Setup

Figure Technology Solutions (FIGR) is a blockchain-native fintech that went public in September 2025 and just filed its first 10-Q as a public company. The Q1 2026 numbers are loud — revenue +97.6% YoY, adjusted EBITDA margin 49.6% — and the stock is up 18% in the last month. At $43.28 the market is paying 14x revenue and 29x EBITDA for what the filing, read against peer disclosures, describes as a HELOC originator with a tokenized funding wrapper.

What the filing says

Q1 2026 ($167M revenue, $82.7M adjusted EBITDA, $9.5B market cap):

• HELOC servicing UPB +72% YoY to $14.6B. Figure Connect (the marketplace platform) volume +237% YoY to $1.61B. Consumer Loan Marketplace volume +112%. Net take rate 3.8% (+20bps).
• YLDS (Yield-bearing Tokenized Deposits) outstanding: $598M vs $2.7M one year ago. $303M of $598M (51%) sits in "Debt, current to related parties." Per 10-K Note 18, this is Signum Ltd. dba Hastra, a DeFi protocol that Cagney described on the Q1 call as "launched by the Provenance Foundation." Per Morpheus Research (Apr 16, 2026), Provenance is led by June Ou — Cagney's wife and Figure co-founder. Cagney on the same call: "available lender supply was 0.9x borrower demand at end of year, now 1.2x" — i.e., Figure itself is the marginal lender.
• Non-accrual loans on held-for-sale: $2.4M UPB → $14.0M UPB in one quarter (+483%). 90+ day delinquency 3.91% → 5.46%.
• 13% of Q1 revenue ($21.1M) from "Toledo Bend Investments," not in the related-party perimeter and with no public footprint (no EDGAR filings, no Form D, no Form ADV).
• OPEN (Figure's blockchain equity ATS, launched Feb 2026) has 2 issuers and ≈$60K daily volume. 86% of Figure's own Blockchain Stock has left the network per Morpheus.
• May 14, 2026 8-K: board removed plan-level CIC double-trigger and added individual CIC agreements with full acceleration for CEO Tannenbaum, CFO Kgil, and CCO Stevens.

What the market thinks

Mean analyst target $54 (+25% upside). 75% bullish (3 Strong Buy / 3 Buy / 1 Hold / 0 Sell / 1 Strong Sell). Mizuho and Needham reiterated $55 on May 13–14, after the Morpheus short report, after the CIC drafting amendments, and after the Q1 10-Q. Bernstein $67 cites $4T tokenized credit TAM.

Options tell a different story. P/C open interest 2.21 (puts 2.2x calls). IV rank 18% (cheap on a 99% idio vol name). Aug 21 expiry — covering the Q2 10-Q — has 21,795 OI. Term structure inverted 109% → 88%. Market-implied P(FIGR < $32.50 by Aug 21) is approximately 18% via put deltas.

The fintech-with-blockchain-optionality comp is HOOD at ≈6x revenue, 25x EBITDA — implies ≈$28. Pure HELOC originator multiple (RKT ≈1.5–2x revenue, ≈10–12x EBITDA) implies ≈$15. The current $43.28 prices roughly $13/share of blockchain narrative premium above what the filing-supported HELOC case justifies.

Why the gap exists

Four structural reasons the disclosure hasn't been synthesized:

1. Cohort comparison reveals what single-name analysis misses. Rocket reports home equity doubling YoY. BETR funded volume +89%. Achieve +91%. Homebridge +160%. SoFi home equity +570%. FIGR's +112-237% volume growth sits inside the peer range. Most of it is sectoral HELOC tailwind from Fed cuts plus $34T of tappable equity — not Figure-specific share gain.

2. Credit divergence requires the same cohort lens. Consumer credit improved across the cycle in Q1 2026: COF domestic card DQ -29bps QoQ, SYF below 2017–2019 historical avg, BofA HELOC DQ 1.92% → 1.78%. FIGR went the opposite direction, with 90+ DQ on 2024 securitization vintages 2x peer pools (RKT, Spring EQ) per Morpheus. This is the underwriting box (no title insurance, AVM-based valuations, mandatory full-draw, alternative FICO) showing through, not the cycle.

3. YLDS structure requires reading the 10-K against the transcript. Note 18 names Hastra. The Q1 call confirms Hastra is Provenance-operated. Morpheus connects Provenance to June Ou. Each piece is in primary sources; the synthesis is not in any sell-side note.

4. OPEN's competitive set is bigger than analysts model. Robinhood already trades 2,000+ stock tokens in the EU including OpenAI and SpaceX. Securitize tokenizes BlackRock BUIDL. tZERO is IPO'ing in 2026 with five years of operating history. OPEN's 2 issuers and $60K daily volume don't survive that comparison.

Risks

1. Q2 prints clean. If 90+ DQ improves and related-party YLDS share drops below 40% as third-party borrowers (Acura, Agora, Credibly) scale, the bear thesis breaks. Our predictions assign 30–35% probability to clean prints across both metrics.
2. HELOC sectoral tailwind continues to dominate the tape. Fed cuts persist, tappable equity unlocks accelerate, and the stock continues to ride the wave with peers.
3. Squeeze tail. SI is 6.4% with 2.8 days to cover — not crowded yet, but RSI 86 and +18% one-month suggest momentum hasn't broken.
4. Toledo Bend resolves benign. If disclosed as a legitimate large-fund counterparty, the 13% concentration risk compresses.
5. Tokenized credit narrative validates. Real adoption parallels exist (Ondo, WisdomTree, BlackRock BUIDL). FIGR's specific YLDS could attract genuine third-party demand and decouple from Hastra.

Catalysts

What would change our mind

• Q2 90+ DQ falls to 4.5% or below. Non-accrual stabilizes or improves. Underwriting box thesis breaks.
• Related-party YLDS share drops below 40%. Third-party borrower count expands meaningfully beyond Acura, Agora, Credibly. Captive structure resolves.
• OPEN announces 3+ new issuer agreements in 90 days. Platform thesis revives.
• Sell-side capitulates. Mizuho or Needham downgrade, or rating distribution shifts to majority hold/sell. Confirms narrative breaking through to informed flow.
• Toledo Bend named as legitimate institutional counterparty — disclosed hedge fund, family office, or insurance company. Concentration risk compresses to noise.
• BlackRock or Franklin announces tokenized-credit partnership with Figure specifically (not just generic RWA exposure). Bernstein TAM thesis validates idiosyncratically.

Memo LR: 0.7 — direction is bearish, market hasn't fully priced the disclosure issues. Aug 21 prints adjudicate within 90 days.

Evidence