Data room / Successor concepts and their reviews

Version 2 review: insurance and actuarial

Review, written by the insurance and actuarial reviewer, 4,701 words, about 20 min to read

concepts/review-insurance-actuarial.md Open the file as written

Insurance & Actuarial Review — HOMEDRAW™ Concept Sheet v0.1

Reviewer role: credentialed actuary (FSA/MAAA-equivalent), insurance-regulatory and insurance-asset-management lens. Date: 2026-08-01. Stage: mini-panel review of the successor concept; focus per mandate is the INSTITUTIONAL FUNDING QUESTION — the concept's central bet. Equity Shield sections of my prior charter are moot (the concept deletes it, correctly).

Materials: concepts/HOMEDRAW-CONCEPT.md; ANALYSIS.md §§1, 3, 8; my Stage 1 report (reviews/insurance-actuarial.md) and Stage 2 cross-exam (reviews/crossexam-insurance-actuarial.md), whose duration and breakeven frameworks are reused and adapted below. All arithmetic recomputed this session in Python (models on file in session scratchpad: homedraw_model.py, homedraw_exits.py, homedraw_funding.py).

Standard deal modeled throughout: $625,000 home, $375,000 senior (60% CLTV) at 3.5% (deep-lock vintage; 6.5% variant shown where material), $1,500/mo × 84 months = $126,000 drawn, 9.5% fixed PIK compounding monthly, 7% costs of sale, 30-year legal final.

Claim tags: SETTLED LAW / LIKELY INTERPRETATION / STATE-DEPENDENT / AMBIGUOUS / SPECULATIVE.

Posture: adversarial, as chartered. Where the concept fixes what I killed, I say so — an adversary who cannot concede verified fixes produces noise, not diligence. Three of my Stage 1 kill mechanisms are genuinely cured here; two new defects of comparable seriousness are introduced. Both are quantified below.


0. Arithmetic integrity of the concept sheet itself (read this first)

Finding 0.1 — The base-case sketch understates its own accrual by ~11%, and every downstream number inherits the error. (MAJOR)

At the sheet's own stated terms ($1,500/mo × 84, 9.5%, compounding monthly), the draw-end balance is $177,993, not "≈ $160K" ($168,168 even at annual compounding — no compounding convention produces $160K). Corrected trajectory:

Year Concept sheet Correct (monthly comp)
7 (draw-end) ≈$160,000 $177,993
12 (sale example) payoff ≈$250,000 $285,680
13 ≈$276,000 $314,033
~14 ("hits $300K cap") year ~14 year ~12.6
30 (legal final) not shown $1,568,913

PIK at 9.5% monthly doubles every 7.3 years. The sheet's year-12 homeowner example overstates the homeowner's net by ~$36K, the Draw-Limit example is mis-timed by ~1.5 years, and — most importantly — the sheet never prints the legal-final balance: $1.57M owed at year 30 on $126K drawn. That number drives Findings 4.1–4.3 and must appear in every consumer scenario table the sheet itself mandates (5/10/20/30-yr disclosures). SETTLED (arithmetic on the sheet's own terms). Severity: major — not because the design breaks, but because a funding pitch built on a concept sheet that understates its own asset balance 11% will be re-derived by every warehouse lender and rating analyst, and the sponsor's credibility problem from Equity Flow (investor O3) repeats on day one.


1. THE SCHEDULE B QUESTION (decisive, per mandate)

Finding 1.1 — Schedule B reporting is genuinely winnable for directly held

whole loans; this is the single largest economic fact in the concept. (Supports thesis, with conditions)

The statutory-accounting question decomposes into three separate gates the concept sheet compresses into one:

(a) SSAP No. 37 classification (NAIC accounting → Schedule B reporting). SSAP 37 defines a mortgage loan as a debt obligation that is not a security, secured by a mortgage or deed of trust on real estate, held directly or via participation. There is no amortization, coupon-frequency, or lien-priority requirement in the SSAP 37 definition; second liens are reportable on Schedule B, and contractually deferred interest is not "overdue" interest (the good-standing tests key off delinquency versus contract terms, and a PIK loan is performing per its contract). Unlike Equity Flow's ride-along — which I killed on this exact point in Stage 1 FX-A ("an appreciation-contingent, no-maturity claim does not become a mortgage loan by being secured by a deed of trust") — HomeDraw is a true fixed-rate note with a stated maturity, standard security instrument, and no contingent payment. My FX-A objection does not reach it. LIKELY INTERPRETATION — with the structural condition that insurers hold whole loans or true participations. If they instead hold notes issued by the funding trust, the paper is a security → Schedule D ABS (rated: fine; unrated: dead), and if they hold fund/LP interests, it is Schedule BA. The concept's funding diagram must say "whole-loan / participation private placement," not "insurance private placement," which usually means notes.

(b) RBC factor. For life insurers, directly held residential mortgage loans in good standing on Schedule B carry a C-1 pre-tax factor of 0.68% ("residential mortgages — all other"); the residential factor is not LTV-graduated or lien-graduated the way commercial CM categories are. Even the downside cases are mild: a Schedule BA mortgage-character vehicle holding primarily junior liens is CM3 at 3.00%, and 2024–2026 NAIC actions extended 0.68% look-through treatment to affiliated (2024) and unaffiliated (April 2026) Schedule BA vehicles holding performing residential mortgage loans. The regulatory current is toward mortgage-loan look-through, not away from it. LIKELY INTERPRETATION on 0.68% for direct holdings; AMBIGUOUS whether a PIK-only junior would be pushed to CM-style treatment on examiner challenge — but even that outcome (3.00%) sustains the thesis.

Net-yield arithmetic (9.5% nominal = 9.92% effective annual at full recovery; 10% cost of capital, same convention as my Stage 1 Finding 3.1):

Treatment C-1 factor Capital drag Net yield
Schedule B residential (this concept's claim) 0.68% 0.07% 9.86%
Schedule BA mortgage-fund CM3 (junior-lien downside) 3.00% 0.30% 9.62%
Stage 1 Equity Flow ride-along (Sch BA other) 30% 3.00% 6.92%
Stage 1 residual treatment 45% 4.50% 5.42%

The inversion the concept sheet hypothesizes is real: ~290–440 bps of net yield swings on this classification. A 9.86% statutory net on a 0.68%-charge asset is not "bond-like return for venture risk" — it is one of the highest capital-adjusted yields available to a life company, which is precisely why gate (c) matters.

(c) State-law admissibility — the gate the concept sheet does not mention. (MAJOR caveat.) Schedule B reporting and RBC are NAIC constructs; whether the loan is an admitted asset within the enumerated mortgage class is state investment law. NY Ins. Law § 1404 permits second-mortgage loans, but its high-LTV residential bucket (up to 90% of value) requires "monthly payments of principal and interest sufficient to repay the loan" within statutory limits — an amortization test a PIK loan fails by construction; a non-amortizing junior at up to ~75%+ projected CLTV falls outside the favored bucket and into lower LTV limits or the leeway basket for NY-regulated companies. STATE-DEPENDENT — and materially so: it does not kill the thesis (the large annuity buyers are domiciled in IA/AZ/DE/MI-style modernized-code states, and Bermuda-reinsured capital prices under BSCR, not § 1404), but it shrinks the buyer universe and must be mapped domicile-by-domicile before the funding deck claims "insurance bid."

(d) The re-classification tail risk. NAIC is actively policing HEI-adjacent paper (SAPWG 2025 items tightening what counts as a Schedule BA mortgage fund; the HEI classification debate that motivated this concept's redesign). HomeDraw is distinguishable — no appreciation share, fixed rate, stated maturity — but a first-of-kind PIK consumer junior at scale invites an SVO/SAPWG look, and an adverse interpretive letter mid-fund re-prices every insurer holder from 0.68% to 3–30% overnight. The mitigation is cheap and mandatory: obtain the classification in writing (domiciliary-regulator concurrence and an SSAP 37 opinion from a national statutory-accounting practice) before insurer capital closes, not after. AMBIGUOUS by nature; manageable. Severity of (c)+(d) combined: major, not critical — the first concept in this file where I can write that sentence about the insurance leg.

Finding 1.2 — What Schedule B does NOT solve: ALM, valuation, and income

quality. (MINOR-to-MAJOR, buyer-dependent)

  • Duration/convexity: a PIK asset's duration ≈ its exit horizon (13–21 years, Finding 3.1) with mortgage-style negative convexity both directions (refi when rates fall; extension when rates rise, since mobility collapses with rates — my Stage 1 4.1 mechanism, which survives into this product, merely capped now at year 30). Payout-annuity and PRT writers genuinely want 15–20-year duration, so this is a fit for a real buyer segment — but C-3 modeling and cash-flow-testing treatment of a zero-cash asset will haircut the bid. LIKELY INTERPRETATION (practice, not law).
  • Statutory income with no cash: accrual income on Schedule B is admitted, but a book that accretes for a decade produces earnings without distributable cash; asset-adequacy testing will require the insurer to fund liability outflows elsewhere. Insurers cap allocation percentages accordingly.
  • Good-standing drift: RBC does not LTV-test residential loans annually, so the accrual-driven CLTV drift of Finding 4.2 costs the insurer no capital until default — an arbitrage that works until an examiner or SAPWG notices (see 1.1(d)).

Bottom line on Question 1: PIK + no scheduled payments + junior lien does not disqualify the loan from Schedule B under current SSAP 37 for directly held whole loans/participations (LIKELY INTERPRETATION); the realistic factor is 0.68% (downside 3.00%), netting the insurer ~9.6–9.9% — an inversion of my Stage 1 30%/45% arithmetic worth ~290–440 bps. The binding constraints move to state admissibility (STATE-DEPENDENT), ALM fit, and re-classification tail. The concept's central accounting bet is, on current guidance, more right than wrong — it is the first load-bearing funding claim in this file to survive contact.


2. RATING FEASIBILITY

Finding 2.1 — Ratability is credible on precedent, but "first to rated

takeout" is false: the rated precedent for THIS EXACT INSTRUMENT already

exists and belongs to a competitor. (MAJOR — kills the moat claim, not the funding claim)

The concept claims its edge is being "first to rated takeout for this paper." The record says otherwise:

  • Finance of America HomeSafe / HomeSafe Second: proprietary, non-recourse, fixed-rate, deferred-payment (PIK), second-lien reverse mortgages — the same instrument as HomeDraw restricted to 55+ — have been securitized repeatedly with KBRA ratings to AAA (recent deal: 2,024 loans, ~$106K average balance, 8.30% WA coupon, AAA/AA senior classes).
  • Point/Redwood (2021 on) and Unlock HEI securitizations were rated by DBRS Morningstar without decades of seasoning, on a harder asset (no stated payments, appreciation-contingent) — via AVM haircuts, stressed-HPA paths, structural credit enhancement, and long trust legal finals.
  • The closed-end-second / HELOC rated market (2023–2026: Rocket, Figure, Saluda Grade programs; DBRS/KBRA/Fitch criteria published) supplies the collateral-file, servicing, and rep-and-warranty template wholesale.

So: can this get rated in fund-one's lifetime? Yes — realistic timeline 3–5 years from first origination, conditional on (i) ~$150–250M of collateral, (ii) licensed origination and a rated-eligible servicer with backup, (iii) legal opinions on state licensing/usury/reverse-statute capture, (iv) enhancement sized to the crossover stress in Section 4. This is not the year-7-8 fiction of Equity Flow — my Stage 1 3.2 objection was specifically the absence of a legal final and of any settlement mechanism; the concept cures the exact defect I named. LIKELY INTERPRETATION (agency practice, precedent-based).

But the moat inverts: the precedent that makes HomeDraw ratable makes it copyable by the incumbent who already owns the rating relationships, servicing stack, and shelf. FoA extends HomeSafe Second below 62 by changing an underwriting parameter; HomeDraw builds all of it from zero. "First to rated takeout" is not credible; "fast follower into an incumbent's rated market with a CAC thesis" is the honest description, and the funding deck should say so.

elsewhere: the concept solves a problem the precedent shows doesn't exist,

and buys a new one. (MAJOR)

The FoA deals are rated with no borrower-level stated maturity at all — the loans are due at death/move-out/sale, and the trust carries a distant legal final. The rating problem my Stage 1 3.2 identified was cured by precedent structuring at the trust level, not by putting a balloon in the consumer note. Meanwhile the loan-level 30-year maturity:

  • Breaks the Reg Z reverse-mortgage classification the sheet hopes for. § 1026.33(a)(2) requires that principal and interest be due "only after" death, transfer, or cessation of occupancy. A stated 30-year maturity makes the balance due while the borrower is alive and occupying → not a reverse mortgage under Reg Z → the § 1026.43 ability-to-repay rule applies to a negatively amortizing, no-income-doc balloon loan — a combination §1026.43 is more or less purpose-built to prohibit. LIKELY INTERPRETATION (regulation text is clear; application untested). Flagged to legal-regulatory, but it is a funding-design finding: the maturity exists for ratability, and ratability doesn't need it.
  • Guarantees that every loan reaching it is distressed (Finding 4.3: $1.57M due; homeowner residual $0 at ≤3% HPA; lender impaired ~10% even at trend). An agency models the legal-final bucket at its true recovery; the maturity contributes stress, not comfort.

Fix: drop the loan-level maturity (restoring the clean §1026.33 posture and the "due when life settles" promise), keep due-at-death/sale/move-out, and put the legal final where the precedent puts it — on the securitization trust (2060s-style final). This strictly dominates the current design for rating, Reg Z, and consumer optics simultaneously. AMBIGUOUS only in that agencies prefer defined terms all else equal; the FoA precedent shows the preference is not binding.


3. DURATION / EXIT MODELING AND THE COST OF CARRY

Finding 3.1 — Rerun of the mobility+mortality model for the new box:

mean exit 13–15 years (vs ~24 for Equity Flow), 6–10% of loans forced out

Changes from my Stage 1 model: any-age box (modeled at 57 average, 52 for the deep-selection case), lock-in used for marketing only (no eligibility gate), no appreciation-share sale penalty (Equity Flow's contract added a second lock-in by taxing the sale 25–43% of appreciation; HomeDraw does not — sale merely stops a 9.5% accrual, which is an incentive to exit), and a refi channel (the borrower can refinance the accrual away when rates permit). Net: mobility on this book is higher than the Equity Flow book. Hazards: Gompertz mortality (μ(57)≈0.6%, μ(77)≈4.4%) + mobility as tabulated + 0.5–1%/yr refi; 30-year truncation with survivors forced out at legal final:

Scenario Mean exit Median P(>20y) P(forced at 30y legal final)
Unselected mover ~7%/yr, age 57 10.0 8 12.4% 2.3%
Base: self-selected 4.5%/yr, age 57 13.2 12 23.9% 6.2%
Lock-in marketing works, 3%/yr 15.3 15 32.8% 10.1%
Deep lock-in 2%/yr, age 52 couple 21.1 25 59.3% 38.2%

Two structural observations. First, the concept genuinely fixes my Stage 1 4.1 category error: this asset prices off a fixed accrual, not off appreciation, so extension no longer destroys value through the discount of a contingent payoff — extension is now ordinary mortgage extension risk, which markets price daily. Second, the marketing self-contradiction survives in softened form: the funnel still targets lock-in (low mobility), i.e., markets for duration, while the credit box needs early exits (Finding 4.2's drift). The better the lock-in targeting works, the worse Finding 4.3 gets — the deep-lock-in row puts 38% of the book at the legal final owing $1.57M per original $126K drawn.

Finding 3.2 — Negative carry: the warehouse phase burns ~$9.6M of cash per

$100M program before any rated takeout, and the pre-rating capital stack

does not clear the consumer rate ceiling. (MAJOR)

A PIK asset pays the warehouse nothing. Modeling a 794-loan, $100M-commitment book originated over 24 months, warehouse advancing 75% of accrued balance at ~SOFR+300 (7.4% cash):

Year Book accrued Warehouse draw Cash carry (equity-funded)
1 $4.0M $3.0M $0.22M
2 $15.8M $11.9M $0.88M
3 $32.3M $24.3M $1.79M
4 $50.5M $37.9M $2.80M
5 $70.4M $52.8M $3.91M
Cumulative 5-yr cash interest, zero asset cash flow $9.6M

Required asset yield by funding phase (effective-annual build-up: funding + 50bps servicing + 50bps through-cycle credit + 40bps origination amortization + 25–50bps OID/extension premium demanded by holders), against the 9.92% EA the 9.5% coupon delivers at full recovery:

Phase Blended funding Required asset yield Margin vs 9.92% Implied consumer APR to break even
Warehouse 75/25 @7.4%/16% 9.55% 11.25% −132 bps 10.35%
Rated takeout 85/15 @6.4%/14% 7.54% 9.45% +47 bps 9.06%
Insurer whole-loan Sch B (net 7.5% req.) 7.50% 9.15% +77 bps 8.88%

Read: at 9.5% the product is underwater by ~130 bps for its entire pre-rating life and clears by only ~50–80 bps after it. The venture is a bet that ~3 years and ~$10M+ of negative carry per $100M buys entry to takeout economics with a sub-100bp steady-state margin — viable only at scale, only if the rating arrives on schedule, and only if takeout execution matches the modeled 85%+ advance at ≤6.5% weighted coupon. Raising the consumer rate to the 9.75% ceiling adds 25 bps; raising it to cover the warehouse phase (10.35%+) prices above FoA's incumbent product and erodes the consumer thesis. Additionally, rated bonds on PIK collateral either accrete (Z-bond pricing concessions) or depend on prepayment cash, and the reverse-mortgage precedent deals rely on ~3-year call/refinancing structures — the "takeout" is not one exit but a chain of refinancings across 20+ years, which jams in 2022-style windows. LIKELY INTERPRETATION (market practice); arithmetic SETTLED on stated assumptions.


4. CREDIT: THE NON-RECOURSE TAIL

Finding 4.1 — Adapted breakeven framework: the −27% breakeven of Equity

Flow becomes a breakeven that MOVES AGAINST THE LENDER ~0.3–0.5pp/yr and

goes positive at year ~14. (CRITICAL for pricing; the sheet does not price it)

Equity Flow's escrow claim was static ($126K), giving a fixed −27.1% breakeven drawdown. HomeDraw's claim compounds at 9.5% against collateral compounding at HPA plus senior amortization. Required home value for full recovery (payoff = balance ≤ V×0.93 − senior), senior at 3.5%:

Exit yr Balance Senior Required V Breakeven drawdown vs $625K Required HPA/yr
7 $177,993 $318,917 $534,312 −14.5% −2.21%
10 $236,422 $290,351 $566,423 −9.4% −0.98%
12 $285,680 $269,573 $597,046 −4.5% −0.38%
15 $379,460 $235,552 $661,303 +5.8% +0.38%
20 $609,037 $170,289 $837,985 +34.1% +1.48%
25 $977,510 $92,565 $1,150,619 +84.1% +2.47%
30 $1,568,913 $0 $1,687,004 +169.9% +3.37%

(6.5%-senior variant: breakevens are uniformly worse — e.g., −11.0% at year 7, +0.76%/yr required at year 15.) The year-7 cushion is roughly half of Equity Flow's (−14.5% vs −27.1%) because the claim is $178K rather than $126K and sits behind a larger senior. From year ~14, any flat-to-down market impairs the loan with zero housing "crash" required; by the legal final the loan needs above-trend HPA (3.37%/yr for 30 years) merely to return par. The 9.5% coupon on the long tail is partially fictional: yield-to-worst, not coupon, is the ratable number. SETTLED (arithmetic).

Recovery grid (payoff/balance):

HPA \ exit yr 7 10 15 20 25 30
+5% 1.00 1.00 1.00 1.00 1.00 1.00
+3% 1.00 1.00 1.00 1.00 1.00 0.90
+2% 1.00 1.00 1.00 1.00 0.88 0.67
+1% 1.00 1.00 1.00 0.88 0.67 0.50
0% 1.00 1.00 0.91 0.67 0.50 0.37

2007-replay paths (peak at origination): national −26.5% trough → recoveries 0.57–0.89 depending on exit year; sand-state −45% → 0.11–0.54. Unlike Equity Flow there is no escrow-first 1.0x — the entire accrued balance is the junior exposure. Portfolio yield net of these losses (probability-weighted over the Finding 3.1 exit distributions; full-recovery benchmark 9.92% EA): −16 bps at 3% HPA on the base book, −55 to −114 bps at 2%, −226 to −334 bps at 0%. The 50 bps through-cycle credit charge in Finding 3.2 is therefore adequate only if HPA ≥ ~2.5% and the deep-lock-in scenario does not materialize; a rating agency will size enhancement off the 0%-HPA and replay rows, not the base case.

Finding 4.2 — The Draw Limit is calibrated to the wrong variable and stops

protecting at exactly the moment risk starts compounding. (MAJOR)

Two defects:

(a) "Projected CLTV ≤75% at draw-end" is an HPA-assumption artifact. Actual draw-end CLTV with the corrected $178K balance: 64.6% if HPA = +3%, 74.2% at +1%, 79.5% at 0%, 91.6% at −2%. A limit computed at closing off a 3% projection binds nothing in the flat/down states where it matters. The "objective number printed in the closing documents" cures the AVM-manipulation findings (genuinely — red-team F1/M1 do not survive), but objectivity was bought by fixing the number in the one dimension (realized HPA) the risk lives in.

(b) The Draw Limit only stops draws; accrual is uncapped for up to 23 more years. Post-draw, CLTV drifts up ~5–6.5 pp/yr net in a flat market (9.5% on a growing $178K+ balance vs ~$19–27K/yr of HPA+amortization cushion). The instrument that manages this risk in the rated precedent is the reverse mortgage principal limit factor: proprietary reverse products advance ~25–40% of value at origination sized so accrual-to-death rarely crosses 100%. HomeDraw's combined 75% projected CLTV is roughly double the PLF discipline the rated market imposes on the same risk. What a rating agency will actually require is either (i) a materially lower Draw Limit — solving for 60% projected CLTV at draw-end cuts the draw from $1,500 to ~$1,199/mo; 55% cuts it to ~$875/mo — or (ii) an accrual cap / rate step-down at a CLTV trigger, which reintroduces a contingent term. The consumer product survives (i) in weakened form; the concept sheet has not priced it. LIKELY INTERPRETATION (agency practice); arithmetic SETTLED.

surviving loan matures impaired, and every surviving homeowner is wiped

out. (CRITICAL as designed; cured by the Finding 2.2 fix + 4.2 recalibration)

At year 30 the balance is $1,568,913. Home value at 3% HPA: $1.52M (net $1.41M) → lender recovery 0.90 at trend HPA; homeowner residual $0. At 2.5%: recovery ~0.77, homeowner $0. At 2%: recovery 0.67, homeowner $0. The homeowner's residual hits $0 at roughly year 23 (2% HPA) to year 25+ (3%); "the homeowner keeps 100% of appreciation" is true only for exits before the crossover — for the 6–38% of the book that reaches the legal final, the product consumes the entire house and then forces its sale within a 12-month window. Balance-weighted, this bucket is far larger than its loan count: each legal-final loan carries 8.8x the draw-end balance. The non-recourse cap converts the homeowner catastrophe into lender severity, so the same cells are simultaneously a UDAAP/headline event (elderly borrower, 30 years of disclosed-but-compounding interest, forced sale, zero proceeds) and a credit loss. No rating committee and no insurer investment committee books this bucket at coupon. The sheet's own #1 open question ("negative amortization is the product") is correct but understated: negative amortization without a PLF-style cap and with a balloon is the product. SETTLED (arithmetic); UDAAP characterization AMBIGUOUS but the reputational pricing is not.


5. VERDICT ON THE FUNDING THESIS — WHO HOLDS THIS PAPER

Buyer map, in order of realism:

  1. Life/annuity insurers via whole-loan or participation private placements (post-rating-era, and possibly before it): the real bid. Schedule B at 0.68% nets ~9.6–9.9% — extraordinary on a capital-adjusted basis — and payout-annuity/PRT liabilities want the duration. Constraints: state admissibility mapping (STATE-DEPENDENT, Finding 1.1(c)), allocation caps for zero-cash assets, written classification comfort (Finding 1.1(d)), and a required net yield ~7.5%+ that the 9.5% coupon covers with room. This buyer exists. It is the first time in this diligence file that sentence is true.
  2. Rated securitization senior/mezz (banks, money managers): feasible on the FoA/HEI/CES precedent in 3–5 years at ~6.2–6.5% weighted, 85–90% advance — the takeout the economics require. Accreting-bond and serial-refinancing frictions are real but precedented.
  3. Credit funds (warehouse-phase whole loans / residuals): will buy at 11–13% required yields — which the 9.5% coupon cannot pay (Finding 3.2). They are bridge capital priced as venture cost, not a hold base. Phantom OID income pushes taxable vehicles toward REIT wrappers (the asset is REIT-qualifying — a genuine structuring asset).
  4. Banks: no. Basel risk weights on negatively amortizing junior-lien consumer credit plus the 2008 institutional memory of option-ARM/neg-am books close this channel except as warehouse lenders (where they are already modeled).

Does the required yield produce a consumer rate anyone would accept? Yes — narrowly, and only in the takeout era: the rated/insurer stack requires an asset yield of ~9.15–9.45% EA against the 9.92% the 9.5% coupon delivers, i.e., the consumer rate the capital stack needs (~8.9–9.1% APR steady-state, ~10.35% in the warehouse phase) brackets the 9.5–9.75% the sheet proposes, and 9.5% is market-consistent with the incumbent (HomeSafe Second WA coupon 8.30% plus its 55+/PLF-conservative box). The funding thesis is not the fantasy its predecessor's was. Its honest weaknesses are: a ~130 bps-negative, ~$10M-per-$100M warehouse valley; a sub-100 bps steady-state margin; a crossover credit tail the coupon does not price; and a moat claim that is false because the incumbent already owns the rated market.


VERDICT: FIX-BEFORE-LAUNCH — and, for the first time in this file, the

fix list is a fix list rather than a kill in fix costume.

What is genuinely cured versus my Stage 1/Stage 2 record: the mortality-vs- mobility category error (fixed accrual replaces contingent appreciation), the no-legal-final rating bar, the Schedule BA 30%/45% capital wall (inverts to 0.68–3.00% for whole loans — LIKELY INTERPRETATION), the Equity Shield unlicensed-insurance exposure (deleted), and the AVM-manipulation surface (deleted). What is not cured or newly created, in required order of repair:

  1. Delete the loan-level 30-year maturity (Finding 2.2): unnecessary for rating per the FoA precedent, breaks the §1026.33 reverse-mortgage posture and triggers ATR on a neg-am no-doc balloon, and guarantees the Finding 4.3 catastrophe bucket. Trust-level legal final instead.
  2. Recalibrate the Draw Limit to PLF discipline (Finding 4.2): ~55–60% projected CLTV at draw-end (draw falls to ~$875–1,200/mo) and/or an accrual cap; without it, no rating committee passes the 0%-HPA row and no UDAAP review passes the year-25 homeowner-wipeout cells.
  3. Convert the Schedule B hypothesis into paper before it is load-bearing (Finding 1.1): SSAP 37 opinion, domiciliary-regulator concurrence, state-by-state admissibility map. The concept's central bet is winnable but currently unwritten.
  4. Re-cut the funding plan around the warehouse valley (Finding 3.2): ~$10M negative carry per $100M program, three-plus years, before a takeout whose steady-state margin is <100 bps — this is the venture's true capital requirement and it appears nowhere in the sheet.
  5. Fix the sheet's own arithmetic (Finding 0.1) and restate the moat claim honestly (Finding 2.1): fast follower vs FoA below age 62, competing on CAC and box, not "first to rated takeout."

The single make-or-break number the sponsor must validate: the rated takeout execution: ≥85% advance at ≤6.5% weighted bond coupon (blended cost of funds ≤7.55%) achievable within ~36–48 months of first origination. At that number the 9.5% consumer coupon clears by ~50 bps and every other finding is a repair; 75 bps wide of it (blended ≥8.3%), the required consumer rate exceeds 10.3%, the product loses to the incumbent on price, and the concept fails on the same axis Equity Flow did — not law, but the cost of capital. Validate it the only way it can be validated pre-launch: a rating-agency feasibility engagement (KBRA/DBRS) plus one anchor insurer's written Schedule B classification comfort, both before the first dollar of warehouse draws.

--- end of review ---