Carvana ($CVNA): Subprime Auto in Prime Clothing – A Forensic Short Thesis (Detailed Breakdown with SEC Data)

I just published a deep-dive bearish analysis on Carvana ($CVNA) over at Abelian Analysis: ["Carvana: Subprime Auto in Prime Clothing"](https://abeliananalysis.com/posts/carvana-short-thesis/).The stock's ripped \~10,000% from the \~$4 lows, trading around $415–470 recently, but the underlying credit picture looks increasingly shaky. The core argument: Carvana has shifted heavily toward subprime-like lending practices while keeping headline metrics (avg FICO \~706) looking prime. This masks rising risk in their securitization-dependent model.Key pillars from the piece (all pulled from public SEC filings like 10-D servicer reports, ABS prospectuses, and Regulation AB II data):

* 100% stated-income loans (no income verification) since 2021 – Code 3 across all trusts. PTI ratios look low (down to 7.2%), but based on potentially overstated borrower income. Reminds of pre-2008 "liar loans."
* FICO illusion – Stable average hides a new sub-550 cohort (3–3.5% of recent pools) and high-FICO borrowers (>740) paying subprime rates (spreads \~7–9% over Fed Funds). Correlation between FICO and pricing collapsed (R² from 0.5–0.6 to \~0.25).
* Underwater collateral exploding – LTVs jumped from 21% in 2021 to 92% in 2025 pools; \~36% originated underwater. Loss-given-default could hit 60–80% on defaults.
* Delinquency manipulation via extensions – Spikes reset the clock just before covenant triggers, keeping reported 60+ day DQ artificially suppressed (visible in monthly 10-D filings across vintages).

Excited to hear everyone's thoughts. Tell me why I'm wrong!