisitions
−$2.58/sh book
Acquired intangibles $59.8M + amortisation ($15.2M in 2026)
Asset in book; amortisation added back to adjusted NI
Excluded from book and from cost (consistent). DTL on them ($13.6M) reversed
−$0.47/sh book
Policy fair value $383.0M (Level 3)
Q2-26 switched from DCF + Monte Carlo to the “Historical Return Method”: cost × risk-grade return (25.1% weighted). Prospective, impact not quantified; LE and discount-rate sensitivities dropped
Base 10% haircut ≈ 40% of the ~$77M embedded mark; bear 20% (≈ cost); bull 0
−$0.29/sh base
Unrealised vs realised gains
Both in revenue and in adjusted NI
Kept, but tracked: +$49.3M net marks FY25, −$42.0M 1H26 as the book sold down. Over 18 months they net to +$7M, so marks are not the problem
Neutral
Stock comp ($15.5M FY25, $13.2M 1H26)
Added back
Real expense. RSUs therefore not added to share count
−$0.20/sh/yr
“Business acquisition & special projects” ($11.8M FY25, $11.2M 1H26; includes Coventry litigation)
Added back
Recurring for a serial acquirer that is also suing its largest competitor: $10–12M/yr from FY27
−$0.09/sh/yr
Tax on the add-backs
Headline $1.00–1.05 FY26 guide is gross. Tax on add-backs only shown in the reconciliation
Everything taxed at 28% (FY25 effective 29.3%)
−$0.16/sh FY26
IRC 162(m) non-deductible exec pay
Added back
Real cash tax
−$0.02–0.03/sh
Related-party realisation (85% of 1H26 realised gains; 828 of 1,059 FY25 sales to Abacus-managed funds)
Recognised at transaction price; same transactions calibrate the marks
No haircut to booked revenue. Forward gain spread normalised to management’s 20% target; bear case = fund-flow stall
Operating raw material (financial-services treatment). Valued at equity level, interest in earnings
In earnings
Series A convertible preferred $5M
Mezzanine
Subtracted as a claim (converts at $10, out of the money)
−$0.05/sh
Adjusted-NI definition change
FY24 adjusted NI shown as $46.5M gross in FY24 10-K, re-presented as $30.7M net in FY25 10-K
Use the net figure for history; ignore the gross headline
Comparability
Balance-sheet bridge (6/30/26, base)
Stated common equity (6/30/26 10-Q)
$434.7M
Less goodwill (Carlisle $101M, Abacus Settlements, FCF, AccuQuote, NIB)
−$252.8M
Less acquired intangibles (management contracts, customer lists)
−$59.8M
Add back deferred tax liability on intangibles (reverses with them)
$13.6M
Less policy mark haircut: 10% of $383.0M fair value, after 26% tax
−$28.3M
Adjusted tangible equity
$107.4M · $1.10/sh
Other investments: carry $73.0M → value $69.0M (Manning & Napier at cost; rest at 80%)
−$4.0M
Less Series A convertible preferred (mezzanine)
−$5.0M
Adjusted NAV (run-off floor before franchise value)
$98.4M · $1.01/sh
Memo: stated book per share / price-to-book
$4.44 · 1.94x
Memo: price / adjusted tangible equity
7.9x
Earnings bridge: company “adjusted” → economic
FY25A
1H26A
FY26E
FY27E base
Company “adjusted net income” — gross (guidance basis)
$85.7M
$47.2M
$101.0M
$101.6M
less tax the company leaves on its add-backs
−$12.1M
−$7.7M
−$15.2M
−$14.0M
Company adjusted NI — net of tax
$73.6M
$39.5M
$85.8M
$87.5M
reverse SBC add-back (after tax)
−$11.6M
−$9.9M
−$20.1M
−$20.9M
reverse “business acquisition & special projects” (after tax)
−$8.8M
−$8.4M
−$14.3M
−$8.9M
reverse 162(m), warrant, credit-loss, NCI, other add-backs
−$3.8M
−$1.7M
−$3.4M
−$3.4M
Economic net income (keeps only the acquired-intangible amortisation add-back)
$49.4M
$19.6M
$47.9M
$54.4M
per share
$0.50
$0.20
$0.49
$0.56
of which net change in unrealised policy marks (after tax)
$36.7M
−$31.3M
—
—
Realised-basis economic net income
$12.7M
$50.9M
—
—
Economic NI = GAAP NI + acquired-intangible amortisation after tax (consistent with excluding those intangibles from book). Every other add-back is reversed. Projections tax all of it at 28%.
What revenue is made of ($M): realised gains vs marks vs fees
realised gains net of premiumsnet change in unrealised marksfees, commissions, originations, tech
Marks added $53.8M in FY24 and $49.3M in FY25, then reversed $42.0M in 1H26 as the book went from 804 to 622 policies. The marks have converted to cash. The question is who supplied it:
1H26 realised gains on policies sold or matured, by counterparty (10-Q note: $143.9M related party vs $25.7M non-related).
Forward slate: base case (bull and bear economic EPS shown)
FY25A
FY26E
FY27E
FY28E
FY29E
FY30E
Capital deployed into policies
$525M
$690M
$770M
$840M
$900M
$950M
Gain spread (policy revenue ÷ capital deployed)
36.4%
33.5%
30.0%
28.0%
27.0%
27.0%
≈ realised gain on cost (spread ÷ 1.34)
32.5%
25.0%
22.4%
20.9%
20.1%
20.1%
Asset-management revenue
$34M
$31M
$40M
$48M
$55M
$61M
Total revenue
$235M
$277M
$288M
$303M
$320M
$342M
Cash operating costs (ex SBC, ex deal/legal)
$104M
$124M
$134M
$143M
$152M
$161M
Interest expense
$38.8M
$38.3M
$40M
$42M
$44M
$46M
GAAP EPS
$0.36
$0.38
$0.45
$0.50
$0.56
$0.64
Company adjusted EPS — gross (how guidance is quoted)
$0.86
$1.03
$1.04
$1.06
$1.10
$1.18
Company adjusted EPS — net of tax
$0.74
$0.88
$0.90
$0.92
$0.97
$1.05
Economic EPS — base
$0.50
$0.49
$0.56
$0.58
$0.61
$0.68
Economic EPS — bull
$0.50
$0.94
$1.21
$1.43
$1.57
Economic EPS — bear
$0.35
−$0.06
−$0.08
−$0.05
−$0.04
FCFE per share — base (after equity tied up in inventory)
$0.40
$0.42
$0.48
$0.57
Price ÷ economic EPS at $8.64
17.3x
17.6x
15.5x
14.8x
14.1x
12.7x
Price ÷ company gross adjusted EPS
10.0x
8.4x
8.3x
8.2x
7.9x
7.3x
FY26E = 1H26 actual + H2 base. Consensus adjusted EPS (gross basis): FY26 ≈ $1.04, in line with base; FY27 ≈ $1.20, between base ($1.04) and bull ($1.38). Shares held at 97.8M basic. Buybacks show up as FCFE paid out, not as a shrinking share count.
Economic EPS path vs the company's adjusted EPS
Excess Return Model: base ($M, from 6/30/26)
H2-26
FY27
FY28
FY29
FY30
Opening operating tangible equity
$34.4M
$40.7M
$55.5M
$71.3M
$84.8M
Economic net income
$28.3M
$54.4M
$57.0M
$60.1M
$66.6M
Equity charge (CoE 11.28% × book)
−$1.9M
−$4.6M
−$6.3M
−$8.0M
−$9.6M
Excess return
$26.4M
$49.8M
$50.8M
$52.1M
$57.0M
FCFE (NI − equity added to inventory)
$22.0M
$39.5M
$41.3M
$46.6M
$55.3M
Opening operating tangible equity $34.4M excludes the $73.0M of non-operating investments, which are added back at value. Terminal: FY31 NI grows 3% and 12% is retained, giving a TV of $729M (74% of value). Rolled forward to 10/21 at the cost of equity. ERM $7.12 · FCFE $7.11.
Life Solutions: FY27E after-tax economic earnings × multiple
−$12.5M × 4x = −$50M
$44.3M × 7x = $310M
$78.5M × 9x = $707M
Other investments at value (Manning & Napier stake etc.)
$51M
$69M
$73M
Policy mark haircut (one-time, after tax)
−$57M
−$28M
$0M
Series A preferred
−$5M
−$5M
−$5M
Equity value per share
$0.11
$4.98
$10.31
Asset management: fee-related-earnings margin 30/35/40%; multiples 10/14/18x (small, related-party-concentrated alt manager). Life Solutions: 4/7/9x, the range for gain-on-sale originators.
Scenarios
Scenario
Value
vs price
Weight
Bear — fund inflows stall after the Lapetus evidence; realised gain ~15%; deployment flat at $600M; costs cut; marks −20%. Going-concern equity ≈ $0.25, so value = run-off floor
$1.26
-85%
25%
Base — deployment grows to $950M by FY30; realised gain normalises to management’s own 20% long-term target by FY29; AUM fees $61M; marks −10%
$6.27
-27%
55%
Bull — today’s ~25% realised gain holds forever; deployment $1.13B by FY30; AM revenue $80M; no mark haircut
$13.24
53%
20%
Probability-weighted
$6.41
-26%
Base: Excess Return Model (≡ FCFE, 60% weight)
$7.12
-18%
Base: Sum of the parts (40% weight)
$4.98
-42%
Base: Adjusted NAV (floor, not weighted)
$1.01
-88%
Driver tornado: swing in value per share
Fund inflows and the gain spread together swing value by about ±$4. The one-time mark haircut is worth ±$0.11, because a mark is only realised once. The flow is what you own.
Sensitivity: steady-state gain spread × cost of equity
Cost of equity ↓ · Steady-state gain spread (≈ realised gain) →
22% (≈16%)
25% (≈19%)
27% (≈20%)
30% (≈22%)
33% (≈25%)
36% (≈27%)
10.00%
$4.33
$5.99
$6.95
$8.32
$9.97
$11.63
10.75%
$4.12
$5.63
$6.50
$7.73
$9.25
$10.76
11.28%
$3.99
$5.42
$6.23
$7.38
$8.82
$10.25
12.00%
$3.84
$5.18
$5.92
$6.98
$8.31
$9.65
13.00%
$3.67
$4.89
$5.57
$6.51
$7.73
$8.96
Spread applies from FY28 (FY27 phased). Red = more than 15% below price, amber = 5–15% below, grey = within ±5%, green = above. Price $8.64 needs ≈33% (≈24% realised) at an 11.3% CoE.
Sensitivity: steady-state gain spread × capital deployed
Capital deployed vs base path ↓ · Steady-state gain spread →
22% (≈16%)
25% (≈19%)
27% (≈20%)
30% (≈22%)
33% (≈25%)
36% (≈27%)
80% of base
$2.45
$2.60
$3.25
$4.17
$5.32
$6.46
90% of base
$2.72
$4.01
$4.74
$5.78
$7.07
$8.36
100% of base
$3.99
$5.42
$6.23
$7.38
$8.82
$10.25
110% of base
$5.26
$6.83
$7.73
$8.99
$10.57
$12.14
120% of base
$6.52
$8.24
$9.22
$10.59
$12.31
$14.03
$2.45 is the run-off floor (adjusted NAV + asset-management contracts at 14x FRE). At management's 20% target (~27% spread), the price needs deployment about 16% above the base path (~$1.1B a year by FY30).
Cost of equity and terminal assumptions
Risk-free (10-yr UST, 10/20/26)
5.12%
Equity risk premium (house)
4.40%
Beta: bottom-up alt managers / specialty finance (regression 0.05 is a thin-float artefact)
1.40
Cost of equity
11.28%
Terminal growth / marginal ROE on retained equity
3% / 25% → reinvest 12% of NI
Terminal value share of ERM (base)
74% — above the 70% flag; it is a flow business with almost no book