Variant Perception

Where We Disagree With the Market

Our sharpest disagreement: the market is pricing Copart as if revenue growth determines per-share earnings, while the buyback engine plus 21-year-clean cash conversion mean EPS compounds 5–7% even at zero topline growth — and consensus FY27 EPS of $1.70 (only 7% above FY25) does not appear to model the FY26 buyback velocity. The Street has compressed CPRT's P/E from 30.5x (10-year average) to 20.2x on one negative print (Q2 FY26 -3.6%) and a flat 9-month line, and is increasingly anchoring the right comp at LKQ (12.9x) rather than historical CPRT. Three of the four "market is right" pillars look softer on the evidence: the competitive read on RB Global's 27% revenue CAGR confuses M&A with share gain, the US service-revenue read ignores the sequential thaw and the mechanical hurricane base-lapse, and the earnings-quality discount embedded in 14x EV/EBITDA is misapplied to the cleanest forensic profile in specialty industrials (18/100 Clean, CFO greater than NI for 21 straight years, zero debt, $4.8B net cash). The single observable signal that resolves the debate is the Q4 FY26 print (early September 2026) — specifically US service revenue YoY and Q4 buyback dollars; both either confirm the mechanical inflection or break it.

This is not a "stock is cheap" memo. The variant view is a measurable gap between the underwriting assumption the market is using (CPRT is a low-growth specialty industrial whose earnings power tracks revenue) and the evidence (CPRT is a duopoly cash-engine whose per-share earnings power tracks a buyback velocity the market has not metabolized).

Variant Perception Scorecard

A variant strength score of 68 reflects three independent disagreements with the market that each have observable resolution signals inside the next four months. Consensus clarity is high (72) because the multiple compression is measured, the analyst distribution is documented (6 buy / 5 hold / 1 strong sell, $41.44 target), and the tape (52-week low, 18% below 200-day SMA, June 2025 death cross) records exactly which direction the market is leaning. Evidence strength is the rate-limiting input: the data behind each variant claim is auditable from one upstream tab and one cross-check (e.g., RBA segment data, the FY25 hurricane base, the 9M FY26 buyback line). The four-month horizon to resolution is unusually short because the Q4 FY26 earnings print and FY26 10-K Note 15 disclosure land together in September 2026 — the same window updates all three variant views.

Consensus Map

The six issues collapse into one underwriting question: should CPRT be priced as a duopoly cash engine in a temporary air pocket, or as a high-quality but ex-growth specialty industrial whose right comp is LKQ? The market has shifted from the first frame to the second over the past twelve months. Our disagreement is targeted at the three pillars holding that shift up.

The Disagreement Ledger

Disagreement #1 — the buyback denominator. Consensus would say the buyback is welcome but does not change the structural growth picture; the multiple compression to 20x is the correct response to revenue flatlining. Our evidence disagrees: at 977M shares and $1.6B deployed in 9 months, the math implies share count down 6-7% annually, and roughly $300M of T-bill interest income on the cash that remains is recurring earnings power. The Street's FY27 EPS estimate of $1.70 is consistent with share count down only ~2-3% (pre-FY26 cadence), not down 6%. If the variant is right, the market is mispricing the denominator — paying 20x earnings that compound 5-7% from share retirement alone. The disconfirming signal is a Q4 FY26 buyback line below $300M, or a Q1 FY27 buyback line that drops without a stated multi-year framework. Neither is in evidence today.

Disagreement #2 — RBA's growth is acquisition, not share gain. A consensus analyst would point to RBA's 27% five-year revenue CAGR versus Copart's 16% and conclude the duopoly is rebalancing. Our evidence disagrees: that CAGR is dominated by the $7.3B IAA acquisition in 2023 that nearly doubled RBA's revenue base overnight. The clean economic test — operating margin — has held a 21pp gap for a decade including the period after RBA paid up explicitly to scale IAA against Copart. If IAA were actually closing the gap on duopoly economics, margins would converge before unit share shifts, not after. If the variant is right, the bear case's "LKQ is the right comp" framing collapses — the moat is intact, the multiple compression is cyclical, and the right anchor remains historical CPRT in the 25-28x range. The disconfirming signal is RB Global's next automotive-segment disclosure: an IAA salvage-segment margin closing to within 5pp of Copart's would break this read.

Disagreement #3 — sequential thaw and mechanical base-lapse. Consensus reads the three-quarter U.S. service-revenue weakness as evidence of a structural demand reset, with the stock at a 52-week low. Our evidence disagrees: Q3 FY26 already showed sequential improvement to -0.4% (versus Q2 FY26 -5.6%), and Q4 FY26 is the first quarter where the FY25 Helene/Milton hurricane base ($56M of CAT volume) lapses — meaning the YoY comp gets mechanically easier. The FY13-14 historical analog showed a similar margin compression-then-revert pattern in under 18 months. The disconfirming signal is straightforward: U.S. service revenue YoY in the Q4 FY26 segment table (early September 2026). A positive print is consistent with the variant; a continued negative print refutes it.

Disagreement #4 — earnings-quality premium not paid. A consensus analyst would say accounting quality is already in the multiple. Our evidence disagrees: CPRT trades at 13.9x EV/EBITDA — the same as RBA (with $1.4B net debt and a 15.5% margin) and only modestly above LKQ (8.6x, with $3.3B net debt). In peer sets where accounting drift normally gets penalized, CPRT — 21 straight years of CFO greater than NI, zero non-GAAP layer, debt-free for four years, 18/100 Clean forensic grade — is being charged the same multiple as the names with the drift. A 2-3 turn premium for the cleanest profile in the comp set is standard; that premium is not in the price today.

Evidence That Changes the Odds

The eight evidence rows split into two groups. Rows 1-3 are the active evidence we are betting on: the sequential thaw, the durable margin gap, and the buyback velocity all point the same direction and resolve in the same Q4 FY26 / Q1 FY27 window. Rows 4-8 are passive evidence — clean accounting, international compounding, the hurricane base lapse, the absent LKQ pilot, and benign DOJ disclosure — that support the variant without requiring active confirmation in the next four months. The DOJ row is the single fragility: if the FY26 10-K Note 15 escalates, the entire variant view becomes conditional on the resolution path rather than the print.

How This Gets Resolved

The resolution map is unusually concentrated: five of seven signals resolve in the September-November 2026 window, and three (Q4 FY26 print, FY26 10-K filing, RB Global Q2 segment disclosure) land within a four-week period. This is what makes the variant view tactically actionable — the same calendar window that updates the bear case on every metric the bear has been pressing also updates the variant on every metric the variant depends on. Either side gets a clean read; neither side has the option of waiting another quarter to argue the data is incomplete.

What Would Make Us Wrong

The variant view depends on the buyback cadence continuing at the FY26 9-month pace. If management throttles the program because the multiple recovers, or quietly returns to the pre-FY26 cash-hoard pattern, the per-share compounding math collapses and the bear case framing (capital allocation was reactive, not regime change) becomes correct. The risk is not zero: the buyback was announced without a multi-year framework, there is no target capital structure, and the same founder-family that hoarded cash for fifteen years still controls roughly 9% of the stock and approved a September 2025 anti-pledging waiver letting Johnson pledge ~11M shares. That waiver is uncomfortable behavior at a multi-year low and it is exactly the kind of evidence that suggests the family is raising cash defensively, not pressing the buyback at a generational entry. The single observable that breaks our view here is a Q4 FY26 buyback below $300M paired with the absence of a multi-year framework in the FY26 10-K MD&A.

The competitive read on RBA depends on the operating-margin gap remaining the right test. If RB Global's next segment disclosure shows the IAA salvage segment margin closing materially — even by 5-7pp — the duopoly is rebalancing and the bear comp (LKQ at 13x) becomes the right anchor. The fragility is that RB Global has been investing in IAA modernization since the 2023 acquisition; three years is roughly the right window for integration synergies to start showing in segment margins. We are betting the gap holds because it has held a decade including the explicit-acquisition period, but a single quarterly segment disclosure could update that view. The disconfirming signal is unambiguous: IAA salvage-segment operating margin within 5pp of Copart's in the next two RBA quarterly disclosures.

The sequential-thaw reading depends on Q4 FY26 US service revenue actually inflecting positive — a mechanical base-lapse helps, but does not guarantee it if underlying demand has reset more severely than three quarters of data show. The FY13-14 historical analog took ~18 months to revert; the FY26 air pocket is currently 12 months in. If Q4 FY26 still prints US service revenue negative YoY despite the hurricane base lapse, the "structural" reading becomes harder to dismiss and the LKQ comp becomes the bear's strongest claim. The signal is clean: a positive US service revenue YoY print validates; a continued negative print refutes.

The DOJ AML matter is the binary risk where our variant view does not survive a single bad outcome. If the FY26 10-K Note 15 escalates to a charging document, consent decree language, market exclusions, or a material accrual, the international buyer pool — which we cite as the engine of revenue-per-car compounding — becomes a constrained variable. We accept the consensus read on direction (most AML matters resolve with KYC enhancements) but acknowledge this is the one event in the next six months that could update the long-term thesis without earnings. Our variant view is therefore conditional on a benign or unchanged DOJ disclosure; any other outcome breaks it.

The first thing to watch is Q4 FY26 US service revenue YoY in the September 2026 earnings release segment table — paired with the Q4 buyback line in the same filing. Both update in the same press release; both either confirm the mechanical inflection and the regime-change buyback, or break them together.