# EQUITY FLOW — PRODUCT/OPERATIONS DILIGENCE REVIEW Reviewer role: veteran fintech / mortgage-servicing operator Material: `disclosures/equity-flow-invention-disclosure.txt` (39 sections, read in full) Date: 2026-08-01 Framing assumption per my mandate: this product has already failed and I am explaining why. The structure is clever. The machine that has to originate, disburse, monitor, pause, un-pause, statement, transfer, and unwind thousands of positions over 25-30 years is not designed, not costed, and in several places is explicitly defined as unfunded ("Fee 5: Amount: None"; "Flow 5: No money flows"). Products die in operations, and this spec has written its operating budget out of the product. Legal/tax claim tags used throughout: SETTLED LAW / LIKELY INTERPRETATION / STATE-DEPENDENT / AMBIGUOUS / SPECULATIVE. --- ## 1. ORIGINATION / SERVICING ### Finding 1.1 — Origination unit economics are negative at the stated fee basis (CRITICAL) **Issue.** The origination fee (Sections 18, 29 Flow 1, 36 Fee 1) is 2-3% of *total projected disbursements*, not of home value. On the spec's own base case ($1,500/mo × 84 = $126,000), that is **$3,150 at 2.5%**. The spec asserts this "at scale, covers origination costs, underwriting, and platform operating expenses." Rebuilt from realistic components, it does not cover origination costs alone, let alone platform opex. **Rebuilt fully-loaded cost per funded deal** (mid-scale operation, ~1,000 deals/yr; homeowner-paid pass-throughs excluded): | Component | Low | High | Notes | |---|---|---|---| | Customer acquisition (blended, per funded deal) | $1,500 | $4,000 | Even "organic" funnels have content/SEO/paid amplification cost; HELOC digital CAC runs $1,500-$3,000 per funded unit; a novel product needing education runs higher. Free-tool funnels dilute but don't eliminate this. | | Sales / advisor time (education-heavy, 30-45 day cycle, Sec. 39) | $400 | $1,200 | Multiple touches; this is not a click-to-close product. | | Underwriting & processing labor (title review, mortgage verification, CLTV, insurance verification, conditions in Sec. 39 a-i) | $350 | $700 | 5-9 hours across processor/underwriter/QC at loaded $70-80/hr. | | Appraisal review + AMC management overhead | $75 | $150 | Fee itself is homeowner-paid (Fee 2), but review/rebuttal labor is not. | | Data & verification vendors (KYC, OFAC, credit/mortgage tradeline pull, AVM at underwrite, flood determination) | $75 | $200 | | | Document preparation, notary/closing coordination, recording follow-up | $200 | $450 | Recording fee is homeowner-paid; coordination labor is not. | | Legal doc suite amortization + per-deal compliance QC | $150 | $400 | Assumes $75-150K per-state doc/enforceability build amortized over volume. | | Fallout absorption (appraisals/title on deals that die post-order; 3-day rescissions) | $200 | $500 | 30-50% post-application fallout is normal for a 30-45 day cycle. | | Platform/technology allocation | $200 | $400 | | | **Total** | **$3,150** | **$8,000** | **Midpoint ≈ $5,300** | At the $3,150 base-case fee, the platform loses roughly **$2,000-$2,500 per funded deal at midpoint cost** — before a dollar of servicing. **The scenario that exposes it.** The minimum-eligibility deal. Section 39(b) allows $300K homes with $50K appreciation. At a 70% payout rate that is ~$35,000 of total disbursements → **origination fee of $875-$1,050** against the same ~$5,000 cost stack. The qualification floor admits deals that lose ~$4,000 each at origination. Because the fee scales with disbursements while cost is nearly flat, the product's economics *invert* exactly where the funnel will produce the most volume (modest homes with modest appreciation). **Quantitative impact.** At 1,000 deals/yr with the spec's example mix: -$2M to -$4M/yr origination P&L. Break-even requires either a fee basis change (percent of home value, like incumbent HEIs' 3-3.9% of *investment amount plus* embedded pricing) or average deal size ~2× the base case. **Severity: critical.** ### Finding 1.2 — The "1.0x, no markup" claim is broken by the fee-netting mechanic (MAJOR) **Issue.** Flow 1 note (Sec. 29): if the origination fee is deducted from disbursements, "the total disbursed to homeowner is reduced but the capital return basis remains the gross amount." The homeowner repays 1.0x on money they never received. On the base case: receives $122,850, repays $126,000 — a de facto 1.026x markup, i.e., exactly the "premium on disbursed capital functions like implicit interest" that Section 11 says was deliberately eliminated. **Scenario.** Any journalist, AG, or plaintiff's lawyer reconciling the ACH ledger against the settlement statement. Also an internal reconciliation problem: servicing must maintain two bases (cash disbursed vs. contractual basis) per position, forever, through assignments. **Impact.** Marketing claim "no premium markup" becomes demonstrably false on every fee-netted deal — UDAAP exposure (LIKELY INTERPRETATION that a "1.0x / no markup" claim contradicted by the contract's own math is deceptive under UDAAP standards) and a permanent dual-ledger servicing requirement. **Severity: major.** Fix is trivial: basis = net cash actually received. ### Finding 1.3 — Servicing is real, perpetual, and explicitly unfunded (CRITICAL) **Issue.** Fee 5 (Sec. 36): homeowner pays no servicing fee; "the investor bears all costs of monitoring, AVM assessments, annual statements, and communication." Flow 5 (Sec. 29): during ride-along, "no servicing fees, monitoring fees, or other periodic charges." Nothing anywhere creates a servicing strip, a fee to the platform from the investor, or a fee retained on Tier 2 sale. The servicing machine has no revenue line. **Rebuilt per-position servicing cost, active term:** | Task | $/position/yr | |---|---| | Monthly ACH origination, returns/NOC handling, account-change management (12 credits/yr, exception rate 2-4%) | $25-45 | | Quarterly AVM on every position (Sec. 33: quarterly pause assessment) | $40-100 ($10-25/pull commercial AVM) | | Insurance tracking incl. additional-insured/loss-payee verification (outsourced tracking runs $15-30/loan/yr; verification of a non-standard endorsement is worse) | $20-40 | | Property-tax delinquency monitoring | $5-15 | | Annual statement production (incl. projected settlement calc per Sec. 31) | $10-20 | | Homeowner communications / call center allocation | $30-60 | | Notices, covenant monitoring (occupancy, listing notices, improvement receipts per Sec. 30) | $15-30 | | Compliance/QA/oversight allocation | $20-40 | | **Total active phase** | **$165-350/yr** | Ride-along phase: drop ACH, keep everything else plus death-monitoring and occupancy verification: **$90-180/yr — for 10 to 25 years — against a position generating zero cash until an unscheduled exit.** Undiscounted servicing drag on a 25-year ride-along: **$2,250-$4,500 per position**, with no contractual payer other than "fund operating expenses" of a Tier 1 fund that intends to *sell the position at year 7-8 and dissolve.* **Scenario that exposes it.** Year 12. The originating fund has sold its book and wound down. The Tier 2 insurer bought "no management, no operational burden" paper (Sec. 8's explicit pitch). Who is verifying insurance, running AVMs for the annual statement the contract still requires (Sec. 31 Record Keeping — an investor *obligation* with a default remedy in Sec. 37), watching for the homeowner's death, and answering the homeowner's calls? The spec's answer is nobody, at a price of $0. **Impact.** Either (a) Tier 2 buyers price in a 25-50 bps servicing cost the originator's exit model doesn't contemplate — directly reducing the sale price that "crystallizes the ~11% IRR" — or (b) positions go unserviced, covenants go unmonitored, statements stop (a contractual investor default under Sec. 37), and the book decays into unlocatable claimants (see 6.2). **Severity: critical.** This is the single largest structural operations gap. ### Finding 1.4 — The market-pause trigger is mis-calibrated to enrollment value, guaranteeing far more pauses than marketed (CRITICAL) **Issue.** Payments are sized on appreciation above *purchase price* (Sec. 4), but the pause triggers when value drops below *enrollment value* — the day-one appraisal (Secs. 16, 29 Flow 3a, 33). A homeowner with $150,000 of pre-enrollment appreciation pauses the moment an AVM prints **1% below the origination appraisal**. The marketing frame ("if appreciation is erased, payments stop") is untrue as specified: payments stop when *post-enrollment* appreciation — which is $0 on day one — goes negative by any amount. **Scenario.** A flat market (0% ± 3% — i.e., most markets in most years). Every position in the book sits at the pause boundary from month one, and quarterly AVM noise (median absolute error 5-8%; PPE10 ~85-90% even for good AVMs in liquid markets) flips positions into and out of pause with no real price movement. Each false pause = a written notice within 10 business days (Sec. 31), a stopped income stream the homeowner budgeted on, a 30-day dispute window, a homeowner-funded $400-700 appraisal, and — because the AVM was noise — a reimbursed appraisal plus resumed payments plus a furious customer. Paused months are *not made up* (Flow 3a: paused months don't extend the term), so AVM noise permanently confiscates contracted payments. **Quantitative impact.** With zero true appreciation and 6% AVM dispersion, expect on the order of 30-50% of positions to print below enrollment value in any given quarter of a flat market. In a real -10% downturn, effectively 100% of the book pauses simultaneously: at 5,000 positions that is 5,000 certified-mail notice packages, a call-volume spike of 3-5× baseline, and a dispute pipeline where each homeowner win costs the investor ~$550 appraisal reimbursement plus $200-400 of internal handling. It is also the moment the product's pitch — "monthly income you can count on" — fails for every customer at once, in a recession, on the record. **Severity: critical.** Minimum fixes: pause trigger only after value falls below enrollment by a buffer (e.g., 5%), resume above a lower band (hysteresis), two-AVM-agreement or index-confirmation cascade before any pause notice, and a platform-paid dispute appraisal. ### Finding 1.5 — Investor payment default terms turn back-office errors into fund-level losses (MAJOR) **Issue.** Secs. 31/37: a missed monthly ACH outside a valid pause has a 5-business-day grace and 15-day cure; **three misses in 12 months = homeowner may terminate with forfeiture of the investor's entire appreciation share.** In payments operations, misses happen for non-credit reasons: stale account numbers, ACH file rejects, bank holidays mishandled, a botched servicing transfer after assignment. The contract prices an ops error at the full return on the position (~$63,650 on the base case). **Scenario.** A Tier 2 assignment where disbursement routing changes mid-term (Sec. 35 expressly contemplates assignment during the payment term, with the assignee assuming disbursement obligations). Servicing transfers are precisely where payment continuity breaks. Three hiccups in a year and the insurer's position is gutted. **Impact.** Requires bank-grade disbursement controls (redundant ACH rails, pre-funding, payment-continuity SLA in every assignment agreement) that the spec nowhere requires. **Severity: major** — good for homeowners, but it must be engineered for, and Tier 2 buyers will diligence it. ### Finding 1.6 — Death detection is unspecified for a product whose settlement trigger is death (MAJOR) **Issue.** Death is one of three organic exit events (Secs. 6, 29 Flow 4b, 32 Event 3) and the ratchet keeps running until settlement — yet no one is obligated to tell the investor the homeowner died. Sec. 30 notice duties bind the homeowner (who is dead); heirs are not parties. The SSA Death Master File is access-restricted for non-financial-institution purposes; commercial deceased-suppression data is imperfect. On a 30-year dormant book, the platform needs a standing decedent-monitoring program (probate docket scraping, obituary/data vendors, periodic occupancy verification), plus a probate-interface team, plus the petition-for-sale escalation (Sec. 32 Event 3) — probate litigation at **$15,000-$50,000 and 1-3 years per contested case**, against an estate whose only asset is the house the investor cannot foreclose on. **Scenario.** Homeowner dies in ride-along year 9; heirs (or a squatter, or a renter the heirs installed in breach of occupancy covenants nobody is monitoring) occupy for six years before a title event surfaces it. The "reasonable period" clock (12 months from death or probate close) never started because nobody knew. **Impact.** Duration extension and value leakage the actuarial Tier 2 pricing model doesn't capture; per-decedent administration cost $2,000-$8,000 uncontested, 10-25× that contested. **Severity: major.** --- ## 2. TECHNOLOGY / DATA ### Finding 2.1 — The entire pause/statement apparatus rests on AVMs used at their weakest point (MAJOR) **Issue.** AVMs are decent at portfolio-level marks and terrible at adjudicating whether one specific house is 2% above or below a fixed threshold — which is exactly the query the pause mechanism (quarterly, per Sec. 33) and the annual "projected settlement calculation" statement (Sec. 31) run. Commercial AVM confidence (FSD) varies by market; in non-disclosure states, rural markets, and unique properties, error widens to 10%+. The spec names no provider, no FSD floor, no market-eligibility screen tied to AVM quality, and no model-governance process. A pause triggered by AVM error is simultaneously a consumer-harm event, a contractual-notice event, and a reimbursable-dispute event (see 1.4). **Impact.** Without an FSD-gated eligibility rule ("no enrollment where AVM FSD > X"), the platform originates positions it cannot competently monitor. **Severity: major.** ### Finding 2.2 — The lock-in score is an unvalidated heuristic underwriting committed multi-year capital (MAJOR) **Issue.** Sec. 23: the score combines rate spread, equity cushion, appreciation, and "estimated holding propensity based on demographic and market factors." There is no backtest — there cannot be one; the product and its hold-behavior data don't exist. Yet Sec. 39(h) makes the score a condition precedent to funding, and Sec. 10(f) sells it to investors as a portfolio-quality control. The core economic bet (long holds, low refi) is being underwritten by an untested composite whose most predictive-sounding input ("holding propensity") is the least defined. **Scenario.** Rates fall 150 bps in year 3. Rate-spread — the score's dominant input — collapses across the entire book simultaneously (it is one macro factor, not a diversified signal). The "seasoning data" pitched to Tier 2 buyers turns out to be one regime's data. **Impact.** Model risk on 100% of AUM; no champion/challenger or validation framework specified. **Severity: major.** ### Finding 2.3 — "Demographic factors" in eligibility is a fair-lending and FCRA landmine (CRITICAL) **Issue.** Using demographic data to score who is offered / approved for the product: - **FCRA**: if any third-party data bearing on personal characteristics is used to determine eligibility for a personal/family/household financial transaction, the platform is plausibly a "user" of consumer reports — triggering permissible-purpose requirements and **adverse-action notice duties** the spec entirely lacks (no adverse-action process exists anywhere in Sections 23 or 39). Tag: **LIKELY INTERPRETATION** (FCRA's transaction definitions are not limited to credit; data vendors' non-FCRA "marketing only" carve-outs collapse the moment the data gates approval). - **ECOA/fair lending**: ECOA applies to "credit." The product's whole posture is "not credit" — but (i) regulators are actively contesting that for HEIs (see 5.1), and (ii) state UDAP and civil-rights statutes reach discrimination in non-credit financial products regardless. "Estimated holding propensity based on demographic factors" is, on its face, a disparate-impact machine: age, familial status, and proxies for both are the obvious predictors of mobility. Tag: **AMBIGUOUS** on ECOA applicability; **LIKELY INTERPRETATION** that state-level discrimination and UDAP exposure exists either way. - **CCPA/CPRA and analogues**: profiling for a financial decision triggers notice/opt-out and, under several state laws (CO, CT, VA), profiling opt-outs for decisions producing "legal or similarly significant effects." Tag: **STATE-DEPENDENT**. **Impact.** The data pipeline as described may be **legally unusable** for its underwriting purpose without an adverse-action framework, a permissible-purpose chain from every vendor, and a fair-lending testing program (disparate-impact monitoring). Cost to build properly: $300-600K initial + $150-300K/yr. Cost of not building it: the first enforcement action. **Severity: critical.** --- ## 3. HOMEOWNER EXPERIENCE ### Finding 3.1 — No hardship provision; the pause script is "the contract says so" (CRITICAL) **Issue.** The spec candidly admits (Sec. 16) the pause removes the homeowner's income "exactly when they might need it most," then provides no hardship mechanism whatsoever — no partial-payment option, no advance against recovery, no waiver process. The operational reality: a CSR telling a 70-year-old in a recession that the payments she budgeted for groceries stopped because a computer valued her house $9,000 below a 2026 appraisal, and that she can pay $600 for an appraisal to argue. That call, verbatim, is a state-AG exhibit and a front-page story. The dispute mechanic is regressive by design: the party who just lost income must front cash to contest a model output. **Impact.** Reputational risk concentrated at the worst possible moment (down market, all customers at once — see 1.4). At minimum: platform-paid dispute appraisals, a hardship review desk, and a pause-communication playbook approved by consumer-protection counsel. **Severity: critical** (shared root cause with 1.4). ### Finding 3.2 — Support volume is materially understated by the payments-out model (MAJOR) **Issue.** Payments-out products generate call drivers lump-sum HEIs never see: "where is my payment" (every ACH delay), bank-account changes, pause notices, annual-statement shock (the statement includes a projected settlement number that grows every year — Sec. 31), ratchet questions ("why did your share go from 27% to 28%?"), buyout quotes, assignment confusion, heirs and probate attorneys. Modeled: **0.3-0.6 contacts/position/month active phase** (vs ~0.05-0.1 for a dormant lump-sum HEI), i.e., 1 FTE per ~700-1,200 active positions plus specialist escalation (valuation disputes, estates, hardship). At 5,000 active positions: 5-8 support FTE + 2-3 specialists ≈ **$700K-$1.1M/yr** — nowhere in any fee. **Severity: major.** ### Finding 3.3 — The rate-and-term subordination carve-out fails in practice because the refi lender decides, not the contract (MAJOR) **Issue.** Sec. 13 promises the homeowner can rate-and-term refi with the Equity Flow interest subordinating. But the party that must accept a recorded shared-appreciation interest remaining on title in second position is the *new first-mortgage lender*. Agency (Fannie/Freddie) underwriting treats existing HEI/shared-appreciation instruments as title conditions that generally must be paid off at or before closing; most lenders will simply condition the refi on settlement. Tag: **LIKELY INTERPRETATION** that mainstream refi lenders will not close over a subordinating HEI memorandum absent a negotiated, GSE-recognized subordination framework that does not exist today. **Scenario.** Rates drop in year 4. The homeowner tries to refi 3.1% → 2.4% (sic — or 6.8% → 5.4%), lender says "pay off the Equity Flow first," which forces a settlement the product promised would never be forced, at an appraised value, with the homeowner paying the appraisal. The product's signature promise — "never traps you in your mortgage" — fails at first contact with a loan officer. **Impact.** Either a standardized subordination agreement gets pre-cleared with lenders/GSEs before launch (long, uncertain), or the marketing must stop promising refi freedom. **Severity: major.** ### Finding 3.4 — Assignment to Tier 2 will read as "my payments got sold to an insurance company" (MINOR-MAJOR) **Issue.** Sec. 35: assignment is free, notice within 10 business days, homeowner cannot block. Mid-term assignment moves the *binding disbursement obligation* to a party the homeowner never chose; the notice changes payment routing (a payment-continuity risk, 1.5) and the counterparty's name from a fund to an insurer. Without a persistent servicer-of-record who answers the same phone number for the life of the position, each assignment manufactures confusion, missed-notice risk (certified-mail addresses go stale), and distrust. The spec's own memorandum design (Sec. 34: investor contact info recorded on title) guarantees stale contact data on 20-year-old recordings. **Impact.** Requires a MERS-like registry / permanent servicer facade — which is more unfunded servicing (1.3). **Severity: major** at scale. ### Finding 3.5 — Homeowner pays ~$1,100-$3,200 out of pocket before the first dollar arrives (MINOR) Appraisal $400-700 + title/recording $200-500 + recommended legal review $500-2,000, plus the origination fee (netted or not), against competitors (HELOCs) that routinely close with zero closing costs in 7-14 days vs. this product's 30-45 days (Sec. 39 timeline). This is a funnel-killer at the application stage and will drive the fallout costs in 1.1. **Severity: minor individually, feeds 4.2.** --- ## 4. SCALE / UNIT ECONOMICS ### Finding 4.1 — The first fund cannot pay for the machine (CRITICAL) **Issue.** Sec. 22: proof-of-concept fund of $25-50M. At the base-case commitment (~$126K/deal), that funds **200-400 positions**. Platform revenue available: origination fees (200-400 × ~$3,150 = $630K-$1.26M, one-time, spread over a 2-3 year deployment) plus, if the platform is also the fund manager, 1.5-2% management fee ($375K-$1M/yr). Against that: the compliance build (5.1: $1.5-3M before first deal), the origination cost stack (1.1: $5,300/deal midpoint → $1.1-2.1M), servicing (1.3), and a core team. A lean 15-20 FTE platform with vendors runs **$4-6M/yr**. The proof-of-concept fund covers perhaps a third of that. The platform needs roughly **$750M-$1B+ AUM-equivalent (6,000-8,000 positions)** before recurring fees plausibly cover a properly staffed operation — and the spec's own sequencing says institutional capital won't arrive until *after* years of seasoning data that the underfunded platform must somehow survive to produce. **Severity: critical.** This is a venture-subsidy plan without naming the venture subsidy. ### Finding 4.2 — Funnel conversion from free-tool user to funded deal will be brutal (MAJOR) **Issue.** The Noteworth education-first funnel (Secs. 23, 27) is real but slow. Modeled honestly: free-tool user → qualified lead 3-8% (the qualification gates are tight: 2.5% rate spread AND $50K appreciation AND 20% equity AND owner-occupied AND clean 12-month mortgage history AND CLTV ≤80%); lead → application 10-20% (novel product, education burden, out-of-pocket costs, spousal co-sign); application → funded 30-50% (30-45 day cycle, appraisal surprises, cold feet in a 3-day rescission window). Net: **0.1-0.8% of tool users become funded deals.** To fund 1,000 deals/yr the platform needs 125K-1M engaged tool users/yr — a real consumer-marketing operation, which reprices "organic" CAC back toward the $2,500-4,000 in 1.1. **Drop-off hotspots:** the out-of-pocket appraisal ask; the moment the homeowner's attorney reads the ratchet; the moment the refi carve-out is explained honestly (3.3); the 30-45 day wait vs. a HELOC's week. HELOCs and HE loans close faster, are legible in one sentence, and the homeowner keeps 100% of appreciation. The one-sentence rebuttal ("no monthly payment back") is strong for cash-poor seniors — the segment where this product most resembles the reverse-mortgage complaint file. **Severity: major.** ### Finding 4.3 — Loss, fraud, and recovery economics rest on court petitions the fund will rarely bring (MAJOR) **Issue.** The investor has no foreclosure right; remedies are notice → cure → demand → arbitration or court petition (Secs. 34, 37, 38). Realistic loss channels: senior-lien foreclosure wiping the junior interest (Event 4 — surplus recoveries in foreclosure are historically pennies), occupancy fraud (enrolling, moving out, renting — detection requires the occupancy monitoring nobody funds, 1.3), value fraud at exit (non-arm's-length sales; Sec. 33's "higher of" remedy requires the investor to *detect* the related party first), and origination misrepresentation (Category C — the only personal-recourse path, against defendants who are, by construction, cash-poor). Model **2-4% of positions hitting a loss/enforcement event over a 10-year life, with net recovery on enforcement events of 20-50 cents on the dollar after $15-50K per-case legal spend.** No loss reserve, no enforcement budget, and no fraud-detection function appears anywhere in the spec. **Severity: major.** --- ## 5. COMPLIANCE OPERATIONS ### Finding 5.1 — "Not a loan, so TILA/RESPA/usury don't apply" is the load-bearing wall, and it is cracking in exactly the states that matter (CRITICAL) Claim-by-claim tagging of Sec. 25: - "Title recording and escrow settlement is established real-estate law in all 50 states" (Sec. 12): **SETTLED LAW** as to the *mechanics* of recording and escrow generally; **STATE-DEPENDENT** as to whether *this instrument* (an "equitable interest in future appreciation" that is expressly "not a lien," Sec. 34) is enforceable at closing in each state. Title companies underwrite off familiar instruments; a sui generis memorandum will get exceptions, escrow delays, and in some states outright refusal to insure over it until underwriting bulletins exist. - "Not subject to TILA/RESPA/state usury" (Sec. 25): **AMBIGUOUS, trending adverse.** Regulators and courts have been actively contesting HEI not-a-loan positioning: state regulators (e.g., Connecticut's banking department) have treated shared-appreciation HEI agreements as mortgage loans requiring licensure; litigation over Unison's product produced rulings recharacterizing an HEI as a loan under state law; and the CFPB has publicly analyzed HEIs as potentially credit under TILA. For the *lump-sum* products. Equity Flow's *monthly stream plus a time-based ratchet* looks more loan-like than any incumbent: periodic advances (negative-amortization optics) plus a claim that grows +1%/yr with duration is functionally accrual. Tag for "the ratchet is not interest": **AMBIGUOUS** at best. - "Not subject to state insurance regulation" (Sec. 25): **LIKELY INTERPRETATION** for the core product; **AMBIGUOUS** for Equity Shield™ (Sec. 17), which is described as functioning "like portfolio insurance" with a funded guarantee — that is a financial-guaranty structure that may itself require an insurance license or a regulated wrapper. **STATE-DEPENDENT.** - "May be subject to state HEI regulation as it develops": **SETTLED** that this is already happening (several states have enacted or proposed HEI-specific statutes imposing disclosure, licensing, and cost caps); **STATE-DEPENDENT** in content. - 3-day rescission "not legally required" (Sec. 39(l)): **STATE-DEPENDENT** — and if any state recharacterizes the product as a mortgage loan, TILA's own rescission and disclosure regime applies with different mechanics, and having offered a voluntary lookalike will not cure the violation. - Ch. 13 non-modifiability because "it is not a debt" (Sec. 32 Event 7): **SPECULATIVE** — the spec itself says "requires validation by counsel." A bankruptcy court that sees monthly advances, a recorded interest, and a growing claim can see a secured claim subject to the code. (Estate/BK reviewer's lane; flagged here because servicing must operationalize whichever answer courts give, per state, per circuit.) - Tax deferral of monthly payments as installment draws on a forward sale (Sec. 24): **AMBIGUOUS** — the spec admits it; operator's point is that the compliance build must include information-reporting infrastructure (1099 posture, escrow reporting) for *both* characterizations, because if the IRS recharacterizes mid-book the platform must issue corrected forms across every open position and vintage. **Cost before the first deal funds:** 50-state enforceability and licensing survey with instrument-level opinions ($400-800K), state-specific document suites for the launch footprint ($75-150K/state), licensing where required (CT-style regimes: 6-12 months), compliance-management system, vendor oversight, complaint program, fair-lending testing (2.3): **$1.5-3M and 9-18 months pre-revenue.** Nowhere budgeted. **Severity: critical.** ### Finding 5.2 — There is no contingency plan for mid-book reclassification, and this product cannot improvise one (CRITICAL) **Issue.** Unique to Equity Flow among HEIs: the investor owes *forward monthly payments for years*. If a state issues a C&D or reclassifies the product as lending mid-operation: - **Continuing to pay** = continuing to originate/advance on an unlicensed loan in that state (each monthly ACH is arguably a new advance). - **Stopping payment** = investor default under Secs. 31/37 — 3 misses and the homeowner may terminate *with forfeiture of the entire appreciation share* across every affected position. A regulator can, with one letter, put the fund's whole return in a state on a 3-payment fuse. The spec contains no severability-by-state plan, no regulatory-event clause in the contract (force-majeure-style suspension with tolling), no reserve, and no CFPB posture (proactive engagement, no-action application, or wait-and-see — each with different cost/risk; the spec says only "early engagement is advisable" with no plan). **Severity: critical.** Minimum fix: a contractual regulatory-event provision negotiated *into the homeowner agreement now* — which consumer counsel will note is itself a disclosure problem ("your payments may stop if a regulator objects" belongs in the marketing that currently says payments are a binding obligation). ### Finding 5.3 — Force-placed insurance and tax advances by a non-lender (MAJOR) **Issue.** Sec. 37 lets the investor force-place insurance and advance taxes, adding both to the capital-return basis. The force-placed market, carrier appetite for naming a non-lender "equity interest holder" as loss payee/additional insured (Sec. 30), and the consumer-protection rules around force-placement are all built for *lenders*. Carriers routinely refuse non-mortgagee loss-payee endorsements or issue them incorrectly; verifying 5,000 correct endorsements annually is exactly the tracking cost in 1.3. Adding advances to the capital-return basis also further erodes the "1.0x of what you received" story (see 1.2). Tags: enforceability of loss-payee status for a non-lien equitable interest — **STATE-DEPENDENT / AMBIGUOUS**; force-placement by a non-lender without a lender's statutory framework — **STATE-DEPENDENT**. **Severity: major.** --- ## 6. ASSIGNMENT / SECONDARY OPS ### Finding 6.1 — Tier 2 economics assume a servicer that no one has agreed to pay (CRITICAL) **Issue.** Sec. 8 sells insurers "no management, no operational burden." Sec. 35 transfers "all rights AND obligations" to the assignee — including annual statements, AVM monitoring, notice handling, death detection, probate petitions, release recording within 15 business days of a settlement the assignee may not even learn about promptly. Insurers do not run consumer servicing shops; they will require a named servicer with an SLA and a fee — industry-analogue servicing strips run 25-75 bps or $75-200/position/yr. That fee must come out of the purchase price, i.e., out of the Tier 1 exit that "crystallizes the ~11% IRR." The IRR claim and the "no burden" claim cannot both survive contact with a Tier 2 diligence team. Additionally, until a licensed/rated servicing arrangement exists, insurers' investment committees will not clear the asset regardless of yield — Sec. 22 admits the "rated structure or actuarial track record doesn't exist yet." **Severity: critical** (compounds 1.3). ### Finding 6.2 — Decades-dormant recorded interests decay: stale contacts, lost claimants, and marketable-title statutes (MAJOR) **Issue.** The memorandum (Sec. 34) records the investor's contact info at origination. After two assignments and 18 years, that recording points to a dissolved fund. Consequences: homeowners and title companies cannot locate the claimant at sale (escrow delays, quiet-title actions *against* the position, settlements closing with funds interpled or escheated); releases don't get recorded (the Sec. 34 remedy — homeowner petitions court at investor's cost — presumes a findable investor). Separately, several states' marketable-record-title and dormant-interest statutes extinguish or require re-recording of old interests after 30-40 years — a 25-30 year ride-along plus probate delay walks into that window. Tag: **STATE-DEPENDENT** — whether an "equitable interest in future appreciation" survives a marketable-title act without periodic re-recording is unresearched in the spec. Fix: a permanent registry/servicer-of-record (MERS analogue), a re-recording calendar per state, and assignment recording standards — all of which are, again, unfunded servicing. **Severity: major.** ### Finding 6.3 — Position packaging and transfer mechanics are unbuilt (MINOR) **Issue.** A Tier 2 sale requires: per-position data tapes (payment history, AVM history, covenant status, insurance status, occupancy verification — data the unfunded servicing function was supposed to be collecting), diligence re-underwriting, assignment instruments recorded per county (at $50-100 × thousands of positions), homeowner notices, and payment re-routing without tripping 1.5. Cost per transferred position: **$150-400 all-in.** Not modeled anywhere. **Severity: minor** individually; it compounds 6.1. --- ## Cross-cutting drafting/spec-integrity notes (operator's punch list) 1. **Appraisal payer contradiction:** Sec. 33 (origination appraisal "Ordered by: the investor… Paid by: included in origination costs") vs. Sec. 36 Fee 2 ("Paid by: Homeowner") vs. Sec. 29 Flow 2 note ("paid by the party ordering it"). Three answers to one question. Ops cannot build a fee matrix from this, and the inconsistency will surface in the first regulator document request. 2. **Pause resumption language:** Sec. 16 ("payments *can* resume") vs. Flow 3a ("resumes when value recovers"). Discretion vs. obligation — a litigation-grade ambiguity on the product's most sensitive mechanic. 3. **Investor annual-statement duty (Sec. 31)** survives into ride-along and assignments but has no funded owner (1.3, 6.1) — the contract obligates a function the fee schedule prices at zero. 4. **"No transfer fee to homeowner when investor assigns" (Sec. 36)** is correct consumer design but confirms no one pays for assignment ops (6.3). --- ## VERDICT: FIX-BEFORE-LAUNCH This is not a kill — the core structural ideas (monthly deployment limiting capital at risk, ride-along decoupling of disbursement from settlement, equity-side ratchet) are genuinely better-designed than the incumbent HEI mechanics, and the enforcement mechanism at sale piggybacks on real title/escrow rails. But as specified, the product fails in operations within one housing cycle, for reasons the spec itself half-admits: 1. **The operating budget is defined as zero.** Fee 5 and Flow 5 price servicing at nothing while Sections 30-33 mandate a quarterly-AVM, insurance-tracked, death-monitored, statement-producing servicing operation for up to 30 years. Until a servicing strip exists — funded in the deal economics and priced into the Tier 2 exit — the product is a 30-year obligation with a 7-year budget. This alone is launch-blocking. 2. **The pause trigger is mis-calibrated (enrollment value, no buffer, AVM single-source) and will fire constantly in flat markets**, converting the product's marquee safety feature into its chief consumer-harm and dispute-cost engine. Fixable with bands, hysteresis, confirmation cascade, platform-paid disputes, and a hardship desk — none of which exist in the spec. 3. **Origination loses money at the stated fee basis**, and loses the most on minimum-eligibility deals. The fee basis, eligibility floor, or both must change before the unit economics can be projected honestly to LPs. 4. **The not-a-loan wall is AMBIGUOUS and eroding** (state HEI statutes, loan-recharacterization rulings, CFPB attention), and this product — monthly advances plus a time-ratcheting claim — is the most loan-shaped HEI yet proposed. With binding forward payment obligations, the absence of a regulatory-event plan (5.2) means a single state action can force the fund to choose between unlicensed lending and contract-wide forfeiture. A contractual regulatory-event clause, a state-by-state launch gate with instrument-level opinions, and a decided CFPB posture are prerequisites, not fast-follows. 5. **FCRA/fair-lending exposure in the lock-in score (2.3)** makes the current data pipeline legally unusable for underwriting until an adverse-action framework, permissible-purpose chain, and disparate-impact testing exist. If the sponsor funds a real servicing entity with a contractual fee stream, recalibrates the pause, reprices origination, drafts the regulatory-event provision, and builds the compliance/data governance layer — roughly **$4-7M and 12-18 months before the first institutional dollar** — this becomes an operable product with a genuinely novel structure. Launched as specified, it fails: first in origination P&L, then in the first flat-market pause wave, and terminally when the first state recharacterization meets a book of binding monthly obligations that no one is left servicing.