Data room / Run 1, stage 2: the nine cross-examinations
Red team
run-1/reviews/crossexam-red-team.md Open the file as written
EQUITY FLOW — STAGE 2 CROSS-EXAMINATION: RED TEAM REVIEWER
Reviewer role: adversarial red team (fraud adversary / competitor / hostile journalist-regulator), cross-examining the eight Stage 1 panel reports against my own Stage 1 review (reviews/red-team.md) and the spec (disclosures/equity-flow-invention-disclosure.txt).
Scope per Stage 2 mandate: (1) disputes with other reviewers' findings and severity calls; (2) new attack surfaces created by other reviewers' proposed fixes; (3) explicit flagging of direct conflicts between any two reviewers. Claim tags: SETTLED LAW / LIKELY INTERPRETATION / STATE-DEPENDENT / AMBIGUOUS / SPECULATIVE. New legal/tax claims are tagged; claims already tagged in Stage 1 reports are cited, not re-tagged.
Standing note: I do NOT dispute the panel's core findings (tax P-1/D-1, estate 4.1, title 1.2/5.1, sec-cf 6.A, investor S1-S5, ops 1.3/1.4, actuarial 1.1-1.4). They held up under cross-examination and my Stage 1 verdict (KILL) is built on them. The disputes below are targeted: places where a reviewer overstated an attack surface a real adversary would not use, understated one an adversary WOULD use, modeled the friendly case, or reached a verdict their own findings contradict.
================================================================================
SECTION 1 — DISPUTES
================================================================================
D1. Retail V1/V2 — the "large fraction of enrollees land in the 12.8%+
APR cells" likelihood claim is overstated; the real surface is bimodal.
Cite: retail-customer.md V1 ("The average American moves every 8-10 years; a large fraction of enrollees will land in the 12.8%+ cells"), V2. Dispute: the national 8-10-year mover statistic does not describe this book. The product's own underwriting (spec Sec. 23, 39(h)) selects for deep rate locks and high holding propensity, and actuarial 4.1 models the selected book at ~2-3.5%/yr mobility with a majority of positions still outstanding at year 20+. On that book, the mid-tenure 7-12-year voluntary exits that populate retail's most expensive cells are the LEAST likely outcome. What actually populates the book, from an adversary's chair, is the two tails: (a) concealed near-term sellers exploiting the sanction-proof non-binding intent representation (my F6) — who exit in flat markets at near-zero cost to themselves, and (b) permanent stayers whose APR-equivalents are the 5.9-8.9% long-hold cells and whose harm is the ratchet/estate machinery, not the headline APR. Retail's 23-35% cells are real and remain first-rate attack-ad material (I use them in C1/H-series), but as a PREVALENCE claim they contradict the selection math. Severity of retail V1 as a consumer-harm finding: stands. Likelihood: downgrade from implied HIGH to MEDIUM for the mid-tenure cells specifically. Flagged as direct conflict DC-3.
D2. Tax A-4 / my own H3 — the mass consumer-audit scenario rests on the
branch the tax reviewer himself ranks second.
Cite: tax-attorney.md A-4 ("deficiency notices go to individual homeowners ... thousands of consumer audits"), P-3 (loan recharacterization is "the most probable" outcome; ordinary-income/NPC treatment is the AMBIGUOUS alternative). Dispute (and self-correction): under the tax reviewer's own most-probable characterization (SAM/loan per Rev. Rul. 83-51), monthly receipts are nontaxable loan proceeds — homeowners get NO deficiency notices on receipt; the detonation lands on the platform and fund (TILA, licensing, usury, phantom CPDI income to LPs) and on homeowners only at exit (nondeductible interest, D-4(b)). The "IRS tells retirees their payments were income all along" scenario (tax A-4; my Stage 1 headline H3, which I tagged MEDIUM-HIGH) requires the ordinary-income branch that tax itself tags AMBIGUOUS and ranks behind the loan branch. A-4's framing presents the lower-probability branch as the headline failure mode. Corrected posture: H3 likelihood downgraded to MEDIUM; the higher-probability tax catastrophe is institutional (fund/LP/platform), which is worse for the sponsor and better for the homeowner than either of us wrote. The consumer-facing tax harm that survives on the likely branch is H-1's whipsaw (tax on gain never kept) — which is exit-time, individually litigated, and slower-burning. Tag on the corrected sequencing: LIKELY INTERPRETATION.
D3. Sec-cf verdict premise — "materially similar characterization risk"
to incumbents is a misreading; the fix-before-launch verdict on the core
contract is built on it.
Cite: securities-consumer-finance.md Verdict para. 1 ("Lump-sum HEIs ... operate today at scale under materially similar characterization risk; the 1.0x no-premium structure genuinely improves the equity characterization versus incumbents"). Dispute: both clauses are contradicted by the tax and title reports on this record. Tax A-2: Equity Flow eliminated the two facts (single lump-sum prepayment; genuine investor downside in home VALUE) on which incumbent comfort opinions rest — it is strictly WORSE positioned, not similarly positioned. Tax D-1: the 1.0x senior capital return is the single most debt-like feature in the structure ("the confession"), and title's central defect notes the 1.0x floor fails the CFPB Roberts loss-sharing test MORE cleanly than the products the CFPB criticized. Sec-cf's own Sec. 3.1 factors list says the same thing; the verdict paragraph then credits the 1.0x as an equity-characterization improvement. The "coherent (if costly) strategy" of excluding CT/MD/WA and keeping a TILA fallback stack does not survive the monthly-advance format sec-cf itself flags as worse (3.1) plus the functional reverse-mortgage statutes it flags as under-analyzed (Sec. 4). A competitor or AG needs to win once; the launch-footprint strategy requires the sponsor to win everywhere it operates, forever. Severity call disputed: the core-contract verdict should be KILL on sec-cf's own findings, not fix-before-launch. Flagged as direct conflict DC-1.
D4. Title 6.2 — "SPECULATIVE / probability low" on a due-on-sale
acceleration wave understates an adversary-endogenous risk.
Cite: real-estate-title.md 6.2 caveat 2 ("Probability low; severity maximal"). Dispute: the probability is treated as exogenous (would servicers happen to notice?). It is endogenous: every holder of a 2.5%-below-market first lien has maximal economic incentive to find an acceleration theory (title's own observation), and incumbents/competitors are natural instigators of the triggering event — circulating the spec's own "prepaid forward SALE of a partial interest" language (which title concedes takes the instrument outside the (d)(1) lien carve-out) to GSE/FHFA servicing-policy desks costs a trade association nothing (my C5). A single servicing bulletin is a de facto national injunction. Adjusted: likelihood MEDIUM (not low) within the first years of visible recorded volume; the SPECULATIVE tag should attach only to the spontaneous-wave variant, not the provoked one. Tag on the provoked-bulletin mechanism: LIKELY INTERPRETATION that GSEs/underwriters issue guidance once petitioned; STATE-DEPENDENT in effect.
D5. Actuarial 1.1 / 2.3 / verdict — the "escrow-first 1.0x recovery is
genuinely robust" conclusion models the friendly case and omits the
legal-defeat channels; Equity Shield underfunding is therefore
UNDERSTATED.
Cite: insurance-actuarial.md 1.1 ("the only material loss mechanism for the capital return is foreclosure wipeout"; distressed sales "almost never breach the floor because of the 20%+ equity cushion"), 2.3 and Verdict ("its escrow-first 1.0x recovery is genuinely robust in sale exits even at -45%"). Dispute, three ways. (a) The loss model is foreclosure-only. It excludes the channels the legal panel established: Chapter 7 discharge of the non-lien claim (estate 4.1 — SETTLED LAW), strong-arm avoidance (estate 7.2), Florida homestead descending to heirs free of the claim (title 5.2), Texas void-encumbrance (title 5.1), nonclaim-barred probate claims (estate 1.1), and related-party exit manipulation with no detection function (my F4, ops 4.3). Every one of those is a 100%-severity floor breach that is NOT a foreclosure and NOT captured in the 3%/6%/12% foreclosure rows. In a recession vintage, consumer bankruptcy filings dwarf completed foreclosures; on this book the bankruptcy channel is the larger tail. (b) The equity-cushion arithmetic uses the spec's 56%-LTV example, but the eligibility box permits 80% post-deployment CLTV (Sec. 39(g)); at the boundary, the modeled -26.5% national path breaches sale recoveries too. (c) The cushion assumes honest enrollment values; systematic appraisal inflation at origination (my F3 — every party at the table benefits from a high number) erodes it book-wide. Consequence: actuarial's already-fatal 0.47x-0.95x reserve coverage is an UPPER bound, and the verdict sentence crediting the core contract's robustness is wrong on the panel's own record. Severity: the correction strengthens actuarial's kill on Equity Shield and removes its stated reason for sparing the core contract. Flagged as direct conflict DC-4.
D6. Investor AS4 — severity MAJOR is understated; the pause moral
hazard is CRITICAL because it is the doom loop's escape hatch.
Cite: investor.md AS4 (MAJOR); C2 (doom loop, CRITICAL); ops 1.4. Dispute: AS4 prices the pause asymmetry as an economics-plus-reputation problem. Combined with C2 and ops 5.2 — as my F1/M1 laid out — the pause is the mechanism by which a distressed fund lawfully avoids ever triggering the three-missed-payment forfeiture: it never defaults, it pauses, using an AVM it selects, against a dispute mechanic that taxes the homeowner ~$550 to appeal. That converts the homeowner's ONLY counterparty-credit protection into a dead letter and makes fund distress invisible until total. A finding that neutralizes the contract's sole credit protection for one side is CRITICAL by any consistent scale (investor rated the doom loop itself CRITICAL; the escape hatch that makes the doom loop survivable-by- fraud inherits that severity). Same upgrade applies to ops 1.4's framing: ops models the pause as accidental AVM noise; the deliberate-abuse channel is the graver surface and calibration fixes do not touch it (see FX-9 and DC-12).
D7. Sec-cf 3.5 — the arbitration-clause comfort ("enforceable
generally") understates the California hole that exists WITHOUT
recharacterization.
Cite: securities-consumer-finance.md 3.5 and checklist row ("Enforceable generally (Concepcion) but void if dwelling-secured loan"). Dispute: McGill v. Citibank, 2 Cal. 5th 945 (2017) — public injunctive relief cannot be waived by contract (SETTLED LAW in California; survived FAA preemption challenge in Blair v. Rent-A-Center, 9th Cir. 2019 — LIKELY INTERPRETATION federally). Sec-cf's own deception catalog (6.A) is the ready-made predicate for a public-injunction claim litigated in open court in the largest target state, years before any recharacterization ruling. The clause's failure mode is therefore not contingent on winning the credit-characterization fight; the waiver already leaks in the market that matters most. My M5 flagged the omission; the cross-exam point is that sec-cf's summary-table row affirmatively overstates the protection.
D8. Ops 4.3 — the 2-4% loss/enforcement-event rate is understated
because it models idiosyncratic events, and the adversary dynamics are
correlated.
Cite: product-operator.md 4.3. Dispute: 2-4% over 10 years is a reasonable static estimate for uncoached counterparties. It ignores contagion: (a) the bankruptcy-discharge playbook (estate 4.1 + my F2/M4) industrializes after the first published win — consumer bankruptcy is an advertised, precedent-driven bar, and the funnel pre-selects cash-poor households; (b) concealed near-term sale is sanction-proof by the spec's own drafting (F6), so in any flat vintage it is the rational default, not a tail; (c) occupancy fraud has no penalty spread (F7 — detection produces the same settlement the homeowner would get anyway). Once any of these publishes, exploitation is correlated across the book, and a static 2-4% becomes a floor with a fat right tail in exactly the vintages where exits cluster. Adjusted: 2-4% base decade, scenario-conditional 10%+ after a published roadmap. Severity: upgrade from MAJOR to CRITICAL conditional-on-contagion.
D9. Ops verdict framing — "monthly deployment limiting capital at risk
... genuinely better-designed than the incumbent HEI mechanics" is
contradicted by ops' own findings and the rest of the panel.
Cite: product-operator.md Verdict preamble; spec Sec. 4(b). Dispute: the design credit adopts the sponsor's framing. Monthly deployment is precisely what creates: the binding 84-month forward obligation and deepest-in-consumer-finance J-curve (investor C1), the doom-loop forfeiture fuse (investor C2), the regulatory-event trap ops itself rated CRITICAL (5.2 — each monthly ACH is arguably a new unlicensed advance), the servicing machine ops priced at zero (1.3), and the primary tax defect (tax P-2: serial variable consideration). "Limits investor capital at risk" is true deal-by-deal and false fund-wide — the fund's committed exposure is identical, plus a default fuse the lump-sum incumbents do not carry. An operator's verdict that the core mechanics are "genuinely better-designed" while rating three of their direct consequences CRITICAL is internally inconsistent; the mechanics are the defect. Flagged as direct conflict DC-5.
D10. Estate 7.4(c) — the SPECULATIVE tag on ratchet-increment avoidance
understates the practical exposure, which does not require winning.
Cite: estate-bankruptcy.md 7.4(c) ("SPECULATIVE as to outcome"). Dispute: the tag is correct as to OUTCOME and misleading as to EXPOSURE. The avoidance argument (each +1% ride-along step is a transfer of the debtor's property for zero new consideration within the 548/UVTA windows) costs a debtor's attorney one paragraph to plead in a case they are filing anyway, and the expected settlement value is positive at near-zero cost. Adversary economics, not merits, drive frequency: expect it as standard-issue leverage in every ride-along-phase consumer bankruptcy the moment a form book includes it. Exposure severity: MAJOR (portfolio friction and settlement leakage on every long-hold bankruptcy), even if no court ever sustains it. The same pleading-cost asymmetry applies to estate 1.3's post-death ratchet accrual in every contested probate.
================================================================================
SECTION 2 — FIXES THAT CREATE NEW ATTACK SURFACES
================================================================================
This is the core cross-exam finding: the panel's repair prescriptions are not neutral. Each material fix hands at least one of my three adversaries a new tool. Stage 1 established the fixes conflict across domains; Stage 2 adds: even where a fix is domain-internally sound, an adversary exploits the fixed product.
FX-1. True-lien recording (performance deed of trust) — title verdict
fix 1; estate required fix 1.
What the fix does for me:
- Regulator/journalist: a recorded deed of trust securing monthly tenure-style advances to an owner-occupant, repaid at sale or death, is a photograph of a reverse mortgage under functional-definition state statutes — Olson stops being an analogy (STATE-DEPENDENT, trending adverse). It also concedes the "security interest in a dwelling" element of TILA/Reg Z, HOEPA, and the 1639c(e) arbitration ban outright — the only remaining fight is "is there debt," which tax D-1 answers against the sponsor. Arbitration clause void if the debt fight is lost: SETTLED LAW contingent. The before/after marketing ("not a lien, not a mortgage" retracted and replaced by a recorded mortgage) is itself the competitor's ad and the AG's exhibit.
- Tax: this is estate's own stated kill condition — the fix detonates Sections 24-25. Also mortgage-recording/doc-stamp taxes become unambiguously due (tax S-2 follow-on), computed on an indeterminate secured amount; several states force a STATED MAXIMUM secured amount — a new, large, scary disclosed number on every consumer's title. STATE-DEPENDENT.
- Fraud (new): future-advance priority. A deed of trust securing 84 months of FUTURE advances raises the optional-vs-obligatory advance doctrine: in many states, optional future advances made after notice of an intervening lien are junior to it — and the market-pause conditionality makes the advances look optional. A homeowner grants a HELOC in month 13; the remaining 71 months of Equity Flow advances may sit behind it. STATE-DEPENDENT (future-advance mortgage mechanics SETTLED LAW; priority of conditional advances varies). Nobody on the panel priced this because nobody war-gamed the fixed instrument.
- Bankruptcy whipsaw persists in the other direction: a true junior lien that is underwater (the -15% recession vintage, 80% CLTV originations) is wholly unsecured and STRIPPABLE in Chapter 13 (SETTLED LAW, majority rule; estate 5.2). The fix trades estate 4.1's discharge wipeout for strip-off wipeout in the same vintages where filings cluster. Chapter 7 genuinely improves (Caulkett — SETTLED LAW).
- Assignee weaponization (my M2) upgrades: the least scrupulous downstream holder now holds foreclosure machinery. The dementia-occupancy headline (H2) becomes a nonjudicial foreclosure story instead of a court petition.
FX-2. Spousal/all-titleholder joinder — estate fix 5; title fix 6.
Necessary (the current screen originates void instruments), but:
- Fraud: forged or impersonated spouse execution is the classic HELOC fraud; a joinder REQUIREMENT creates a void-instrument attack surface at scale if execution is remote/notary-shopped — every position now has two signatures to attack instead of one. STATE-DEPENDENT (RON standards).
- Journalist/regulator: the non-borrowing 80-year-old spouse co-signing an instrument the investor page prices off mortality tables; capacity and undue-influence claims double; in CA the co-signing spouse is a second 15610.30 elder-abuse plaintiff. STATE-DEPENDENT.
- Divorce: with both spouses bound, the non-joinder voidability escape closes for homeowners while the investor's consent-veto over decree-ordered substitution (estate 3.4(b)) now operates against two bound parties — the family-court collision worsens.
FX-3. Hardship provisions / medical occupancy carve-out / hardship desk
— ops 3.1; estate fix 4.
- Fraud: a documentation-based medical/hardship exception is an occupancy- fraud amplifier (F7): fabricated hardship letters extend absentee occupancy indefinitely while the home runs as an STR, against a monitoring function priced at zero (ops 1.3). "Acceptable explanation" standards are unauditable.
- Recharacterization: any hardship ADVANCE against future recovery is itself an extension of credit secured by the position — the hardship feature is a loan even if the base product wins its fight. LIKELY INTERPRETATION.
- Regulator: a discretionary grant/deny desk creates servicing-side disparate-treatment/disparate-impact exposure (FHA 3605 reach per sec-cf 3.4 logic extends to servicing decisions — LIKELY INTERPRETATION), and denial logs are discoverable. Journalist: the first denied dementia case under a marketed hardship program ("they had a hardship desk and used it to say no") is a worse headline than the absence of one.
FX-4. Escrowed committed-amount payments — spec Sec. 24's own
alternative; estate fix 6 (escrow/LC-backed commitment).
- Tax boomerang: the current defense against day-one taxation is that the homeowner cannot demand the undisbursed balance and the pause is a substantial limitation (tax P-4). Funds SET APART in escrow for the homeowner attack exactly that element (Treas. Reg. 1.451-2 — framework SETTLED LAW; constructive-receipt application to a funded irrevocable escrow: AMBIGUOUS, adverse-trending). The consumer-protection fix undermines the tax timing defense.
- If instead the escrow stays investor-controlled to dodge constructive receipt, it fails its purpose: fund creditors reach it in fund bankruptcy and the homeowner is still unsecured (estate 6.2 unsolved). The fix protects the homeowner only in the characterization that hurts the homeowner at tax time. Pick one.
- Economics/competitor: day-one full funding recreates the lump-sum J-curve, erasing the marketed "limited capital at risk" rationale (spec Sec. 4(b)) — the fixed product converges with the incumbent escrow-wrapper (my C2), i.e., the fix is a confession that the competitor's version is the correct design, minus the tax defect.
- AML: pre-funded escrows converting lump sums into clean monthly ACH are a cleaner laundering rail than the current structure (F10 amplifier), and the escrow agent is a new insider-fraud locus.
FX-5. Licensed-carrier Equity Shield — sec-cf gating item (vi);
actuarial fix (i) alternative.
- The honest premium IS the kill: a licensed carrier prices the correlated tail at the 6-12%-of-deployed capital-equivalent actuarial computed; fund economics cannot pay it. More dangerous: every carrier DECLINATION becomes discoverable evidence that the sponsor knew the risk was commercially uninsurable ("we shopped the guarantee; no carrier would write it") — a scienter gift in any later floor-shortfall litigation.
- The carrier's underwriting imports the entire diligence file into a third party's records, outside privilege.
- Regulatory: marketing a wrapper engineered to deliver "favorable regulatory capital treatment" (spec Sec. 17) invites NAIC/SVO scrutiny of the whole asset class in a period actively hostile to structured capital-relief wrappers — the fix summons the very regulator actuarial 2.3 warned the product currently avoids.
- Competitor: only insurer-affiliated incumbents can execute this fix cheaply; it converts the differentiator into an incumbents-only feature.
- Journalist: the investors get an insurance policy; the homeowners' counterparty risk (H4 — checks stop, encumbrance stays) remains naked. "They insured Wall Street's downside, not grandma's" writes itself.
FX-6. Hold-to-maturity / evergreen 20-30-year fund with honest 6-8% net
marketing — investor verdict; actuarial fix (ii).
Honest, and it maximizes three of my Stage 1 fraud surfaces:
- M2 maximized: a manager locked in for 25 years whose ONLY IRR lever is homeowner exit timing becomes the covenant-enforcement adversary by design — occupancy technicalities, consent denials, probate petitions are its alpha. "Duration is the feature" marketing versus a holder economically desperate to shorten duration.
- F11 maximized: decades with no exits, sponsor-controlled marks, and periodic fundraises is the maximal wash-trade/validation-mark incentive environment.
- F1 maximized: the zombie evergreen fund of year 15 still holds the pause switch, now with no exit pressure relieving it.
- Orphan-book risk (ops 6.2) becomes the base case: the fix presumes a servicer, registry, and manager that outlive a quarter century, none of which is funded (ops 1.3).
- And because the honest version cannot raise (investor B5; actuarial 3.3), the realistic outcome of mandating honest docs is honest docs plus optimistic side letters — a securities-fraud shape, not a fix.
FX-7. Honest APR-equivalent disclosure and contract-accurate
pause/ratchet/fee disclosure — retail verdict minimum bar; sec-cf
gating items (ii)/(viii).
- Evidentiary boomerang: a sponsor-published APR-equivalent concedes that a finance charge exists in substance; usury and recharacterization litigation gains a defendant-authored number to test against caps. LIKELY INTERPRETATION (evidentiary weight, not dispositive).
- The abusiveness irony: disclosure cures the DECEPTION prong and sharpens the ABUSIVENESS prong. A disclosed 23-35% worst-case cost sold through a funnel that pre-selects people who cannot refinance is, on paper, "taking unreasonable advantage of the consumer's inability to protect their interests" — no deception element required. Retail's own verdict concedes honest disclosure collapses the base to those with no alternatives; a regulator reads the post-disclosure funnel data the same way. LIKELY INTERPRETATION.
- Competitor: the mandated disclosure IS the attack ad (HELOC ~8.5% flat vs. "up to 34.7%"). No copywriting survives it.
- Convergence trap: emerging state HEI statutes require APR-equivalent tables; complying voluntarily walks the product into the regulated bucket it claims to sit outside.
FX-8. Minimum-term early-exit adjustment (higher share on exit before
year 3) — floated in investor-page FAQ; flagged unsupported by sec-cf
B11; implicitly endorsed as an adverse-selection mitigant by investor
AS2.
Do not implement. It is the single worst fix on the table:
- Substance: a prepayment penalty. If the product is recharacterized, Reg Z/HOEPA prepayment-penalty restrictions apply directly (SETTLED LAW, contingent); pre-recharacterization it is fresh evidence of lending economics. LIKELY INTERPRETATION.
- It misses the fraudster entirely: the concealed-sale exploit (F6) operates in FLAT markets, where a "slightly higher appreciation share" is a higher share of approximately zero. The surcharge binds only on honest early exits in hot markets — which are death, divorce, and job-loss sales. The fix taxes widows and misses adversaries.
- Journalist: "Early-exit surcharge on a widow's home sale" — while the homeowner page still says "settle whenever life settles" and "no exit fees." Implementing B11 converts a marketing inconsistency into a contract term with UDAAP exposure attached.
FX-9. Pause governance repairs (buffer, hysteresis, two-AVM cascade,
platform-paid dispute appraisals) — ops 1.4 fixes; my M1 partially.
I proposed some of this, so cross-examining myself with the panel's versions: the ops fix set treats the pause as a calibration problem. It is a CONTROL problem.
- Relocation, not removal: whoever selects the two AVM vendors and the band parameters still owns the switch (F9); a cascade the fund configures is still shoppable in distress (F1). Without an independent valuation agent with contractual standing — which nobody budgets (ops 1.3) — hysteresis is governance theater.
- Platform-paid dispute appraisals create a dispute-mill channel: free disputes plus volume-paid appraisers is the F3 collusion economics reversed into the dispute lane. Manageable, but it must be designed as a fraud surface, not a goodwill feature.
- The buffer has a fund-side cost nobody netted: paying into confirmed declines worsens the negative convexity investor C3 identified, which INCREASES a distressed fund's appetite for the other lawful stop switch — see FX-10.
FX-10. Regulatory-event suspension clause — ops 5.2 minimum fix.
Ops flagged the disclosure problem; the adversarial reading is worse:
- A consumer contract clause suspending payments upon regulatory action is a written admission of characterization risk, quotable in every proceeding and every article ("even the contract says the payments stop if a regulator decides it is a loan").
- It hands the investor a SECOND lawful payment kill-switch with a trigger the investor side can help manufacture: a distressed fund (F1) benefits from inviting or amplifying regulatory noise, citing an inquiry, and suspending book-wide with tolling. The clause must therefore be drafted with a homeowner-protective trigger standard (final agency action, not inquiry) — no panel report specifies this.
- Marketing becomes unwritable: "binding obligation" copy cannot coexist with two disclosed suspension switches (AVM print; regulator letter).
FX-11. Payoff/estoppel machinery, servicer-of-record with recorded POA,
MERS-like registry — title fixes 3-4; ops 3.4/6.2 fixes.
- Fraud: a RECORDED limited power of attorney to execute subordinations and releases, held by a thinly capitalized servicer, is a high-value instrument-fraud target — one corrupt or spoofed servicer signature produces facially valid releases/subordinations, and downstream bona-fide-purchaser protection makes the loss stick against the fund. STATE-DEPENDENT (BFP protection). This is how title-fraud rings already attack HELOC releases; the fix centralizes the target.
- Deadline-plus-penalty estoppel regimes invite homeowner-side gaming: request floods timed against a skeleton servicer, manufactured missed 10-day deadlines, captured claim haircuts.
- Journalist: the MERS analogy arrives pre-loaded ("the industry that brought you robo-signing now runs a private registry of home-equity claims"); registry standing challenges are STATE-DEPENDENT and were litigated for a decade last time.
- Competitor: a single servicer-of-record everyone depends on is a chokepoint — acquirable, poachable, or starvable.
FX-12. State exclusions (TX; FL homestead restructuring; CT/MD/IL/ME/WA
licensing) — title fix 2; estate fix 5; sec-cf launch-footprint
strategy.
- The exclusion of TX and FL removes roughly a quarter of the target inventory including the sizing tool's own promoted metros, shrinking investor M1's already-shrunk SAM and concentrating the book in California — which is simultaneously the McGill state (D7), the elder-abuse-statute state (H2), the highest-state-tax state (tax S-1), the insurance-crisis state (my M8), and one housing factor (actuarial 1.3's correlation problem worsens with concentration).
- The exclusion list is an admission map: each excluded state's rationale is a memo other states' regulators can adopt by reference, and a competitor talking point ("effectively barred in seven states including the two largest").
FX-13. Fee and basis repairs (capital-return basis = net cash, ops 1.2;
origination fee re-based to home value, ops 1.1).
- Basis-net-of-fee fixes a deception finding (retail T6) at the cost of fund return — fine, but note it and re-cut the investor deck or the 10b-5 problem migrates.
- Fee-on-home-value (2.5% of the 625K example is 15,625 vs. 3,150) quintuples the consumer-borne cost, walks straight into points-and-fees and HOEPA math if recharacterized (SETTLED LAW frameworks, contingent), and makes retail's APR-equivalent table materially worse. The origination-P&L fix and the consumer-cost findings are mutually antagonistic — see DC-8. There is no fee level that satisfies ops 1.1 and retail V1 simultaneously.
Fix-stack conclusion.
Apply the full panel fix set — lien-recorded, all-spouse-joined, hardship-desked, escrow-funded, carrier-wrapped, hold-to-maturity, APR- disclosed, state-limited, fee-repriced — and the surviving artifact is a licensed, disclosed, mortgage-shaped, small-footprint, long-duration product: precisely the incumbent lump-sum-with-monthly-draw product of my C2, delivered later, at higher cost of capital, by a first-time sponsor with a discoverable file. The fixes do not rescue the product; they complete the competitor's argument. Stage 2 verdict unchanged: KILL.
================================================================================
SECTION 3 — DIRECT CONFLICTS (all reviewer pairs)
================================================================================
DC-1. SEC-CF vs TAX on the 1.0x capital return. Sec-cf (Verdict para. 1) says the 1.0x no-premium structure "genuinely improves the equity characterization versus incumbents." Tax (D-1, A-2) says the 1.0x senior capital return is "the confession" — the single most debt-like feature — and leaves the product strictly worse-positioned than incumbents. Incompatible because the same term cannot be both the best pro-equity fact and the dispositive anti-equity fact; a characterization opinion cannot be written until the sponsor knows which reviewer is right. (Title's central defect sides with tax.)
DC-2. TITLE/ESTATE vs TAX on the mandatory fix. Title (verdict fix 1, "Pick the lien") and estate (required fix 1) make a recorded true security instrument the load-bearing repair; tax's verdict makes non-lien, non-security economics a condition of any surviving equity characterization, and estate's own kill condition concedes the collision. Incompatible because the fix required for bankruptcy/probate/title survival is the kill condition for tax/regulatory survival: the two verdicts of "fix-before-launch" (title, estate) and "kill" (tax) are the SAME finding with opposite labels.
DC-3. RETAIL vs ACTUARIAL on who is in the book. Retail (V1) says a large fraction of enrollees land in 7-12-year exits at 12.8%+ APR-equivalents, citing 8-10-year average US mobility. Actuarial (4.1) says the lock-in-selected book runs 2-3.5%/yr mobility with the majority of positions outstanding past year 20. Incompatible because the same selected population cannot be predominantly mid-tenure movers and predominantly 20-year holders; each report's headline harm scenario assumes a book composition the other report's model excludes.
DC-4. ACTUARIAL vs ESTATE/TITLE on whether the 1.0x recovery is robust. Actuarial (1.1, 2.3, verdict) models floor loss as foreclosure-only and calls escrow-first recovery "genuinely robust in sale exits even at -45%." Estate (4.1, 1.1) and title (5.1, 5.2) establish that the claim is dischargeable in Chapter 7, barrable by probate nonclaim statutes, void against Texas homestead, and defeated by Florida homestead descent. Incompatible because a recovery cannot be actuarially robust and legally voidable in the two largest homestead states plus every consumer bankruptcy; the actuarial loss model and the legal reports price different products.
DC-5. OPS vs INVESTOR/TAX on the monthly-disbursement mechanic. Ops (verdict preamble) credits monthly deployment as "genuinely better-designed" risk limitation. Investor (C1, C2) identifies the binding 84-month forward commitment as the deepest J-curve in consumer finance and the doom-loop trigger; tax (P-2, verdict para. 2) identifies the monthly cadence as one of the two features that destroy the tax characterization. Incompatible because the same design element is rated the product's chief virtue by one reviewer and its primary structural defect by two others — and ops' own findings 1.3/5.2 side with the critics.
DC-6. SEC-CF vs ESTATE on the class waiver. Sec-cf (3.5, checklist) says the arbitration clause and class waiver are "enforceable generally" absent recharacterization. Estate (6.2) says the class waiver "may not survive in consumer-finance form in several states" and does not bind AGs. Incompatible in emphasis and in output: sec-cf's summary table offers comfort that estate's fund-insolvency scenario expressly denies — and neither addresses the California public-injunction hole (D7), which resolves the conflict against comfort.
DC-7. SEC-CF vs TITLE on enforcement conventionality. Sec-cf (verdict para. 1) calls recorded-encumbrance enforcement "conventional." Title (1.1, 2.1, 7.2-adjacent via estate) finds the memorandum may be OUTSIDE some state recording acts (no constructive notice), that no title company will administer the settlement formula, and that no payoff-demand machinery exists. Incompatible because an instrument cannot be conventionally enforceable and sui generis unadministrable; sec-cf's verdict borrows the incumbents' deed-of-trust chassis that this spec expressly refuses to use.
DC-8. OPS vs RETAIL/TAX/SEC-CF on the origination fee. Ops (1.1) says the fee is too SMALL and must be re-based upward (percent of home value) for the platform to survive. Retail (T6), tax (I-1), and sec-cf (3.2) treat the existing fee as, respectively, a hidden consumer markup, evidence of a lending trade or business (points/OID), and a HOEPA points-and-fees problem. Incompatible because the remedial directions are opposite: the fee level that fixes the P&L aggravates every consumer-side and characterization finding, and no reviewer proposed a fee that satisfies both constraints (because none exists).
DC-9. OPS vs RETAIL/INVESTOR on the nature of the pause defect. Ops (1.4) frames the pause as accidental mis-calibration (AVM noise, wrong baseline) fixable with buffers and hysteresis. Retail (T9/P2) and investor (AS4) frame it as a counterparty-controlled discretionary switch with one-sided economics. Incompatible as to remedy sufficiency: if the defect is calibration, ops' fixes suffice; if the defect is control — as the distressed-fund scenario (D6/F1) demonstrates — a perfectly calibrated switch still belongs to the party with the incentive to flip it, and the fix set must include independent governance with forfeiture teeth that no report other than mine specifies.
================================================================================
CLOSING POSITION
================================================================================
Cross-examination strengthens the Stage 1 red-team verdict rather than softening it. Two panel reports (actuarial, sec-cf) spared the core contract on premises their fellow reviewers refute (DC-1, DC-4, DC-7); the two "fixable" legal verdicts prescribe each other's kill conditions (DC-2); and every material fix on the table either arms an adversary (FX-1, FX-3, FX-5, FX-6, FX-8, FX-10, FX-11), detonates another domain (FX-1, FX-4, FX-13), or completes the competitor's product for them (FX-4, FX-7, FX-12, fix-stack conclusion). Verdict: KILL, unchanged.
END OF STAGE 2 CROSS-EXAMINATION — RED TEAM