Fit

Fit

Does not fit the framework (P1 not met); contested: X1, X2, X3, X4, S1.

Expand Energy clears the universe (U.S.-listed on Nasdaq, ~$22.0B market cap) and trips no disqualifying exclusion, but it fails the one pure gate: the year-10 durability test. A commodity price-taker whose free cash flow ran from $8M to $2,302M in six years, and whose predecessor went through the last down-cycle in Chapter 11, cannot support "year-10 adjusted FCF higher, with very high conviction." Confidence is low — the name-mask probe flipped the exclusion labels, so a prior_driven_risk flag stands. Five criteria (X1, X2, X3, X4, S1) are recorded contested.

The answer, in one screen

The gate decides it. Everything below the gate — the setup, the yield, the balance sheet, the diagnosis — is real analysis, and some of it is favourable, but none of it offsets a failed P1. That is the framework's own rule: P1 not_met -> does_not_fit (gate; nothing offsets it).

Sources: market cap and adjusted yield derived from company filings and market data (fit_features); the trial probability from the report's adversarial diagnosis trial. Adjusted FCF = reported FCF − stock-based comp − 5-yr avg acquisition spend.

Universe and exclusions

Universe — clears both lines. Expand Energy is ordinary U.S. common stock listed on Nasdaq under EXE, incorporated in Oklahoma, with principal offices in Oklahoma City and all operations onshore in the United States [1] — not an ADR, not a Chinese issuer, so U1 passes. At the $91.52 close on 2026-07-24 the ~240.37M shares put market cap near $22.0B, 2.2x the $10B floor; U2 passes even though the stock sits ~29% below its December 2025 peak.

Exclusions — none fires, but each is recorded contested. The company is a natural-gas exploration-and-production business, not an auto manufacturer (X1 does not apply) [2]. It trades near 1.8x sales and ~5x EV/EBITDA on a drawdown chart, the opposite of an extreme multiple-to-sales darling, so the consensus-saturated-story exclusion X4 does not trigger [3]. Operations are entirely onshore and gas sells at U.S. index prices; China appears in the 10-K only as generic "China-Taiwan relations" geopolitical risk, so the S1 China-dependence sensitivity is effectively zero [4].

Here is the decisive point about the "contested" label: it is not disagreement about the facts. On X1, X2, X3, X4 and S1 both model families concluded the exclusion does not fire — one family recorded that as not_met, the other as not_hit. The tally treats the differing labels as contested and, combined with the name-mask flip, that is what pins confidence at low. No juror found Expand to be an auto company, a promotional-CEO story, a structurally declining business, a saturated darling, or China-dependent.

Pattern match

Of the framework's four setups, Expand is closest to cyclicals at the bottom — a large, essential, capital-heavy producer priced off a soft commodity strip. But it fails that pattern's own entry checks. The precedent (large banks whose loss-rate data contradicts the macro fear) turns on mean-reverting fundamentals plus capitulation; here the fear gauge never reached capitulation (peak volume only 1.43x normal, see the Dislocation tab), and the "cyclical bottom" is a commodity price the company does not control rather than a mis-forecast that reprices. It is emphatically not the healthcare/insurance forecasting-error pattern (no 1:1 guidance-cut anchor) and not a quality tech monopoly on a fear dip (no monopoly, no franchise pricing power). It fits none of the four cleanly.

The pillar ledger

Source: deterministic fit tally; per-criterion arithmetic from the surviving claim ledgers cited in each section below.

Year-10 gate (P1) — the decisive result

Not met, unanimously. All four jury seats and the name-masked seat returned not_met; the trimmed-mean year-10 probability is 0.435 with a spread of 0.06 — cross-family agreement, tight. Year-10 revenue is plausibly higher on the LNG/power/industrial demand build and rising production (1,335 Bcfe in 2023 to 2,622 Bcfe in 2025). What cannot be asserted with very high conviction is year-10 adjusted free cash flow: reported FCF ran $22M (2020), $1,140M, $2,302M, $551M, $8M, $1,839M (2025) — a ~288x span [5] — because the number is set by an unforecastable Henry Hub price, and the predecessor filed Chapter 11 in 2020 [6]. Two of the framework's five conviction sources are genuine — natural gas is an essential product, and the business is capital-heavy ($24.4B net property, plant and equipment on $28.3B assets, an investment-grade survival barrier). But the market-structure source fails: at ~6% of U.S. output Expand is the largest producer yet still a price-taker with no regulatory entry barrier [7], and the long-operating-history source is qualified by a 2021 fresh-start reset and the October 2024 Southwestern merger [8]. Under the framework, any proper doubt resolves the gate to not met. Full treatment: the Durability tab.

Strongest counter-fact: consensus free cash flow of roughly $2.6B-$2.9B a year across FY2026-FY2029 sits above the current figure, and the demand tailwind (LNG export growth, coal retirements, data-center load) is structural [9]. It moves the volume question; it does not settle the price, and no consensus estimate extends to year 10.

FCF consistency (P2) — not met

The deterministic fcf_stability feature is not_computable (stock-based comp absent from the feed), so the test was run by hand on reported FCF [10]. The rolling five-year average moved 45% between the only two post-emergence windows ($805M for FY2020-24 to $1,168M for FY2021-25). Real acquisition cash ($1,967M in 2022, $459M in 2024) would push a properly adjusted weakest year negative. That is unpredictable, not the healthy every-5-to-8-years cadence the framework tolerates for insurers and banks. Detail on the Yield tab. Counter-fact: no reported year in the window was negative, and consensus projects a steadier $2.6B-$2.9B ahead.

Dislocation and yield (P3a, P3b, P3c, P3d)

The setup is soft. The only dated adverse event is the 9 February 2026 CEO exit and Houston-relocation announcement (a −6.5% session on 2.16x volume); the rest of the ~29% peak-to-trough fall tracks a weak gas tape, so P3a is not met. The capitulation gauge topped at 1.43x normal volume against a ≥2x reference line, so P3b is not met — 1.43x is well short of true fear (Dislocation). On valuation, the framework-adjusted FCF yield computes to 5.6% — 440 bps below the 10% moderate-class bar — because a $563M five-year-average acquisition charge dominates the adjustment; P3c is not met [11]. The balance-sheet class is moderate (net debt $4,393M / ~$5.0-5.5B EBITDA = 0.8-0.9x), so the 10% bar — not the 25% levered bar — applies [12]. The forward path is the favourable leg: consensus forward FCF yield runs 11.5-13.1% on the current market cap, clearing the bar unadjusted, so P3d is met at p=0.585 (spread 0.04) [13]. Counter-fact and the honest limit: mid-cycle normalized adjusted yield lands at 9.6%-11.2%, straddling the bar — it clears 10% only if Expand is finished as a serial acquirer and if Henry Hub averages ~$3.50+, neither of which the company controls (Yield).

Balance sheet and self-help (P4a, P4b, P4c)

Outlast (P4a): met. Investment-grade at ~0.8x leverage, no note maturities before 2029, a fully undrawn $3.5B revolver [14]. Capital allocation was pointed at debt paydown first — the framework's specific worry at a dislocation — but the debt-first phase is largely complete: the $1B full-year reduction target was met in Q1 2026 and management said it could "rebalance that with share buybacks and shareholder distributions" [15].

Repurchase engine (P4b): not met — a hard-fail. The share count rose rather than fell, and the driver is stock-financed acquisitions: the all-stock Southwestern merger issued ~95.7M shares, dwarfing the ~5M cumulatively retired by buyback [16]. A rising share count from serial acquisition is disqualifying by the framework's own rule; this is the same mechanism that drops Accenture once adjusted. Counter-fact: the buybacks are executed, not just authorized (~$1.5B of real repurchases 2022-2025), and the count rise is a one-time step from an accretive merger, not a chronic SBC drip. Dividend cover (P4c) is not_applicable — the base dividend yields ~2.5%, too small to carry the return case. Detail on Self-Help.

Diagnosis (P5) — met, and it is favourable

The independent trial ruled the impairment more likely temporary than permanent: P(temporary) = 0.66 (mean 0.65, spread 0.07, per-judge 0.61-0.68), not contested, order-stable (temporary-first 0.68 vs permanent-first 0.635, gap 0.045). The near-term earnings hit behind the drawdown is close to zero — full-year 2026 guidance unchanged, Q1 2026 EPS beat by 5.4%, and FY2026 consensus above FY2025 actuals on revenue, EPS and FCF; the cut landed only in outer-year (FY2027) EPS, down ~17% from $10.07 to $8.36. Both sides of the trial are laid out on the Damage Math tab. Counter-fact and calibration: two exhibits from the temporary brief failed the judges' quote-check — an "estimates rose" claim contradicted by FY2027 EPS falling to $8.36, and a $2.21/MMBtu figure misattributed from a peer's benchmark — so the 0.66 rests on the surviving one-year FCF snapback and ~13% forward yield, and the permanent side's fixed midstream commitments held it below ~0.7.

Instrument context (I1) — not verifiable

The tally records I1 as not_verifiable: listed-option existence and current implied volatility could not be confirmed from a citable source inside the run corpus (the web-research pipeline failed on a billing error, and the Clock tab's option facts rest on public aggregators rather than a primary feed). This criterion never blocks the pillar verdicts; here it is moot, because the P1 gate already decides the fit.

What a 3x-in-3-years would require

The tally did not compute a re-rating price: "Re-rating math unavailable because the applicable bar or normalized adjusted FCF is missing" — the balance-sheet-scaled bar resolved to moderate but the framework-adjusted, normalized FCF is not_computable in the feature file (stock-based comp missing across FY2020-2025), so there is no adjusted-yield anchor to invert into a target price.

What can be said, as framework arithmetic rather than a target: at the 10% moderate-class bar, a 3x re-rating would need normalized adjusted FCF to hold near the mid-cycle $2.5B (acquisitions rolled off) and the multiple to expand toward the bar — i.e. consensus would have to concede a durable, through-cycle gas price around $3.50+/Mcf, which the Yield tab prices at roughly a 70-75% chance within 1-2 years. The base-rate context from the Clock tab is sobering for a "3x" ambition: in Expand's ~5.5-year post-bankruptcy history the deepest drawdown is only 33%, and comparable episodes round-tripped in 2 to 26 months — there is no precedent for the 60-70% forced-selling swing the framework's biggest wins came from. None of this is underwritten as a recommendation; the gate has already ruled the name out.

Contested and undetermined

Five criteria are recorded contested — X1, X2, X3, X4 and S1 — in every case because the two model families labelled a non-firing exclusion differently (not_met vs not_hit), not because they disagreed on the underlying facts. The name-masked seat returned not_met on all four gate exclusions, which is what raised the prior_driven_risk flag. Nothing was marked cannot-determine: every pillar reached a verdict, and the two features that could not be computed (adjusted-FCF yield, FCF stability) were rebuilt by hand from filed statements rather than left open.

Provenance

Source: fit tally provenance block and the adversarial trial tally.

Two families of models voted; they agreed on the verdict that matters — the year-10 gate failed on every seat, masked and unmasked, at a tight 0.06 spread — so the direction of the answer was pressed hard and did not move. Confidence is nonetheless low, and honestly so: the name-mask probe changed how the exclusions were labelled, which the framework treats as evidence that priors, not pure evidence, may be doing some of the work on those five contested criteria.

Falsifier ledger

These are the standing what-would-change-this conditions carried from the jury and the diagnosis trial's flip-conditions. They are redundant by design — harvested independently from multiple seats — and reproduced verbatim so each threshold, direction, and window stands on its own:

Data gaps

The run could not answer several things, and the verdict is stated with them in view: the deterministic feature file could not compute adjusted FCF, adjusted-FCF yield, FCF stability, balance-sheet class or revenue trajectory (stock-based comp and annual revenue absent from the XBRL feed, EBITDA missing for FY2025), so every adjusted figure was rebuilt by hand from filed 10-K cash-flow statements. There is no primary regulatory short-interest series (the feed returned zero rows), no Henry Hub price strip staged in the corpus, no independent year-10 gas-price forecast, and no confirmable listed-option/implied-volatility feed. The tradable history is only ~5.5 years post-Chapter-11 and the entity was reconstituted by the October 2024 Southwestern merger, so all base rates rest on a short, structurally shifting sample rather than a long franchise curve.