# HOMEDRAW CONCEPT v0.1 — INVESTOR / CREDIT DILIGENCE REVIEW (mini-panel, Stage 1) Reviewer: senior credit + venture/PE lens (adversarial). Date: 2026-08-01. Materials: concepts/HOMEDRAW-CONCEPT.md (artifact under review); ANALYSIS.md sections 1/3/8 (prior kill record); my Stage 1/2 Equity Flow reports (reviews/investor.md, reviews/crossexam-investor.md). All quantitative claims below rebuilt independently in Python this session (model script preserved in the session scratchpad, homedraw_model.py; parameters: 1,500/mo x 84 draws, 9.5% monthly-compounded deferred, 2.5 points financed on actual draws, 625K home, 350K senior at 3.5%/30yr, 7% costs of sale, 50bps servicing strip, 5,300/deal fully-loaded origination cost per the prior ops review). Framing per mandate: assume the product has already failed; explain why. Where the redesign genuinely cures a prior kill, I say so — an adversarial review that refuses to concede real fixes produces noise, not diligence. --- ## PRELIMINARY: WHAT THE REDESIGN ACTUALLY FIXED (conceded up front) The concept sheet is a serious response to the record. Conceded as cured or materially improved: the characterization fork (it is a loan, everywhere, in every forum — the central Equity Flow kill is gone); the tax wager (interest is interest; no §453 fantasy; no exit double-count); the market-pause one-sided option (replaced by a fixed dollar Draw Limit printed at closing); the occupancy-default/dementia surface (deleted); the Equity Shield (deleted); arbitration (deleted); the homeowner-as-unsecured-creditor defect (funding trust); TX (excluded); the gross-fee basis (points on actual draws). The lender IRR is the note rate by construction, so the deal-level return engine is no longer an HPA bet. This is a coherent instrument. The failures that remain are therefore different in kind: they live in the CAPITAL STACK, the REGULATORY DEFINITION of the borrower box, DEMAND against a cheaper substitute, and DEFENSIBILITY. Several are severe, and two are structural dilemmas of the same shape as the old characterization fork: the product needs two mutually exclusive things to be true at once. --- ## MARKET M1. The sheet's own base-case arithmetic understates the product it is selling. Rebuilt: 1,500/mo x 84 at 9.5% monthly-compounded = 177,993 at draw-period end (182,442 with 2.5 financed points), not the sheet's "~160K"; year-12 payoff = 286-293K, not "~250K"; year-13 = 314K, not "~276K". The sheet's numbers reproduce only at a ~7.4-7.5% rate. Every consumer example in the concept understates the debt by 11-14%. For a product whose stated #1 risk is negative-amortization optics and whose flagship consumer protection is an honest scenario table, shipping the concept sheet with flattering arithmetic is the exact Equity Flow disease (my prior A3/O3) recurring at birth. Impact: consumer disclosures, Draw Limits, and CLTV projections are all mis-anchored ~12% low. Severity: MAJOR (trust-destroying; trivially checkable). M2. The statutory disclosure table is a conversion killer at the sheet's own rate. The 5/10/20/30-year dollar table the concept (correctly) commits to shows: yr 5 = 117K owed on 90K drawn; yr 10 = 242K; yr 20 = 624K; yr 30 = 1.57-1.61M owed on 126K received. A 40-year-old reading "you could owe 1.6 million dollars on 126,000 of draws" next to an 8.5% HELOC offer converts at HECM-like rates or worse — and HECM, with FHA insurance, a federal counseling apparatus, and a 62+ cohort with no alternatives, took 35 years to reach ~30-60K units/yr nationally. The honest product and the sellable product are again in tension — not by deception this time, but the funnel consequence is the same (prior DC-4/DC-9, relocated). Severity: MAJOR. M3. The borrower box is large on collateral, small on behavior. Under-62 mortgaged owner-occupants at CLTV <= 60% with clean 12-month history: plausibly 15-20M households on collateral criteria. But the converting subset is (a) wants committed monthly income, (b) cannot or will not make HELOC payments (else the HELOC wins — see D-side comparison, AS1), (c) accepts open-ended compounding at 9.5%, (d) in-footprint (ex-TX, ex-functional-reverse-statute friction states). The 15-year-old HEI industry, selling a no-payment product with none of this one's honesty burden, does ~20-30K deals/yr TOTAL. Realistic obtainable volume for a new entrant with a scarier disclosure stack: low thousands of deals/yr at maturity. That number matters because the funding thesis needs collateral mass (S2). Severity: MAJOR. M4. Melting-ice-cube TAM inherited unchanged. The funnel premise is the 2020-22 rate-lock anomaly. Every 50bp rate decline shrinks the locked-in pool, revives cash-out refi and HELOC competition, and prepays the book (fine for the note, fatal for the pipeline). A venture needing 4-6 years of origination to reach its first rated takeout (S2) is short rates on the asset side and long rates on the demand side simultaneously. Severity: MAJOR (CRITICAL for the venture timeline). M5. Competitive convergence has already started. Unison launched a deferred-interest second-lien hybrid ("equity sharing home loan" — reduced current rate plus an appreciation share) in 2024; proprietary reverse lenders (Longbridge/Ellington, Finance of America) already run deferred-payment non-recourse programs with servicing and securitization infrastructure and have every capability to drop the age gate on a non-FHA product. The "under-62 tenure niche" is unoccupied mostly because incumbents priced the same math and passed, or are entering sideways with a subsidized-rate version that reads cheaper to the consumer than 9.5% full-freight PIK. Severity: MAJOR. ## UNIT ECONOMICS U1. Origination is still subsidized, now with worse cash timing. Fully loaded origination cost ~5,300/deal (prior ops review, adopted). Revenue: 2.5 points on ACTUAL draws, financed — i.e., 37.50/mo added to the balance, ~3,150 nominal collected over 7+ years, PV ~2,400 at any reasonable discount. Platform cash contribution per deal at closing: roughly minus 5,300, recovered to roughly minus 2,900 PV over a decade. The prior product at least collected its (indefensible) fee up front. Severity: MAJOR (platform must be venture-subsidized for its entire pre-takeout life). U2. No gain-on-sale engine exists until the paper trades above par. A lending platform's income statement is gain-on-sale plus servicing. Servicing is 50bps on a PIK balance — cash income ~600-900/loan/yr against a live draw-disbursement obligation. Gain-on-sale requires selling 9.5% paper to buyers whose required yield is BELOW 9.5%. For unrated/first-time-sponsor 10-25yr PIK second liens, required yields start above the coupon: sales happen at 93-97, i.e., LOSS-on-sale (C3). The platform has no revenue engine until after the rated curve exists. Break-even is not a deal count; it is a calendar date 4-6 years out, funded by equity the whole way. Severity: CRITICAL. U3. Loan-level economics are fine — and that is the ceiling, not the floor. Rebuilt holder IRR net of the 50bps strip, including financed points: 10.06% at a yr-7 exit, 9.67% at yr 12, 9.50% at yr 24 (3% HPA), i.e., the note rate plus ~50bps of points, minus tail losses. Portfolio-weighted across an exit distribution (mean exit ~18-22 yrs on a lock-in-selected book) and an HPA distribution centered at 3%: 9.2-9.5%, with a credit-loss drag of ~40-60bps/yr concentrated entirely in the long-hold/low-HPA tail (D1). There is no upside case: every good outcome returns 9.5%; every bad outcome returns less. The asset is a short option straddle on housing with a 9.5% cap. Severity: structural fact, priced in C-section findings. ## CAPITAL STRUCTURE (funding and exit — the load-bearing section) C1. The concept's own funding trust destroys its own claimed pricing edge (edge #3 vs cure of estate 6.2 — mutually exclusive). The sheet promises a bankruptcy-remote trust "holding committed capital for the full draw schedule." That is 126,000 of cash per deal at closing, deployed at 1,500/mo, undrawn balances earning ~4%. Rebuilt trust-level IRR: 6.9% at a yr-7 exit, 7.8% at yr 12, 8.2-8.5% at yr 18-24 — i.e., the prefunding commitment converts a 9.5% note into a ~7-8.5% vehicle. This is exactly my Equity Flow FX-5 finding: escrowing the draws deletes the staged-deployment advantage. The sheet claims edge #3 (ratable draws halve outstanding capital) and the consumer protection simultaneously. Pick one. If "committed capital" is softened to a warehouse commitment or LC, the homeowner is again exposed to a funding counterparty and the estate 6.2 cure is cosmetic. Severity: CRITICAL (it is the old design-fork shape: two advertised features, one balance sheet). C2. Warehouse-era carry is negative under realistic terms. Banks do not warehouse zero-cash-flow, unrated, novel-asset-class consumer seconds for multi-year aggregation periods; the realistic lender is a credit fund at SOFR+450-600 (9.5-10.5% today), 50-65% advance on DRAWN balance only (undrawn commitments get no credit), interest PIK'd or paid from equity. Rebuilt levered equity IRR, hold-to-exit, 3% HPA, origination cost included: 8.5% with a friendly bank line at 8.25%/60%; 6.1% at a 10% non-bank line. Equity in the bridge vehicle earns BELOW the note rate — leverage is negative or nil — while accruing phantom taxable income (O2) and bearing 100% of the novel-asset, regulatory, and extension risk. No rational LP funds that bridge at 9.5% note pricing. Severity: CRITICAL. C3. Takeout below par is annihilating; takeout above par is unearned. If the exit sale (whole-loan or securitization residual math) clears at 95-97 price at year 4, bridge equity IRR is minus 3.9% to plus 0.9%. Solving the stack backward: for bridge equity to earn 15% with a realistic (10%, 60%) warehouse and a 95 takeout, the note rate must exceed 15% — there is NO consumer-sellable rate at which the bridge works unless the takeout is at/above par. Par-plus takeout requires the market's required yield on seasoned rated paper to be at or below 9.5% — possible in the terminal state, unavailable in years 1-5 when the bridge needs it. The business is a bet that the takeout arrives before the bridge equity is exhausted. Severity: CRITICAL. C4. Even the terminal (rated) state has thin residual economics at a 9.5% note. Asset net yield: 9.5% + ~50bps points amortized - 50bps servicing - 40-60bps tail losses = ~8.5-9.0%. Optimistic rated execution (65% seniors at 6.75%, 15% mezz at 9.5%): residual to the 20% first-loss piece = ~13.9% — but as pure PIK, no cash for 10-20 years, first-loss against the correlated low-HPA tail. Realistic first-shelf execution for 12-18yr-WAL PIK bonds with 30-year extension (60% seniors at 7.5%, 15% mezz at 10.5%, 25% retained): residual = ~10.1% PIK. First-loss, zero-coupon, 20-year-duration residual paper needs high-teens; it gets 10-14 in accrual form. The sustaining note rate is therefore ~11-13% under realistic execution — against a consumer ceiling of maybe 10-10.5% next to an 8.5% HELOC (AS1). The two ranges do not overlap except under best-case rated execution that itself requires years of seasoning financed through C2/C3. Severity: CRITICAL — this is the kill arithmetic of the concept, and it survives every legal fix. C5. Interest-rate risk is unhedged and doubly convex against the sponsor. The asset is a fixed 9.5% accrual with 10-30yr uncertain duration. Rates up: funding cost rises, borrowers extend (lock-in deepens), PIK compounds toward the non-recourse cap, bond extension risk widens spreads. Rates down: the book prepays at par (return capped, reinvestment down) AND the origination funnel dies (M4). A first-time sponsor cannot hedge 20-year convexity on a PIK book from a venture balance sheet; nothing in the sheet prices it. Severity: MAJOR. C6. The Draw Limit is ambiguous at the one point where it matters. "Draws stop at the Draw Limit" — but the base-case sketch has the balance "hitting a 300K cap via accrual after year ~14," after draws already ended at month 84 near 180K. If ACCRUAL stops at the Limit, the holder's yield on the adversely-selected long tail collapses (a balance frozen at 300K from yr 14 to a yr-24 exit is roughly a 5.5-6% deal); if accrual continues, the Limit is decorative after the draw period and the consumer faces the 1.6M yr-30 number. One reading breaks the note economics on the stayers; the other breaks the consumer story. The spec must pick, and both picks cost. Severity: MAJOR (spec defect; must resolve before any modeling is meaningful). ## ADVERSE SELECTION AS1. The qualifying borrower's rational choice is the HELOC; the converting borrower is the one you don't want. Rebuilt matched-draw comparison (PV of borrower costs at a 4% personal discount rate): yr-7 exit — HomeDraw 43K vs HELOC 32K (+35%); yr-12 — 103K vs 68K (+52%); yr-20 — 224K vs 114K (+97%). The only thing HomeDraw sells that the HELOC doesn't is the absence of the payment (892/mo interest-only at full draw). So conversion self-selects for households that cannot service 892/mo — i.e., income-constrained borrowers whose senior mortgage is the next payment they miss. The clean-12-month box screens documents, not trajectories (prior AS1, intact). The book you actually originate is the ATR-problem book (O1) and the senior-delinquency-tail book (D2), by construction. Severity: CRITICAL as a joint demand/credit finding — the demand IS the adverse selection. AS2. Non-recourse plus 100% appreciation retention hands the borrower the long option. Stay forever: the cap eats the lender (D1). Prepay the moment cheaper money exists: the lender gets par on the shortest, best deals. The lock-in score maximizes selection of stayers — the exact cohort that rides the balance into the non-recourse cap and the 30-year maturity wall (D3). Same selection engine as Equity Flow, now pointed at the cap instead of the ratchet. Severity: MAJOR. AS3. Mobility misrepresentation is now mostly harmless — conceded. A borrower who secretly plans to sell in year 2 repays 9.5% paper at par; the lender is fine. The adverse selection that used to be about early exit is now entirely about LATE exit. (Noted as a genuine improvement.) ## DEFAULT / RECOVERY D1. The non-recourse cap converts the tail into designed principal loss. Rebuilt combined CLTV path at 3% HPA: 56% at closing, 62% at yr 7, 71% at yr 21, 95.6% at yr 28. First cap-breach year (payoff < balance on a clean sale): yr 14 at 0% HPA, yr 18 at 1%, yr 23 at 2%, yr 28 at 3%, never (through 30) at 4%+. Yr-30 recovery on the claim: 0.36x at 0% HPA, 0.49x at 1%, 0.65x at 2%, 0.88x at 3%. With 29-50% of a lock-in-selected book reaching the 30-yr maturity, expected loss is ~40-60bps/yr of IRR — tolerable on average, but perfectly correlated (one macro factor: HPA), landing entirely on the securitization residual/first-loss holder, and paired with OID already taxed on interest never collected (O2). A 9.5% cap with a 5%-severity-weighted left tail is money-good only in the rating agency base case — which is not the case they rate to. Severity: MAJOR (drives C4 subordination levels). D2. Senior-default channel: real but survivable — conceded as improved. At 56% combined CLTV, a senior foreclosure typically leaves junior surplus recovery of 30-70% after arrears and auction discount, versus ~0% for Equity Flow's non-lien. Bankruptcy: standard second-lien treatment; not strippable in Ch 13 while any collateral value supports it (origination CLTV <= 60% makes strip-off rare early) [SETTLED framework]. The legal-recovery architecture is genuinely fixed. Residual exposure: the AS1 book concentrates senior delinquency in recessions — same-factor correlation with D1. Severity: MINOR-MAJOR (cyclical). D3. The 30-year maturity wall breaks the product's own design rule (B: no forced sale). P(reaching maturity) on the modeled book: 29-50%. At maturity the borrower is 60-75 years old, owes 0.9-1.6M, is income-poor by selection (AS1), cannot refinance, and after the 12-month workout faces foreclosure by the lender. The product recreates the reverse-mortgage maturity-foreclosure headline for a younger cohort, at scale, on schedule, ~year 31 of the program — with the added feature that at 0-2% HPA the borrower has no equity left (CLTV > 100%) and simply loses the house the product was sold as protecting. "No forced sale" is true for 30 years and false at the exact moment the promise matters. Severity: MAJOR (consumer/reputational; also the legal hinge of O1). ## SECONDARY MARKET / FUNDING EXIT S1. The rated-securitization thesis is credible in kind, not in timeline — and the comps prove the gap. Honest comps: Point (founded 2015) reached its first securitization in 2021 (unrated, with Redwood) and rated HEI executions arrived 2023-24 (Unlock/DBRS et al.) — 6-8 years, with strategic-balance-sheet sponsorship. Figure did HELOC securitizations in ~2 years — but on CASH-FLOWING, fast-revolving collateral. Proprietary reverse securitizations (the closest structural comp: PIK, non-recourse, no payments until death/sale) are rated — off MORTALITY curves for 62+ borrowers. HomeDraw's under-62 collateral has no mortality anchor: exits are behavioral (mobility), the WAL band is 8-25 years, and the rating must run to a 30-year legal final under stressed HPA — where D1 shows the cap breaching. Expect: first rated deal year 4-6 at the earliest, low single-A/BBB senior caps, 30-40% credit enhancement, and senior yields that violate the C4 arithmetic in early shelves. Severity: CRITICAL as a timeline finding (the takeout arrives after the bridge equity is dead), even granting the thesis in principle. S2. Collateral mass builds glacially because the product IS a trickle. Pool balance grows at 1,500/mo/loan plus accrual. A 250-300M securitizable pool requires roughly 4,000-6,000 active loans plus 2-4 years of seasoning-by-accrual; at M3's realistic origination pace (low thousands/yr from a cold start, less in year 1-2) the first deal-sized pool exists in year 3-5. The staged-draw feature that makes the consumer economics work makes the securitization calendar hostage. Severity: MAJOR. S3. Committed future draws inside a securitization are a structuring trap. At any takeout date, most loans carry undrawn committed advances (a consumer-protection- critical obligation). Either (a) the trust prefunds remaining draws — restoring lump-sum capital outstanding and dragging trust yield to the C1 7-8% range inside the deal; or (b) the sponsor retains the draw obligation — and rating agencies will not give value to a committed consumer-facing advance obligation from an unrated first-time sponsor without a fully-funded reserve, which is (a) again. Figure-style HELOC deals don't solve this: their draws are borrower-optional; these are lender-committed. The claimed edge #3 dies at exactly the moment the claimed edge #4 activates. Severity: CRITICAL (the two halves of the funding thesis cancel). S4. The Schedule B insurance thesis is the one genuinely good funding idea — and it is unproven and demand-limited. A real mortgage loan with defined terms plausibly sits on Schedule B at RBC charges near 68bps (uninsured residential, in good standing) versus the 30%+ equity-bucket charge that killed the Equity Flow Tier 2 thesis [LIKELY INTERPRETATION — NAIC classification of a neg-am, deferred-pay, non-recourse consumer second as a "mortgage loan in good standing" needs a written SVO/state-examiner answer, not an assumption; interest admitted-asset treatment for decades-deferred accruals likewise]. Annuity writers can hold accrual assets and can absorb OID (O2). But: whole-loan buyers will demand seasoning, servicing continuity, and yield — a life insurer's par bid for novel 15-25yr PIK seconds will start at 10%+ all-in, i.e., above the sellable note rate, until performance exists. The thesis is an anchor-tenant question (see verdict condition 2), not a market. Severity: MAJOR (as currently evidenced: asserted, not demonstrated — same epistemic status as Equity Flow's Tier 2 story, with better bones). S5. If no exit materializes by year 6-8: the platform (U1/U2) has consumed 30-60M+ of equity subsidizing origination and warehouse carry; the bridge vehicle holds 7-8.5%-yielding (C1) phantom-income paper with 15-20yr remaining WAL; the only buyers are opportunistic credit funds at 85-92 — crystallizing losses on the equity that was told the note rate was the floor. This is the same terminal state as Equity Flow S5, arrived at more slowly and more honestly. Severity: CRITICAL (it is the base case absent condition 2 of the verdict). ## DEFENSIBILITY DF1. "First to rated takeout" is an anti-moat. Rating-agency criteria, once developed for sponsor #1, are published methodology — a public good. The pioneer pays 4-6 years of negative carry, legal spend, and criteria development; the fast follower (Figure, Rocket, a Longbridge/Ellington, a Unison pivoting its 2024 hybrid) arrives with existing shelves, servicing platforms, cheaper warehouse, and the published criteria, and prices through the pioneer's note rate immediately. In lending, the first mover in a new asset class historically funds the market's education and the incumbents' entry memo (my prior DF3, fully intact). Severity: MAJOR. DF2. The product is a contract, the funnel is a website, the data is commodity. No patent survives Alice for a draw schedule plus deferral [LIKELY INTERPRETATION]. The Noteworth lock-in funnel is replicable by any servicer or portal that already knows every borrower's rate, balance, and payment history (Zillow/Rocket/servicers hold strictly better targeting data — prior DF2, intact). Nothing in this concept is defensible except execution speed toward a takeout that, per DF1, transfers its value to competitors on arrival. Severity: MAJOR. DF3. The niche is real but contestable from three directions at once: HECM tenure holds 62+; Unison-style subsidized-rate hybrids attack the same under-62 borrower with a lower headline rate (trading appreciation share for coupon — a BETTER consumer story than 9.5% full-freight); and any rate decline hands the borrower a cash-out refi. The unoccupied space is "under-62, income-poor, equity-rich, rate-locked, neg-am-tolerant" — occupied by nobody partly because it is the ATR danger zone (O1). Severity: MAJOR. ## OTHER (legal/tax — tagged; not my lane for final opinions, but these gate the economics) O1. THE REG Z DEFINITION TRAP — the concept's central unresolved fork. Reg Z §1026.33 defines a reverse mortgage as a nonrecourse dwelling-secured obligation in which payment is due ONLY after death, transfer, or the consumer ceasing to occupy the dwelling as principal residence. Two of HomeDraw's proudest features break the definition: (a) the stated 30-year maturity makes payment due while the borrower may still occupy; (b) the no-occupancy-default feature (borrower may move out/rent with no consequence) means occupancy cessation is NOT a maturity event. Outside §1026.33, the ability-to-repay rule (§1026.43) applies to this closed-end dwelling-secured loan — and ATR cannot be satisfied by the collateral itself. "No income/DTI test" underwriting of a balloon obligation is then a Reg Z violation for precisely the house-rich/income-poor borrower the product targets. [LIKELY INTERPRETATION on classification; LIKELY on the ATR consequence.] The fork: drop the stated maturity to regain the reverse-mortgage shape (due at sale/death only arguably still fits "only after" the enumerated events [AMBIGUOUS]) — and lose the stated legal final the rating thesis (S1) depends on; or keep the maturity and underwrite income — and lose the borrower box. The securitization edge and the regulatory exemption are fighting over the same contract term. This is the successor to the old characterization fork, and it is unresolved on the face of the sheet ("to be confirmed... may actually help"). Severity: CRITICAL until resolved in writing. O2. Phantom income (OID) shapes — and shrinks — the holder universe. Deferred compounding interest on a debt instrument is OID accrued into taxable income annually with zero cash [SETTLED LAW — §§1272-1273]. Consequences: taxable funds and individuals pay tax for up to 30 years on cash they may partially never receive (D1 cap losses are recovered only as later bad-debt/loss deductions — a timing and character mismatch); leveraged tax-exempts convert the OID into UBTI via debt-financed-income rules [SETTLED framework]; a REIT wrapper must distribute 90% of phantom income with no cash [SETTLED], which is unworkable at scale. The natural holders are life insurers and foreign investors under the portfolio-interest exemption [LIKELY INTERPRETATION — registered form, non-bank, <10% holders]. That is a real but narrow base, and it is the SAME base the funding thesis already depends on — concentration, not diversification, of the exit. Severity: MAJOR. O3. Usury and compounding: 9.5% + 2.5 points on junior liens does NOT get DIDMCA §501 first-lien preemption [SETTLED — preemption is first-lien-limited]; state licensed-lender exemptions accommodate 9.5% in most states, but interest-on-interest for consumer mortgages requires express statutory/contract support in several [STATE-DEPENDENT]. Manageable with a state-by-state chart; unbudgeted. Severity: MINOR-MAJOR. O4. State functional reverse-mortgage statutes can capture the product regardless of the age gate (the Olson v. Unison logic runs stronger, not weaker, against an actual deferred-payment loan) — counseling, specific disclosures, and in some states annuity-purchase and vesting rules follow [STATE-DEPENDENT]. TX exclusion [SETTLED as to §50(a)(6)/(t) mechanics for this draw structure — correctly conceded by the sheet]. Severity: MAJOR in enacting states (cost + conversion friction), correctly flagged by the sheet. O5. UDAAP framing: a non-recourse, negative-amortizing, no-income-doc, balloon consumer mortgage is the assembled feature list Dodd-Frank was written against. Even fully compliant, the first recession cohort of D3 maturity-wall or cap-breach borrowers is a pre-written enforcement narrative. The honesty of the disclosure stack mitigates deception theories, not abusiveness theories ("takes unreasonable advantage of inability to protect interests" fits an income-less borrower compounding at 9.5% for 30 years) [AMBIGUOUS — genuinely]. Severity: MAJOR. --- ## VERDICT: FIX-BEFORE-LAUNCH — with embedded kill conditions, and a kill as a standalone venture at the stated note rate. Why not kill outright: the redesign genuinely resolves the prior product's fatal architecture. The instrument is legally coherent, the lender return is contractual rather than an HPA bet, expected credit losses (~40-60bps/yr) are absorbable within the note rate, and the closest structural comp (proprietary reverse securitization) exists and is rated. This is a real product that a real balance sheet could run. Why not acceptable: the capital stack does not close on the numbers as specified. At a 9.5% note: bridge equity earns 6-8.5% unlevered-to-negatively-levered (C1/C2), any below-par takeout destroys it (C3), and even clean rated execution leaves a 10-14% PIK-only first-loss residual that needs high-teens (C4). The sustaining note rate (11-13%) exceeds the consumer-sellable ceiling (~10-10.5% against an 8.5% HELOC), and the borrower who converts anyway is the ATR-violation borrower (O1/AS1). The claimed edges cancel pairwise: staged deployment vs the funding trust (C1); staged deployment vs securitization prefunding (S3); stated legal final vs the reverse-mortgage exemption (O1); first-to-rated-takeout vs published-criteria free-riding (DF1). And the concept sheet's own base-case arithmetic is ~12% flattering (M1), which this panel has seen before. As a venture-funded standalone: KILL — the bridge consumes 30-60M+ of equity over 4-6 years to build an asset class whose economics transfer to cheaper-capital incumbents at the moment of proof. As a program inside an incumbent with an existing shelf, servicer, and warehouse (or with the conditions below): fix-before-launch. THE THREE THINGS THAT MUST BE TRUE (each in writing, before a dollar of origination; failure of any one converts this verdict to KILL): 1. THE REG Z FORK IS RESOLVED WITHOUT KILLING EITHER THE BOX OR THE RATING. Written counsel opinion plus rating-agency pre-engagement feedback establishing EITHER (a) a §1026.33-conforming structure (no stated maturity; due at sale/death) that at least one agency will still rate to an actuarial/behavioral final for under-62 collateral, OR (b) an ATR-compliant underwriting method (non-collateral asset-based) that demonstrably still contains >= 50% of the target borrower volume. If neither exists, the product has no lawful buyer-of-size and no lawful borrower-of-need simultaneously — the old fork in new clothes. 2. AN ANCHOR TAKEOUT EXISTS BEFORE SCALE, NOT AFTER. A committed forward-flow whole-loan agreement (or equivalent) from at least one investment-grade life insurer for >= 200-250M at par or better with note economics <= 9.75%, including written confirmation of Schedule B / mortgage-loan RBC and admitted-asset treatment for the deferred accrual. This converts the C2/C3 bridge from a 4-6-year negative-carry bet into a funded pipeline, and it is the only version of the funding thesis that does not depend on a rating timeline the comps say you will miss. 3. THE CONSUMER ACCEPTS THE SUSTAINING RATE WITH THE HONEST TABLE IN THE FLOW. A funnel test (real applicants, real statutory-format 5/10/20/30-year dollar table, corrected arithmetic per M1) demonstrating economically viable conversion at a note rate >= 10.5% among borrowers who PASS the condition-1 underwriting screen. If conversion only exists at 9.5% or only among ATR-failing borrowers, the demand is either unfinanceable (C4) or unlawful (O1), and the honest version of this product — like the honest version of the last one — has no customers. One sentence for the parent process: Equity Flow died because it was a wager dressed as a product; HomeDraw is a product dressed as a business — the product is real, and the business, as capitalized and priced in this sheet, is not. --- end of review ---