Fit

Fit

Does not fit the framework (P1 not met); contested: P2

Molina clears the universe screen and trips none of the exclusions, but the year-10 durability gate (P1) does not clear — and by the framework's own construction that gate decides fit whatever the other pillars show. Confidence is high: two model families agreed, the trial was order-stable, and load-bearing spreads were at most 0.15. No exclusion hits, no watchlist-only flag, no prior-driven risk; one pillar, P2, is contested.

Universe and exclusions

Here is the decisive framing point: nothing here softens the screen, and nothing here disqualifies Molina either. It is a US company incorporated in Delaware, common stock listed directly on the NYSE under MOH — not an ADR, not a Chinese issuer [1]. Market capitalization is roughly $11.9 billion (52.9M shares at the 2026-07-16 close of $224.82), above the $10 billion floor — but by only about 19%, and at the February 2026 trough of $122.65 the same share count implied about $6.5 billion, below the line. The screen is met on today's price, not settled.

The exclusion list is clean, each item checked against the record rather than waved through:

• Auto OEM / capital-cycle car maker — not triggered. Molina is a government-sponsored managed-care insurer, sector Health Care [2].
• Structural decline — not triggered. Revenue has not fallen for three consecutive years; the only declines were FY2018 and FY2019, reversed, with revenue rising every year since to $45.4 billion.
• Consensus-saturated darling — not triggered. The stock trades at ~0.26x sales, is down 66% peak-to-trough, is a consensus Hold, and forward EPS has been cut ~53%. This is the opposite of a bottom-left-to-top-right chart.
• China dependence — not raised. Operations span 21 US states; essentially all premium is US Medicaid, Medicare and Marketplace [3].
• Promotional CEO without skin in the game — not triggered, with a genuine caveat in the same breath: the long-tenured CEO (since November 2017) delivered acquisitions and buybacks as promised, but missed initial full-year EPS guidance two years running (FY2024 $22.65 vs at-least-$23.50; FY2025 $11.03 vs at-least-$24.50), and insider ownership is low at 1.44% [4]. Credibility is dented; the exclusion still does not trip.

Pattern match

This is the framework's third setup — a healthcare/insurance forecasting error — but only in part. The forecasting-error half fits the template: an insurer misjudged cost trend, took four dated 2025 guidance cuts, and the market repriced the stock; premiums reset on an actuarial calendar, and Medicaid reprices roughly 60% of revenue each January against a cost base management puts about 20% above three years ago [5]. The 2026 Medicaid MCR guide of 92.9% embeds 4% rate increases against 5% trend, with a roughly $2.50-per-share embedded-earnings drag set to reverse in 2027 [6].

The part that does not fit the template: about half the original revision is a Marketplace withdrawal, not a forecasting miss. Membership is being cut roughly two-thirds and about $2.3 billion of premium pulled as enhanced ACA subsidies expire at the end of 2025 [7]. A clean Centene-style repricing case would have the whole industry reprice and mean-revert; here, one large driver is a structural exit rather than a timing error, which is why the diagnosis below leans the way it does.

Pillar ledger

Year-10 durability gate (P1) — not met

This is the gate, and it is where fit is decided. The disqualifier is not tripped: revenue grew from $17.8 billion (FY2016) to $45.4 billion (FY2025), zero consecutive declining years [8]. Market structure and entry barriers support conviction on the revenue leg: government managed care has consolidated to a handful of national plans — Centene, CVS Health, Elevance, UnitedHealth, Humana [9] — down from a field of 500-plus Medicaid contractors two decades ago [10], and entry is gated by state certificates of authority, provider networks, competitively bid contracts, and significant statutory capital [11].

What the gate demands — very-high conviction on both revenue AND free cash flow ten years out — is not available. Revenue is re-competed on 3-5 year state contracts rather than owned; the government counterparty is legislating a Medicaid contraction (OBBBA: a 15-20% Expansion enrollment cut by 2029, plus provider-tax and payment limits phasing in from 2028 over 5-15 years with "uncertain" impact) across the gate's own window [12]; and the FCF leg is volatile — reported FCF was negative in two of the last ten years — and not measurable on the framework's adjusted basis. The gate is binary and fails on any proper doubt; the doubt is real. The strongest surviving counter-fact sits in the same treatment: the OBBBA Expansion cut is only about 4-5% of the 4.57M Medicaid base, and revenue has more than doubled across the decade — so the gate fails on conviction, not on evidence that revenue is set to shrink. The full treatment is on the Durability tab. The vote was unanimous (a/b/c/d all not-met), spread 0.06.

FCF consistency (P2) — contested

Rolling five-year average reported FCF has stayed positive and range-bound ($614M-$1,334M) across the decade despite two negative single years (FY2018, FY2025), which fall about seven years apart — the 5-8 year cadence the framework treats as inherent to insurers [13]. On that evidence the consistency test is met. The counter-fact in the same breath: the framework's own preferred series — rolling five-year ADJUSTED FCF, net of stock-based compensation and trailing acquisitions — is not_computable from the feature file, so the two Claude jurors read "met" while the two Codex jurors returned cannot-determine. That is the split recorded below; the Yield tab carries the FCF series.

Dislocation and yield (P3) — dislocation met, yield short of the bar

The dislocation is genuine and severe: a 66% fall from a $360.77 peak (2024-09-16) to a $122.65 trough (2026-02-11), triggered by four dated 2025 cuts to full-year adjusted EPS guidance ($24.50 → $22 → $19 → $14, actual $11.03) [14]. The fear gauge fires: traded volume spiked to 5.57x the pre-peak median, clustered on the cut dates, with a single ~10.3M-share session (~25x normal) on the Q4 loss day. P3a and P3b are met, each on a survived claim; the Dislocation tab carries the anatomy.

The yield is where the entry trigger falls short. Molina reads as a net-cash, fortress-class balance sheet — long-term debt $3.77B against $4.25B of cash [15] — which selects the 8-9% reference bar. On current and trailing adjusted FCF the yield is far below any bar (negative to ~5.5%); only on a normalized/consensus basis does it reach ~8.1%, at the floor of the fortress bar and about 190 bps short of the 10% default bar (P3c, not met). Consensus forward FCF does clear the 10% bar raw (~10.6% three-year average) and the fortress bar after adjustment (~8.1%), which is why the forward-path criterion (P3d) is met, spread 0.04 — but the reconciling fact is that most of the "real" yield lives in a consensus recovery, not in trailing cash. The Yield tab shows the full computation.

Balance sheet and self-help (P4) — able and willing, with one qualifier

Molina can comfortably outlast a one-to-two-year cost problem: no senior note matures before 2028, the $1.25B revolver is undrawn to 2030, debt-to-capital is ~48% against a 60% covenant, and FY2025 operating income covered interest ~4.1x [16] (P4a met). The repurchase engine is executed, not just authorized — $1.0B of buybacks in each of 2024 and 2025, share count cut 8.3% in one year (57.7M to 52.9M), $500M of live authorization — so the share count is falling and the SBC/serial-acquisition hard-fail does not apply [17] (P4b met). Dividend cover is not applicable — Molina pays no common dividend (P4c).

The counter-fact carried in the same treatment: the ~$8.3B of consolidated cash overstates buyback capacity because most of it is regulated statutory capital at the plans; only $223M sat at the unregulated parent at year-end, and the real fuel is the ~$985M annual upstream dividend flow [18]. Management ranks repurchases third behind organic growth and distressed-plan acquisitions, which compete for the same ~$1.5B annual flow [19]. The Self-Help tab carries the maturity schedule and buyback record.

Diagnosis (P5) — leans permanent, uncontested

The profile's adversarial trial rules the probability the impairment is temporary at 0.37 — a lean toward permanent, with the ruling uncontested (per-judge 0.37 / 0.29 / 0.38; spread 0.09). The best-case framework setup is a short-term earnings fall the market anchors to; here the news plausibly impairs intrinsic value by an amount closer to what the price did, because roughly half the revision is a structural Marketplace exit rather than a timing miss. P5 is not met, spread 0.09. The counter-fact in the same breath: the Medicaid driver has a genuine self-correcting repricing mechanism (about 60% of revenue reprices each January), so a materially better 2027 rate cycle would move the ruling. The Damage Math tab presents both cases and the ruling.

Instrument context (I1) — options exist, but the read is not verifiable

Long-dated listed options exist: LEAPS expiring 2027-01-15 and 2028-01-21 (the latter ~18 months out), with the liquidity typical of a large-cap index constituent. Current 30-day implied volatility is 63.86% (dated 2026-07-21), in the elevated 60-70 reference band rather than the up-to-~50-55 acceptable range, and inflated by the 2026-07-22 earnings print the options are positioned around. Because per-strike liquidity and a settled, non-event IV level could not be pinned to a citable source, the tally marks I1 not-verifiable — but the watchlist-only case (no qualifying long-dated options) does not apply, since the options are there. The Clock tab carries the expression-context line.

What a 3x would require

The framework's target test — the price at bar-yield on normalized adjusted FCF, and what consensus would have to concede — cannot be rendered as the profile's re-rating arithmetic: the tally records the re-rating math as unavailable because the applicable bar or normalized adjusted FCF is missing (adjusted FCF is not_computable where stock-based compensation is absent from the feed). What can be shown is the intrinsic-value band the trial and consensus imply, against the current price and the February trough.

Value scenarios derived from consensus normalized EPS discounted at a back-solved ~9.8% rate, probability-weighted by the trial's p_temporary of 0.37; current price ($224.82, 2026-07-16) and February trough ($122.65) from the profile's deterministic feature file.

At the February trough the gap between price damage and plausible value damage was wide; at $224.82 the price sits inside the probability-weighted band of roughly $217-$265, so the obvious mispricing that existed at $122.65 has largely closed. On this name's own base rate a full round-trip to the prior $360-$420 highs is a two-year-plus process: the one comparable mature-era episode (the 2015-2018 ACA scare, -47.6%) took about 29 months to round-trip, and the partial recovery already in hand (+83% off the trough in about five months) has run faster than that base rate. The Clock tab carries the episode history.

Contested and undetermined

One pillar is contested: P2, FCF consistency. The vote split is met / met / cannot-determine / cannot-determine — the two Claude jurors read the rolling reported-FCF series as stable and met, while the two Codex jurors could not resolve it on the framework's adjusted basis. The named missing datapoint on the cannot-determine side is a Complete rolling 5-year adjusted FCF stability series, including SBC and acquisition adjustments. — equivalently, a rolling 5-year adjusted FCF stability series with SBC and trailing acquisition adjustments — which the feature file returns not_computable. No other criterion is contested, and none was left as an overall cannot-determine verdict.

Provenance

Source: fit_tally.json provenance block and refutations.json (the profile's deterministic tally and skeptic ledger).

The verdict was pressed hard: two independent model families scored the checklist blind to the reader's framework, the trial was rerun with the two briefs in reversed order and moved only 0.035, and a name-masked seat reproduced the same gate result — so the P1 failure is not an artifact of one model, one ordering, or knowing the ticker. Of 41 claims the skeptic checked, none were refuted; six were weakened but survived, and one (the implied-vol read) is unverifiable.

Falsifier ledger

These are the standing conditions that would change the read — thresholds, directions, and windows carried forward from the tally. Adjusted FCF or EBITDA sliding where flat-or-better was underwritten is the framework's own core falsifier; the name-specific conditions follow.

Data gaps

What the run could not answer, carried from the tally:

• Adjusted FCF and its stability series are not_computable: stock-based compensation is absent from the financial feed for every year, and acquisitions defaulted to zero. All adjusted figures here are reconstructed from the filed 10-K cash-flow statements, not from the deterministic feature, and a complete rolling 5-year adjusted FCF series with SBC and trailing acquisition adjustments could not be assembled.
• balance_sheet_class returned unknown (EBITDA absent from the income feed); the net-cash / fortress classification was computed here from the filed balance sheet.
• The 2027 Medicaid rate-cycle outcome — whether states fully fund the ~20%-higher cost base — is unknown; the clock's central mechanism can only be tracked forward against dated catalysts.
• Q2 FY2026 results and the reset full-year guidance (2026-07-22) post-date the corpus, so the candidate re-rating quarter's outcome is not yet in evidence.
• Current market cap could not be pinned to a single dated source around the 2026-07-22 print; the verdict uses the 2026-07-16 feature-file close per the profile contract.
• Official short interest is unavailable (FINRA returned no position rows), so seller composition relies on insider Form 4s and index membership only.
• Per-strike option liquidity and a settled, non-event implied-vol level were not verifiable; only the LEAPS listings and the earnings-inflated 63.86% spot IV were confirmed.
• The capitulation gauge anchors the peak at the 2024-09-16 local high ($360.77); shares traded higher ($419.53) in March 2024, so the full retracement from the all-time high is deeper than the gauge's -66%.
• No national Medicaid managed-care market-share table is in the corpus; market structure is evidenced by named competitors and relative revenue scale.
• No pre-2016 revenue/FCF history is indexed to test durability across the 30-50 year window despite the 1980 founding.