Fit

Does Cigna fit the system?

Leans toward fitting the framework; watchlist-only: no qualifying long-dated options; contested: P1, P2

Cigna clears the universe screen and trips none of the hard exclusions; the dislocation, yield, self-help and diagnosis pillars land on the fit side of the reference lines. Confidence is low: the name-masked juror flipped the year-10 gate (P1) to not-met — the prior-driven-risk trigger. Two pillars are contested, durability (P1) and FCF consistency (P2), and no qualifying long-dated options could be verified from a citable source, so the framework's watchlist-only rule applies.

The verdict below renders the deterministic tally. It is not overturned, softened, or extended here; every number keys to a surviving claim or a tally entry.

Sources: market cap and drawdown derived from company filings and price feed, as reported; adjusted FCF yield derived from FY2025 cash-flow statement [1]; temporary probability from the blind three-judge trial.

Universe and exclusions

The universe screen passes on both lines. Cigna is US common stock on the NYSE (ticker CI, incorporated in Delaware in 1981) — a domestic common share, not an ADR and not a Chinese issuer [2]. Market capitalization is roughly $76.5B — 268.563M shares at $284.85 on 2026-07-22 — about 7.6x the $10B floor even after the sector de-rating. Both criteria are met.

No hard exclusion hits; the tally records an empty exclusion list. Here is what was checked, and why each clears:

• Auto / OEM (X1): not triggered. Cigna is two segments — Evernorth Health Services (pharmacy benefits) and Cigna Healthcare (medical benefits) — with no vehicle, hardware, or capital-goods line [3].
• Market darling (X4): not triggered. At 12.8x FY2025 EPS ($284.85 / $22.18) and 0.28x sales, Cigna is the cheapest positive multiple in its large managed-care peer group — below Elevance (15.4x) and far below UnitedHealth (32.6x) and Humana (40.4x). This is the opposite of a consensus-owned story; the counter is that low price/sales is structural to pass-through managed care and carries little darling signal, so the low P/E is the informative gauge.
• Structural decline (X3): not triggered. Revenue rose in all nine year-over-year steps, from $39.8B (FY2016) to $274.9B (FY2025), with zero consecutive-decline years [4]. The counter-fact in the same breath: two of the largest steps are inorganic — the FY2019 Express Scripts consolidation and the FY2024 Centene PBM contract (Pharmacy Benefit Services revenue +46%) — but no single-year revenue decline appears anywhere in the record.
• China dependence (S1): flag raised and cleared. China is a 50%-owned equity-method insurance JV carrying a negative ~$(0.3)B book value and no consolidated revenue [5]; International Health premiums are ~1.5% of revenue [6]. Foreign operations were ~53% of 2025 pre-tax income [7], which sounds large, but that reflects international tax structuring and a depressed 2024 domestic base, not China revenue or assets.
• Promotional CEO (X2): not triggered. On the five most material 2024–2026 commitments management missed only the 2024 stop-loss EPS target ($27.33 vs at-least-$28.40, called a disappointment) [8] and kept or exceeded the rest; CEO Cordani owns 0.23% of shares under an 8x-salary holding rule [9] and bought stock in the open market near the October 2025 low [10]. The counter: 0.23% is a professional-manager stake, not founder-scale skin in the game.

Pattern match

This fits the reader contract's third setup — healthcare / insurance forecasting errors — the pattern whose own precedents name Centene ($90 to $25 on a guidance cut, recovered toward $60) and "the cheaper twin over the darling (Cigna over UNH)." The pattern's specific checks:

• A repricing mechanism exists. Stop-loss and individual-exchange books reprice every 1 January — the same books whose 2024 cost overrun drove the drawdown — and the 2026 round is booked and guided to a 83.7%–84.7% medical care ratio [11].
• Regulatory friction gates entry. Insurer and HMO licensing and NAIC risk-based-capital minimums bar casual entry [12] [13] — the same apparatus that now legislates PBM economics onto a fee basis.

The reset was a forward-trajectory downgrade rather than a current-year miss: the deepest single day (30 Oct 2025, -17.4%) was a beat on the quarter that reaffirmed 2025 guidance while resetting the 2026 Evernorth/PBS growth path [14]. That is the pattern's signature, with one honest departure from it — see Clock: the sell side never capitulated.

The pillar ledger

Source: deterministic tally (ruchir/fit_tally.json); pillar arithmetic from the surviving claims cited in each treatment below.

Year-10 durability (P1) — contested

The gate is contested: the two Claude jurors read it met, the two Codex jurors not-met (trimmed-mean probability 0.735, spread 0.08), and the name-masked probe flipped it to not-met. The fit-side arithmetic is a three-firm pharmacy-benefit oligopoly, insurer and PBM licensing barriers, revenue that rose every year to $274.9B, and forward consensus FCF near $9–10B through FY2029; Evernorth supplies 83% of segment revenue but only 64% of pre-tax income at a ~3.1% margin [15], with ACA minimum-MLR floors and RBC capital gating entry [16].

The strongest surviving counter-fact sits in the same breath, and it is why conviction on the free-cash-flow leg is capped rather than maximal: federal law effective 2028 prohibits Part D PBM compensation "directly or indirectly linked to the list price" and requires PBMs to "remit 100% of certain rebates, fees and other remuneration to plan sponsors" [17], and Cigna concedes such reform "may adversely affect our ability to price our pharmacy products and services appropriately" [18]. The exposed Pharmacy Benefit Services unit earned about 31% of FY2025 pre-tax adjusted income ($3,506M of $11,374M); retained rebates are estimated under a tenth of Evernorth pre-tax profit, the phase-in runs to 2028–2029, the FTC matter settled in February 2026 without penalty [19], and Cigna is pre-empting the change with a "no rebate, no spread, fully transparent" fee-based model [20]. The full treatment is on Durability.

FCF consistency (P2) — contested

Contested for a different reason: the two Claude jurors read it met, the two Codex jurors could not determine it. The framework's own adjusted-FCF stability metric is not computable — stock-based compensation is missing for FY2016–2019, so no five-year adjusted-FCF window exists — which is the datapoint the two Codex jurors flagged as missing. Assessed on the available proxy, operating cash flow is always positive and range-bound at roughly $7–12B after 2019 with zero negative years [21]. The counter-fact carried in the same treatment: the verdict-bearing feature itself is unavailable, so the rolling-average stability the framework specifies cannot be measured directly; and Cigna lacks the healthy 5–8 year underwriting-loss cadence the framework treats as a positive for banks and insurers. See Durability.

Dislocation and yield (P3) — met

All four dislocation-and-yield criteria are met. The drawdown runs on dated triggers — a two-year PBM margin reset at Q3 2025 and a whole-industry PBM political selloff in December 2024 — for a 33.4% peak-to-trough fall ($366.85 to $244.41). The capitulation gauge is the honest weak spot: sustained 20-day volume peaked at only 2.12x the pre-peak median, short of the 3–5x that marks forced selling, though the single session of 30 Oct 2025 spiked to ~7.4x before fading. On the framework's adjusted basis, FY2025 adjusted FCF of $7,496M yields 9.80% on $76.5B — 20 bps under the 10% moderate-balance-sheet bar — while the three-year average of $8,358M yields 10.93%, straddling the line; the adjustment nets SBC of $291M against reported FCF [22]. The forward path clears it (P3d probability 0.775, spread 0.03): consensus vendor FCF of $9.0–10.3B in FY2026–FY2028 yields 11.8–13.4%. The counter-fact: a plainer cash-from-operations-less-capex proxy yields only ~9.1% in FY2025 before crossing later, so whether the near-term forward yield clears 10% depends on which FCF definition proves truer. Full workings on Dislocation and Yield.

Balance sheet and self-help (P4) — met

Both live criteria are met; dividend safety (P4c) is not applicable. FY2025 net debt of $22.7B against ~$12.4B EBITDA is 1.84x — moderate, which selects the 10% bar — with only $550M due in 2026 and $2,359M in 2027 against ~$8.4B of annual free cash flow [23] [24]. The repurchase engine is real and executed: Cigna spent $3.6B, $7.0B and $2.3B on buybacks in 2025/2024/2023 and cut its share count 29% from a 379.8M post-Express-Scripts peak to 268.6M, with SBC of only $291M a fraction of repurchases — clearing the framework's hard-fail test for a rising share count [25]. The counter-fact: the buyback cadence is lumpy (halving from $7.0B to $3.6B), and the September 2025 refinancing repriced debt to 4.5%–6.0% coupons above the sub-3% notes it replaces, so interest expense drifts up as the book rolls. Full treatment on Self-Help.

Diagnosis — temporary over permanent (P5) — met

The diagnosis criterion is met (probability 0.62, spread 0.10). The blind three-judge trial ruled the damage temporary over permanent at a 0.62 probability, with an order-stability gap of 0.02 (temporary-first mean 0.62, permanent-first mean 0.64) — not contested. The price erased ~$27.5B of market value from the peak, while the near-term problem plausibly destroys only $0.7–8.8B of NPV under conservative assumptions, leaving a gap of roughly $19–27B [26] [27]. The strongest surviving counter-fact: if the PBM re-rating is deeper than the disclosed Q1 2026 run-rate implies — Evernorth margin already fell 5.3% to 3.1% since 2019 — and terminal growth re-bases negative, intrinsic value could approach today's price; at $284.85 the market already implies about -1.3% perpetual earnings growth [28]. Both cases are argued fairly on Damage Math.

Instrument context (I1) — not verifiable

The instrument criterion is not verifiable, and this is what drives the watchlist-only flag. The Clock tab surfaced web-sourced facts — listed options extending to January 2028 (~18 months out) and 30-day implied volatility of 39.9% as of 21 July 2026, below the ~50–55 reference line — but those facts are not anchored to any filing, so the verification layer could not confirm them from a citable corpus source. The framework's own rule then governs: absent verified qualifying long-dated options, the name is a watchlist candidate, not a live one. Stated as a framework fact, not advice.

What a re-rating would require

The framework's target test — the price implied by valuing normalized adjusted free cash flow at the applicable bar yield — cannot be computed deterministically here. The tally returns re-rating math as unavailable: "Re-rating math unavailable because the applicable bar or normalized adjusted FCF is missing." Adjusted FCF is not_computable in the feature file (SBC absent for FY2016–2019), so neither the normalized adjusted-FCF base nor the implied market cap at the bar is machine-derived.

The closest available reconstruction, carried on Yield, is a three-year average adjusted FCF of $8,358M against the 10% moderate bar — a 10.93% yield, implying the current $76.5B market cap already sits near fair on a bar-yield basis, so the re-rating case rests on multiple recovery rather than a deep-yield entry. Consensus forward FCF at 11.8–13.4% and consensus EPS re-accelerating from ~$30.4 (FY2026) to ~$41.9 (FY2029) is what the sell side already models; the setup asks the buy side to concede that the 2026 pause is not permanent.

Base-rate context from Clock: in the modern managed-care-scale era, Cigna drawdowns of comparable depth round-tripped in roughly 11–36 months.

Source: derived from Cigna price history, as reported; current episode drawdown from the capitulation gauge.

The median completed round trip is ~19 months; the current episode has already run ~22 months from its September-2024 peak with price still 22% below it, tracking the slower end. Stated as arithmetic against the framework's own reference lines — not a recommendation.

Contested and undetermined

P1 (year-10 durability) — contested. Both readings survived. Fit-side: revenue and adjusted FCF more likely than not higher on a three-firm oligopoly and licensing barriers, with forward consensus FCF near $9–10B. Not-fit side: 2028 federal delinking and 100% rebate pass-through reshape the economics of Pharmacy Benefit Services, ~31% of FY2025 pre-tax adjusted income, capping free-cash-flow-leg conviction. Vote split 2 met (Claude) / 2 not-met (Codex); trimmed-mean probability 0.735, spread 0.08; masked verdict not-met.

P2 (FCF consistency) — contested. Fit-side: ten-year operating cash flow always positive, range-bound $7–12B post-2019. Undetermined side: the framework's adjusted-FCF stability metric is not computable — the missing datapoint is a complete five-year adjusted-FCF window with SBC for FY2016–FY2019 — so two of the four jurors returned cannot-determine rather than a verdict. Vote split 2 met / 2 cannot-determine.

No criterion aggregated to a cannot-determine verdict.

Provenance

Source: deterministic tally provenance block (ruchir/fit_tally.json), as computed.

Two model families sat the jury — Claude in two seats, Codex in two — and their split on the P1 gate, plus a name-masked probe that flipped that same gate to not-met, is exactly what caps confidence at low; the tally records the basis as name-mask divergence or load-bearing probability divergence exceeding 0.20. The evidence trail was pressed hard: of 17 fully checked claims, 12 survived, 2 were weakened to a narrower reading, none were refuted, and 3 were unverifiable from the corpus (chiefly the web-sourced instrument facts and the external PBM concentration figure).

The falsifier ledger

These are the standing conditions that would change the read. The ledger carries several jurors' restatements of the same underlying conditions; each is preserved verbatim.

Framework templates:

• adjusted FCF or EBITDA declines where flat-or-better was underwritten
• revenue declines for a third consecutive year
• capital allocation pivots to debt paydown over repurchases
• share count inflects upward
• the industry repricing cycle fails to materialize where industry-wide mean reversion was underwritten

Name-specific, with thresholds and windows:

• FY2026 (Jan 2027): Cigna Healthcare MCR prints above 84.7% or consolidated EPS below $30.25 — stop-loss repricing failed to stick.
• FY2026-27: Evernorth/PBS absolute pre-tax income declines year-over-year beyond the disclosed transitional costs — the compression is structural, not investment spend.
• Post-Aug-2028: after rebate remittance and Part D delinking take effect, PBM margin steps down and does not recover, and enterprise FCF/EPS stall.
• Medical customers fall materially below ~18M beyond the deliberate exchange exit — franchise erosion, not portfolio shaping.
• FY2026 Cigna Healthcare MCR above 84.7% or adjusted EPS below $30.25, showing repricing failed.
• FY2027 Pharmacy Benefit Services pre-tax adjusted income declines despite transition spend no longer becoming a year-over-year headwind.
• By year-end 2028, Evernorth rebate-free adoption is materially below 50% or retention falls below 95%.
• Post-2028 PBM profit per claim remains below 2025 levels after rebate-remittance and delinking are fully implemented.
• FY2026 (reported Jan 2027): Cigna Healthcare MCR prints above 84.7% or adjusted EPS lands below $30.25 — repricing failed to stick.
• FY2027: EverNorth pre-tax adjusted earnings decline year-over-year and margin stays below ~3.3% after transition spend rolls off — investment framing was cover for a structural reset.
• Rebate-free/fee model or 2028 regulation drives PBS profit-per-claim materially below the 2025 base with retention slipping under ~95% — economics genuinely re-based lower.
• Total medical customers fall well below ~18M and keep falling beyond the deliberate exchange exit — franchise erosion, not portfolio shaping.

Data gaps

What the run could not answer:

• Adjusted FCF, adjusted FCF yield, balance-sheet class and float-retirement years are not_computable in the deterministic feature file (missing SBC for FY2016–2019 and a clean FY2025 debt/cash pair); every adjusted figure here is reconstructed from filed cash-flow statements and notes and flagged as derived.
• The rolling five-year adjusted-FCF stability series — annual SBC and a five-year acquisition window — does not exist in the corpus, so the P2 stability metric could not be scored on the framework's own basis.
• Year-10 (FY2035) is beyond all filed guidance and sell-side consensus (which runs to FY2029), so the free-cash-flow-leg conviction rests on structural reasoning about post-2028 PBM economics rather than a filed figure.
• The ~80%-of-claims PBM concentration and the "retained rebates under a tenth of Evernorth pre-tax" magnitude rely on external/web sources; filings establish the three-firm structure qualitatively but do not quantify it.
• Reported short interest and holder-level disclosures returned no rows for CI, so seller identity (forced/anchored vs informed) cannot be evidenced directly.
• Point-in-time CapIQ estimate-revision history is not in the corpus (only a current consensus snapshot), so the exact date-by-date path of estimates versus price cannot be plotted.
• Exact option open-interest and implied volatility by expiry were not obtained from a citable source; instrument availability is web-sourced and unverified, which is what places the name on watchlist-only footing.
• The stop-loss ~100 bps margin-recapture commitment (2026–2027) and the FY2026 EPS guide are still pending in the supplied call history, so their outcomes cannot yet be verified.