Fit

The answer

Does not fit the framework (P1 not met); contested: X2, P2, P3c, P3d, P5

The year-10 gate is not met, and under the framework's own order of operations that settles it: nothing offsets a failed gate. Confidence is low, on the tally's computed basis — name-mask divergence or load-bearing probability divergence exceeded 0.20. The prior-driven-risk flag is set. No exclusion screen was hit and the watchlist-only flag does not apply. Five criteria came back contested: X2, P2, P3c, P3d and P5.

The gate rule the tally applied, in its own words: P1 not_met leads to does_not_fit, with nothing able to offset it. Everything below the gate — an 11.2% adjusted free-cash-flow yield, a volume spike at 7.4 times the pre-fall median, a share count down 19.7% in five years — is recorded, and none of it changes the answer.

Universe and exclusions

Both universe tests are met and no exclusion was hit. Fiserv's common stock is primary-listed on the Nasdaq Global Select Market under the ticker FISV, having transferred from the New York Stock Exchange on 11 November 2025 where it traded as FI [1]; the issuer is incorporated in Wisconsin and is neither an ADR nor a Chinese issuer. Market capitalisation is $30.56 billion — 549.0 million shares at the 29 July 2026 close of $55.67 — which is 3.06 times the $10 billion floor. The cushion is the residue of a collapse rather than a stable state: the same shares were worth roughly $130 billion at the March 2025 peak, and a further 67.3% decline from here would be needed to breach the floor.

Four exclusion screens and the China sensitivity flag were checked against evidence. The company manufactures nothing: all $21,193 million of FY2025 revenue is merchant acquiring and financial-institution processing [2], so the car-company exclusion has no subject. Structural decline was checked and not found on the filed annual record — revenue rose in every fiscal year from 2017 to 2025 — and the trial's probability that the impairment is temporary, 0.61, sits well above the 0.35 line at which a strongly-permanent diagnosis becomes a hit. The consensus-saturated-darling screen fails in the other direction: at 1.44 times FY2025 sales Fiserv is cheaper than each of its four listed comparables, which run 2.1 to 5.0 times, trades at roughly 6.8 times guided 2026 earnings, carries a Hold consensus with 69.4% of ratings neutral, and is 80.2% below its peak. On China, the FY2025 Form 10-K discloses no China revenue and no China assets, the renminbi is absent from the five currencies management names as material, and all non-US-and-Canada revenue is 16% of the total spread across EMEA, Latin America and Asia-Pacific combined [3] — an inference from silence rather than a disclosed zero, and recorded as such in the data gaps.

The promotion screen, X2, is the one that split the jury, and it split across model families. Here is the deciding point. In February 2025 management guided FY2025 to 10-12% organic revenue growth and $10.10-$10.30 adjusted EPS [4], reaffirmed the Clover target in July with the statement that "only the timing of realizing them has been extended" and that "we are maintaining our guidance for $3.5 billion of Clover revenue this year" [5], then on 29 October 2025 reset the same year to 3.5-4% organic growth and $8.50-$8.60 adjusted EPS [6]. Free cash flow came in at $4.44 billion against roughly $5.5 billion originally guided [7]. All fifteen current directors and executive officers together hold 324,298 shares [8], about 0.06% of the company. Both prongs of the test are evidenced, which is what two seats found. The counter-fact sits in the same treatment and is why the other two seats did not: the chief executive who made those promises resigned on 12 June 2026 [9] and the president resigned for good reason on 7 July 2026 [10], the reset baseline was then beaten — FY2025 adjusted EPS of $8.64 against the $8.50-$8.60 range, free cash flow of $4.44 billion against $4.25 billion guided, at approximately 93% conversion [11] — and insiders bought about $1.72 million of stock in the open market at the June 2026 low. The tally records the result as contested, not as a hit.

memory access out of bounds

Verdicts and vote splits from the run's deterministic tally; the NASDAQ listing record from Item 5 of the FY2025 Form 10-K [12], total revenue and operating income from Item 1 [13], and beneficial ownership from the 2026 proxy statement [14].

Pattern match

Fiserv fits none of the reader contract's four setups, and in three of them the mismatch is not close.

The large-bank cyclical pattern needs a bank: a lender whose monthly loss-rate data can be read against macro fear. Fiserv sells to banks and is examined alongside them, but it carries no credit book to price. The high-dividend pattern needs a dividend, and Fiserv has never paid one and states it does not anticipate paying one [15]; the framework's dividend-cover criterion, P4c, is recorded not applicable for exactly that reason.

The insurance-forecasting-error pattern is the closest in shape and still fails its own check. The shape rhymes: a single year's guidance cut, and a share price taken down roughly with it as though the cut were permanent. But that pattern's specific check is a repricing mechanism — premiums reset, the book readjusts, and mean reversion is structural rather than discretionary. Fiserv has no equivalent. Its recovery is a company-specific revenue and margin bridge: management's own second-half 2026 waterfall builds from low-single-digit first-half growth to 6-8%-plus in the second half, of which roughly half — about 3.5 percentage points — is a non-recurring 2025 comparison lapsing rather than new demand [16]. Half of the mechanism is arithmetic; the other half has to be executed.

The quality-monopoly-on-a-fear-dip pattern splits the company in two. One half qualifies structurally: the US bank and credit-union core-processing market is a four-vendor set — Fidelity National Information Services, Fiserv, Corelation and Finastra, per Jack Henry's own Form 10-K [17] — and Fiserv is examined directly by the Federal Reserve, the FDIC and the OCC as a significant service provider under the Bank Service Company Act [18]. That protected half is the one contracting: Financial Solutions organic revenue fell 6% in the March 2026 quarter with all three business lines negative, and segment adjusted operating margin fell to 38.1% from 47.5% [19]. The growth half is described by Fiserv's own Item 1 as a market that "is fragmented, highly competitive, and served by a multitude of large and small businesses" [20], where the company's own estimate of its revenue share is 7% in US small-and-medium business and 2% in global enterprise [21]. The precedent behind that pattern is a specific, testable outside fear against an intact franchise. The fear here is about the franchise's own printed numbers.

The pillar ledger

Year-10 durability — P1 not met

This is the gate, and it is the only criterion in the run that is binary by construction. All four seats returned not met, cross-family, at a trimmed-mean probability of 0.605 with a spread of 0.09 — the tightest agreement on any probabilistic criterion in the run. The masked seat returned the same answer at 0.63.

The revenue leg is not what fails. Revenue rose in each of the nine fiscal years from FY2017 to FY2025, $5,696 million to $21,193 million [22], so the framework's disqualifier — high-single-digit revenue decline for three consecutive years — is not triggered, and even the depressed 1-3% 2026 guidance compounded for a decade leaves year-10 revenue above today's.

The cash leg is what fails. Adjusted free cash flow on the framework's basis — reported free cash flow less share-based compensation less the trailing five-year average of acquisition spend — fell 20.7% in FY2025, from $4,297 million to $3,408 million, built from $6,062 million of operating cash flow, $1,763 million of capital expenditure, $357 million of share-based compensation and $820 million of acquisition payments [23]. In the March 2026 quarter consolidated operating income fell 34% to $918 million on revenue of $5,027 million [24], an operating margin of 18.3% against 27.2%. Of the framework's five conviction sources, capital intensity is not one here — capital expenditure including capitalised software ran 8.3% of revenue, which is not the capital-heavy essential the framework treats as structurally defended — and market structure covers only the shrinking half of the company.

The strongest surviving counter-fact sits in the same treatment, and it is a commitment rather than a result: at the 14 May 2026 Investor Day management committed to a 2027-2029 adjusted revenue CAGR of 4-6%, a 2029 adjusted operating margin above 37%, approximately 90% free-cash-flow conversion and over $13.5 billion of cumulative free cash flow across 2027-2029 [25]. The gate asks for very high conviction and resolves genuine doubt downward; a plan presented by a chief executive who resigned 29 days later [26] is evidence, not conviction. The full treatment is in Durability.

FCF consistency — P2 contested

Two seats found the consistency standard met and two returned cannot-determine, and the disagreement is about which series to measure rather than about the numbers. On the reported basis every rolling five-year average rose: $2,505 million (FY2018-22), $3,022 million (FY2019-23), $3,619 million (FY2020-24), $3,830 million (FY2021-25) — up 52.9% across the windows, with a coefficient of variation of 9.8% and no negative year in eight [27]. The framework asks the question on adjusted free cash flow, and that series cannot be built: the two seats from the second model family recorded the missing datapoint as five consecutive years of adjusted FCF, including SBC and trailing acquisition-average deductions, for rolling 5-year stability, and a complete rolling five-year adjusted FCF series with SBC and acquisition-spend data. Reconstructed from the filings, adjusted free cash flow exists for FY2023-FY2025 only, and the FY2023 term is distorted to $18 million by the $15,083 million First Data purchase sitting inside its five-year window.

The counter-fact: the most recent year broke the trend the average records. Reported free cash flow fell 15.1% in FY2025 and the reconstructed adjusted figure fell 20.7%, so the rising average is being carried by FY2023 and FY2024.

Dislocation and yield — P3a and P3b met, P3c and P3d contested

The event and the capitulation are unanimous. A dated, filed trigger exists: the 29 October 2025 release cut FY2025 adjusted EPS guidance from a $10.225 midpoint to $8.55, a 16.4% reduction, and the shares fell 44.0% in that session, the largest one-day decline in a price record running back to 1990 [28]. The counter-fact in the same treatment: the medium-term 9-12% organic growth target was withdrawn on the same call, and the shares had already fallen 46.9% from the March 2025 peak before any guidance changed. On capitulation, the measured volume multiple is 7.39 times the 180-day pre-peak median against a 2-times reference line — 41.4 times on the reset day alone, with 67.2% of the share count trading in the four weeks that followed. The counter-fact: volume has never normalised, running 1.7 to 4.9 times the pre-fall median in every month since April 2025 and still 2.4 times in July 2026, which reads as a register still changing hands rather than a single flush.

P3c and P3d both split 2-2, and both splits turn on one unresolved input: which bar applies.

The yield itself is not in dispute. Adjusted free cash flow of $3,408 million on a $30,563 million market capitalisation is 11.15%. Against the framework's 10% moderate-balance-sheet bar that is 115 basis points through; against the 25% levered bar it is 1,385 basis points short. The classification is genuinely unsettled. Total debt at 31 December 2025 was $28,997 million — $1,239 million short-term and current maturities plus $27,758 million long-term [29] — against $798 million of cash and EBITDA of $9,025 million, giving 3.12 times, just over the rule's 3.0-times cut-off. Excluding $2,112 million of finance-lease and other financing obligations the same computation gives 2.89 times; the consensus data set gives 2.89 times on its own definitions; and management reports 3.0 times at year-end 2025 against a standing 2.5-3.0 times target [30]. The two seats from one model family read the class as moderate and recorded P3c met at plus 115 basis points; the two from the other applied the literal 3.12-times reading, took the 25% bar, and recorded not met at minus 1,385. A third figure cuts against both: the three-year average adjusted yield as the formula literally computes it is 8.42%, 158 basis points below even the moderate bar, though excluding the 2019 First Data window it is 11.71%.

P3d carries the widest disagreement in the run — a trimmed-mean probability of 0.42 on a spread of 0.67, with seat readings of 0.72, 0.73, 0.06 and 0.12. The underlying arithmetic is the same for all four: consensus free cash flow of $3,880 million for FY2026 and $4,280 million for FY2027 becomes 10.33% and 12.25% after deducting share-based compensation and averaged acquisition spend, which clears 10% and falls 1,467 and 1,275 basis points short of 25%. The skeptic weakened this claim rather than letting it stand unqualified, precisely on that point. The counter-fact: FY2026 headroom over the moderate bar is 33 basis points, and the plan requires roughly 400 basis points of second-half margin expansion that no reported quarter has yet shown. The full workings are in Yield, the drawdown anatomy in Dislocation.

Balance sheet and self-help — P4a not met, P4b met, P4c not applicable

P4a failed unanimously, and only on its second leg. The outlasting leg passes comfortably: 2026 maturities are $1,239 million against total debt of $28,997 million [31], an $8.0 billion revolver runs to August 2030, and the sole financial covenant is a 3.75-times leverage test [32] against roughly 15% of headroom. The allocation condition is what fails. In the fourth quarter of 2025 Fiserv repurchased 3 million shares for approximately $200 million and paid down over $1 billion of debt [33]; in the March 2026 quarter it repurchased 3.3 million shares for approximately $200 million again while stating that "we are focused on managing our leverage ratio and remain committed to returning capital to shareholders" [34]; and in June 2026, with the shares at their $47.18 low, it issued EUR 1.0 billion of new notes to fund a tender for existing ones. The levered exception does not rescue it: of its three legs, the yield leg fails at 11.15% against roughly 25%, free cash flow as a share of revenue fell from 24.7% to 20.3%, and only the share-count leg holds. The counter-fact: the same Investor Day commits the "Majority of free cash flow directed to stock buybacks while targeting low-end of 2.5-3.0x" over 2027-2029 [35] — buybacks promised, but sequenced behind the leverage target.

P4b was met unanimously because the framework's hard-fail condition is a rising share count and the count is falling: diluted weighted-average shares went from 683.4 million in FY2020 to 549.0 million in FY2025, a compound rate of minus 4.29%, on $22,987 million of executed repurchases, including 32.2 million shares for $5.6 billion in FY2025 alone [36]. The counter-fact is the price paid and the timing: the average cost of those 174 million shares was about $132 against today's $55.67, quarterly spend fell 90.8% from $2,164 million in the March 2025 quarter to $200 million in the March 2026 quarter while the price fell 72.6%, and no shares were bought in December 2025 or January 2026. P4c is not applicable — no dividend has ever been paid [37], so the coverage test has no subject and the entire shareholder-return channel runs through the repurchase decision. The maturity schedule and buyback record are worked in Self-Help.

Diagnosis — P5 contested

All four seats recorded P5 as contested and all four carried the same number rather than re-deriving it: the adversarial trial's probability that the impairment is temporary is 0.61, on a per-judge range of 0.62, 0.38 and 0.61, a spread of 0.24 and a reading-order gap of 0.125 between the judge who read the temporary brief first and the two who read the permanent brief first. Nine exhibit quote-checks failed across both briefs, several as materially out of context.

The arithmetic that supports the temporary reading is the size of the gap. Equity value fell $101.3 billion from the March 2025 peak. The present value of the cash the reset actually removed computes to $50.7 billion even under a permanent level shift measured against the growth algorithm management withdrew — a $50.6 billion gap, or 2.0 times; measured against the milder anchor the ratio is 4.5 times. The counter-fact in the same treatment: enterprise value fell $96.7 billion, or 62.2%, not 76.8%, because net debt rose $4.6 billion over the window, and the March 2025 peak multiple of 23 times was struck on a guidance path management later attributed in part to an Argentine inflation windfall. Both cases are put fairly in Damage Math, which reports the ruling and does not override it.

memory access out of bounds

Source: the run's adversarial trial tally; recorded probability 0.61, spread 0.24. No filing page applies.

Instrument context — I1 not verifiable

All four seats returned not verifiable, and the reason is sourcing rather than absence. The single docket entry on listed options was ruled unverifiable for want of a corpus-verifiable source on a load-bearing input, so its content — a chain running to 21 January 2028, 72,599 contracts of open interest on that expiry, and implied volatility of 46.98% at the long end — cannot support a finding under the run's verified-evidence-only rule. The same entry puts that longest expiry at 17.7 months from 31 July 2026, short of the framework's 18-month preference. This criterion never blocks a pillar; it drives the watchlist-only overlay, which applies only to a fits or leans-fit verdict and is therefore recorded false here. The dated facts and their sources are set out in Clock.

memory access out of bounds

Verdicts, splits, probabilities and spreads from the run's deterministic tally; the arithmetic is drawn from the surviving claims cited above, principally the FY2025 Form 10-K cash-flow statement [38] and debt note [39].

What a 3x in three years would require

The framework's target test cannot be stated as arithmetic in this run. The tally records the re-rating math as unavailable because the applicable bar or normalized adjusted FCF is missing: the bar, the normalized adjusted free cash flow, the implied market capitalisation at bar-yield and the upside to it are all null. Both inputs are missing for the same reason the P3c and P3d splits exist — the balance-sheet class is unresolved between the 10% and 25% bars, and the feature file returns the framework's adjusted free-cash-flow series as not computable. No price at bar-yield is asserted here, and none should be inferred from the figures above.

What can be stated is the distance and the base rate. A tripling from the 29 July 2026 close of $55.67 is $167.01, which would still leave the shares roughly 30% below the March 2025 peak of $237.79. Consensus is nowhere near that: the mean twelve-month target of $66.63 is 19.7% above the close and 72.0% below the peak, and FY2027 adjusted EPS estimates have drifted from $9.15 to $8.93 over 180 days rather than turning up. Against the company's own record, this drawdown has no precedent — at 80.2% it is 1.56 times deeper than the worst of the six prior 30%-plus episodes since 1990. Those six round-tripped in a median of 19.8 months, but the two nearest in depth, in 2002 and 2007-08, each took 41 to 42 months, and both were market-wide falls with market-wide recoveries. The episode record is set out in Clock.

memory access out of bounds

Derived: peak, trough and current prices from the run's capitulation gauge; consensus mean target as of 30 July 2026; the three-times figure is 55.67 multiplied by three and is arithmetic, not a target.

Contested and undetermined

Five criteria are recorded contested. No criterion was returned cannot-determine at the tally level, so the overall verdict carries no undetermined component — though P2 is contested precisely because two of the four seats individually returned cannot-determine on it.

memory access out of bounds

Splits and probabilities from the run's jury dockets and deterministic tally; the underlying figures are cited in the pillar sections above.

Provenance

memory access out of bounds

Source: the run's jury dockets, trial tally and refutation ledger. No filing page applies.

Two features of that table carry more weight than the rest. The verdict-bearing criterion was pressed hardest and held best: two independent model families reached the same not-met answer on the year-10 gate, at probabilities within nine points of each other, and the name-masked seat reached it too. The confidence tier is nonetheless low, because the tally computes confidence from disagreement wherever it appears, and disagreement here is real — a 0.67 spread on the forward-yield path, a 2-2 split on which balance-sheet bar applies, and a name-mask probe that came back divergent on the exclusion screens. On X1, X3 and X4 the masked seat recorded the same substantive no-hit under a different label; on X2 and on the P3d probability the divergence is genuine.

memory access out of bounds

Source: the run's jury dockets. Seats A, B and the masked seat are one model family; seats C and D are the second. No filing page applies.

The falsifier ledger

Eighteen standing conditions were recorded, several of them near-duplicates arising from different judges' rulings on the same question. They are the run's what-would-change-this conditions, and they run in both directions: some would break the recovery reading, others would confirm it. The first of them resolves within a week of this writing, when the June 2026 quarter reports on 6 August 2026.

memory access out of bounds

Summarised from the run's falsifier ledger; the exact recorded wording of each entry follows below. No filing page applies. The 2H 2026 revenue bridge behind entries 6, 10, 14 and 15 is management's own [40]; the 2029 algebra behind entries 9, 11 and 18 is the 14 May 2026 medium-term outlook [41].

The entries as recorded, verbatim. Bracketed document-and-page tokens are the run's internal source keys, not links; each resolves to the filing page cited elsewhere on this tab.

Data gaps

The run could not answer the following, and several of them bear directly on the contested criteria above.

• The framework's own yield basis is not computable from the feature file. Adjusted FCF, the adjusted yield, the yield baseline, FCF stability, the balance-sheet class and float-retirement years all return not computable, because the structured cash-flow feed carries no share-based-compensation or acquisitions line for any fiscal year. This is a feed gap, not a disclosure gap — Fiserv discloses both lines in every filed cash-flow statement — and every adjusted figure on this tab was rebuilt from the filings and labelled as derived. Two consequences are load-bearing: the applicable bar (10% versus 25%) is unresolved, which is what P3c and P3d split on; and the missing datapoints recorded against P2 are five consecutive years of adjusted FCF, including SBC and trailing acquisition-average deductions, for rolling 5-year stability, and a complete rolling five-year adjusted FCF series with SBC and acquisition-spend data.
• The adjusted series cannot reach far enough back. FY2018 acquisition spend is not in the corpus, so adjusted FCF is computable only for FY2023-FY2025, and the FY2023 window is distorted by the $15,083 million First Data purchase. The feature file also records a five-year acquisition average of zero for FY2022-FY2025, which the filings contradict.
• The revenue series in the feature file stops at FY2021, because the structured income file carries no revenue for FY2022-FY2025. The disqualifier flags were computed on that truncated window; extending the series from the filings does not change either flag.
• The quarter that tests the whole recovery has not printed. No Q2 FY2026 results exist as of 31 July 2026 — the latest statements are for the quarter ended 31 March 2026 — so the second-half margin and revenue inflection that both guidance and consensus depend on has no reported data point, and repurchase activity through the 22 June 2026 low is unobservable.
• Web research was unavailable for this run (the search provider returned insufficient credit). No third-party record of the pre-reset FY2026 consensus path could be gathered; the option-chain and implied-volatility facts were sourced directly from Stock Options Channel, Barchart and AlphaQuery with publication dates recorded, which is why I1 is recorded not verifiable rather than resolved.
• Consensus history is short and partial. The estimate feed's revision history extends only 180 days, to 31 January 2026 — three months after the reset — and covers only FY2027 and FY2028, so the consensus path through the decisive 2025 revisions cannot be reconstructed; the company's own guidance midpoints stand in for it. Consensus free-cash-flow means carry no analyst count, so the breadth behind the forward yield path cannot be verified.
• Positioning data is thin. The official short-interest feed returned zero rows for FISV, plausibly a consequence of the November 2025 ticker change, and third-party trackers disagree materially; only the qualitative conclusion — short positioning under about 4% of shares outstanding — is supportable. Institutional flow is observable only at the 5% Schedule 13G threshold and at proxy dates, so most of the 2.50 billion shares traded since the peak cannot be attributed to identified holders.
• Competitive and share data comes from the company and its peers, not from an independent source. Fiserv's own Form 10-K names no specific competitors, only categories; every named competitor here comes from peer filings. All market-share percentages are Fiserv's own Investor Day estimates produced by a proprietary methodology, and the corpus holds no independent third-party share series for either merchant acquiring or core processing. Peer price-to-sales comparisons are not like-for-like at the revenue-definition level.
• Two disclosure-level gaps remain open. No China revenue or asset figure is disclosed anywhere, so the S1 conclusion is an inference from silence rather than a disclosed zero; and Fiserv publishes no adjusted-EBITDA reconciliation in the FY2025 Form 10-K, so the EBITDA behind the 3.12-times leverage reading is built from reported operating income plus depreciation and amortisation while management's own sub-3.2-times measure uses an adjusted figure the filings do not reconcile.