Investment Thesis
Carvana has deliberately marketed subprime auto ABS as "prime" β and the SEC-filed loan-level data proves it is fabricated.
Every Carvana loan across every trust uses unverified stated income (100% Code 3 under Reg AB II). Nearly 39% of loans in 2025-P4 were originated underwater β two to four times the 8β17% industry norm documented by the CFPB. Average FICO appears stable at 706 only because Carvana has engineered the distribution: deep subprime borrowers (FICO below 550) appear in pools for the first time in company history while high-FICO borrowers with hidden risk factors accept rates no genuine prime borrower would tolerate. Loan extensions spike precisely when delinquency approaches contractual trigger thresholds, mechanically suppressing investor protections. When ABS investors reprice this paper as subprime β or trigger thresholds are finally breached β the securitization flywheel that funds over half of gross profit seizes. The $5.6B debt stack due from 2028 makes refinancing a second existential threat.
Core Allegations
π
Deception
100% Stated Income β "Liar Loans" at Scale
Every single loan across every Carvana Auto Receivables Trust (2021-P1 through 2025-P4) is coded Income Verification Level Code 3 under Reg AB II β borrower-stated, never independently verified. Carvana's own legally binding SEC disclosures confirm no income is verified for any loan in any pool. The payment-to-income ratios management cites as a credit quality advantage are built on numbers typed into a text box.
π
Misrepresentation
The FICO Illusion β Engineered to Appear Prime
Average FICO holds at 706 not through credit quality but through deliberate portfolio construction: deep subprime borrowers (FICO below 550) appear in 2025 pools for the first time in company history (3.1β3.5%), offset by a surge of high-FICO borrowers who accept subprime rates β 26.8% of 740+ FICO borrowers paying above 12% APR. The RΒ² between FICO and spread collapsed from 0.50β0.60 to 0.25 by 2024-P4, meaning credit scores explain barely a quarter of pricing. Carvana prices on hidden risk factors invisible to ABS investors.
π
Financial Risk
Collateral Illusion β A Third Underwater From Day One
Approximately 27β39% of Carvana loans have been originated with LTV above 100% in every single vintage since 2021 β a structural feature, not a recent deterioration. Industry-wide, only 8β17% of auto loans carry negative equity at origination. The FDIC explicitly flags routine above-100% LTV as a subprime indicator. On underwater defaulted loans, Loss Given Default jumps to 60β80% versus 20β30% for positive-equity loans β nearly tripling realized loss severity.
β±οΈ
Pattern of Concern
Trigger Management Via Extension Spikes
Loan extensions (which mechanically reset the delinquency clock without curing borrower distress) spike from 10β15 per month to 150β200+ per month precisely as reported 60-day delinquency approaches contractual trigger thresholds. The pattern repeats across multiple vintages β including the highest-quality 2021 pools. Adjusted delinquency (counting extended loans) consistently exceeds trigger levels while reported delinquency stays just below. The report does not allege fraud; the pattern is presented for investor scrutiny.
βοΈ
Structural Risk
Securitization Flywheel β Single Point of Failure
Carvana does not hold its loans. It originates them, securitizes them, and uses the proceeds to fund the next round. Over half of gross profit derives from this cycle. If ABS investors reprice the paper as subprime β or if junior tranches fail to sell β the flywheel seizes and Carvana cannot originate new loans. The $5.6B debt stack (9β14% rates, PIK periods ending, maturities beginning 2028) must be refinanced at the worst possible moment if the repricing cascade is already underway.
π
Alternative Data
Three-Source Convergence Confirms Real-World Stress
"Car repossessed" Google searches trend at multi-year highs (RΒ²=0.73, p<0.001, +7.3 index points/year). Carvana appears in rising related queries alongside confirmed subprime lenders Global Lending Services and Exeter Finance. A 309-video YouTube sentiment pipeline shows 93%+ bearish signals on inventory, demand, and repossessions (all p<0.001), with measured and sensational creators converging on identical direction. The SEC data, the search data, and industry practitioners all agree.
Underwater Loan Originations vs. Industry Benchmark
Source: SEC Form ABS-EE loan-level data (all Carvana Auto Receivables Trusts) Β· CFPB 2024 Data Spotlight on Negative Equity
Mean Spread Over Fed Funds Rate β Selected Vintages
Source: SEC Form ABS-EE loan-level APR data Β· Carvana Auto Receivables Trusts 2021β2025. Prime auto lending typically runs 3β6% spread.
FICO β APR Spread Explanatory Power (RΒ²) by Vintage
Source: SEC Form ABS-EE data Β· RΒ² measures what share of APR pricing variance is explained by credit score. Prime auto lending benchmark: 0.50β0.85.
Narrative Arc & Catalyst Timeline
2021β22
The Golden Age β and the Hidden Flaw
CVNA stock hits $370. Revenue grows from $5.6B to $13.6B. Average FICO 705, no loans below 550. But even at peak quality, ~37% of loans originated underwater β already 2β4x the industry norm. The CFPB would later find only 8β17% of auto loans industry-wide carry negative equity at origination.
2022β23
Near-Death β Stock Crashes 99%, $5.2B Debt Restructured
Fed hikes from 0% to 5.25%. CVNA collapses from $370 to $3.55. 4,000+ employees laid off. $5.2B debt restructured to avoid bankruptcy β new notes at 9β14% interest with PIK provisions and maturities starting 2028. The company that exists today was shaped by this survival imperative.
2023β24
The High-Spread Workaround β Prime Scores, Subprime Rates
Carvana targets high-FICO borrowers with hidden risk factors (high DTI, thin files, undisclosed derogatory history) who can't access bank financing despite their scores. At peak, 26.8% of 740+ FICO borrowers pay APRs above 12%. RΒ² between FICO and spread collapses from 0.55 to 0.25. The "prime" ABS label is maintained by packaging these borrowers alongside genuinely low-risk accounts.
Late 2024
Pivot to Volume β Deep Subprime Enters for First Time
Banks and credit unions re-enter the market, compressing spread income. To maintain volume, Carvana loosens standards. Sub-550 FICO borrowers (deep subprime, 40%+ historical default rate) appear in pools for the first time in company history at 3.5% of 2025-P1. Average loan terms extend to 73 months, slowing principal paydown on already-underwater collateral.
Feb 4, 2026
Report Published β Short Entered at $471
Abelian Analysis publishes thesis. CVNA trading ~$415. Short position taken with $490 call hedge (March 20 expiry). Three independent data sources confirm real-world stress: SEC loan tapes, Google Trends repossession signals, and 309-video YouTube dealer sentiment analysis.
Feb 18, 2026
Catalyst #1 β Q4 2025 Earnings
Consensus expects $5.25B revenue (+48% YoY). The ABS data shows how volume was achieved: pool sizes grew 79% by accepting borrowers Carvana never touched in 2021. Revenue may be hit. Credit quality won't be.
Mar 2026
Catalyst #2 β 2026-P1 Loan Tape & Updated Servicer Reports
The next ABS issuance loan-level disclosure. If sub-550 cohort grows, LTV ratios worsen, or high-FICO borrower spreads remain elevated β thesis validation. First monthly servicer reports for the new vintage follow in mid-March.
Q2βQ3 2026
Catalyst #3 β Peak Loss Period for 2024β2025 Vintages
Auto loan losses peak at 18β24 months post-origination. The 2024 vintages β originated at peak high-spread strategy with FICO explaining barely a quarter of pricing β enter this window in April 2026. Loss acceleration should eat into subordination buffers, pressure ABS spreads, and make extension-based trigger management increasingly untenable.