JPMorgan Chase & JPM

Reference table

1. What the four segments are, and what they earn

JPMorgan Chase reports three lines of business plus Corporate. The 2025 Form 10-K (filed 2026-02-13) gives net income by segment, which is the number that matters for a bank and which the vendor payload does not carry at all.

Two things the table says immediately. First, firmwide net income FELL 2% in FY2025, and the reason is entirely in the Corporate row: FY2024 Corporate carried the $7.9 billion Visa gain. Section 2 is about the fact that the same thing has just happened again. Second, the Commercial & Investment Bank is now half the firm's earnings — JPMorgan is closer to a capital-markets business with a very large deposit franchise attached than to a lender.

The vendor's seg_prod block AGREES with the filing on revenue, which is worth stating because it usually does not. For FY2025 it carries Consumer & Community Banking $76.029B, Commercial And Investment Bank $78.454B, Asset and Wealth Management $24.073B, Corporate $7.025B and reconciling items −$3.134B — summing to $182.447B, matching the 10-K's managed net revenue of $182.4B to within a rounding error. Note what that proves: the segments sum to net revenue of $182.4B, not to the vendor's own inc_a.revenue line of $279.745B, which is gross interest income plus noninterest revenue. Any ratio built on that $279.7B figure — priceToSalesRatio, evToSales — is measuring a different quantity from the one the Firm reports (Section 8).

A comparability break in seg_prod. FY2023 and earlier use the old four-segment structure (Consumer & Community Banking, Corporate & Investment Bank, Asset & Wealth Management, Commercial Banking); FY2024 onward use the merged Commercial & Investment Bank. The series is not continuous across 2023-24 and no growth rate should be computed across that boundary. seg_geo is populated and sensible — FY2025 North America $139.689B, EMEA $24.478B, Asia Pacific $14.065B, Latin America $4.215B — though the labels were also renamed between FY2024 and FY2025.

Q1'26 by segment (10-Q, filed 2026-05-01): CCB net income $5.0B (+12%) on $19.6B; CIB net income $9.0B (+30%) on $23.4B, with investment-banking revenue up 38% and advisory fees up 82%, ranked #1 globally for IB fees per Dealogic; AWM $1.8B (+12%) on $6.4B; Corporate $699M against $1.7B, the fall driven by the absence of a $588M First Republic gain. Firmwide Q1'26 net income $16.5B, up 13%, EPS $5.94, ROE 19%, ROTCE 23%.

2. The 36% beat — what it actually was

This is the analytical core of the dive, and it resolves cleanly.

Q2'26 was reported 2026-07-14 at EPS of $7.59 against a $5.59 estimate — a 36.0% beat. A firm of this size, covered by 61 analysts, does not out-earn consensus by 36% on operating momentum. So the question is what the item was.

The Firm disclosed it ten weeks in advance. From the Q1'26 10-Q, Recent events (and again in Note 2):

> "On April 13, 2026, Visa commenced an exchange offer expiring on May 8, 2026 for any and all outstanding shares of Visa Class B-1 common stock and Visa Class B-2 common stock... The Firm has tendered its 18.6 million Visa B-2 shares, and that tender is pending Visa's acceptance. Upon acceptance by Visa of the Firm's tender, the Visa C shares received by the Firm would be recognized at fair value, which is expected to result in a gain that may be recorded as early as the second quarter of 2026."

Note 2 adds the mechanism that makes the gain nearly all profit: the Visa B-2 shares "are held at their nominal carryover basis." A near-zero-cost asset marked to fair value is a gain almost equal to its value.

And the Firm has done this before, at a size that anchors the estimate. The 2025 Form 10-K explains FY2025's noninterest revenue as being held back by "the absence of the $7.9 billion net gain related to Visa shares recorded in the second quarter of 2024."

Now the payload arithmetic, which independently reproduces exactly that shape. The vendor's revenue for a bank is gross interest income plus noninterest revenue, so noninterest revenue is recoverable as revenue − interestIncome, and net revenue as revenue − interestExpense. Both reconcile to the filing for Q1'26 (derived NIR $24.470B against the 10-Q's "$24.5 billion"; derived NII $25.366B against "$25.4 billion"), which validates the method before applying it to Q2.

Read the row that matters: net interest income moved $145 million and noninterest revenue moved $7.366 billion. Expenses rose $1.8B, the tax rate went up 360 basis points, and the provision did not move at all.

The four candidate explanations for a beat this size, tested:

1. Reserve release — the single most common driver, and the one the brief flagged first. Ruled out. The derived provision was $2.515B against $2.507B in Q1'26. A release would show the provision falling sharply, not being flat to the eighth of a percent. For scale, the FY2025 provision averaged $3.55B a quarter.

2. Tax item — ruled out. The effective rate rose from 19.5% to 23.1%. A tax benefit would cut it.

3. Clean operating outperformance — partly, but nowhere near enough. Year-on-year, noninterest revenue rose $10.163B (+46.9%); take out $7.366B and the residual is roughly $2.8B of genuine year-on-year fee growth, about +13%. Real, healthy, and consistent with CIB's +30% Q1 net income. Not 36%.

4. A one-off, non-operating gain — the disclosed Visa exchange. Right size, right quarter, right line, disclosed in advance, with an identical precedent at $7.9B.

The EPS bridge. Tax-effecting the $7.366B step-up at the quarter's 23.1% rate gives roughly $5.66B after tax, or about $2.05 per share. So:

> $7.59 reported − ~$2.05 one-off ≈ $5.50-5.60 "clean" — against a $5.59 consensus.

A 36% beat becomes an in-line quarter. The underlying business did not disappoint; it simply did not surprise.

Two honesty caveats, both material. First, no filing in this set reports the Q2'26 results at all. The 8-K of 2026-07-14 is an Item 7.01 Regulation FD filing furnishing investor-presentation slides; the extracted text contains the cover page and forward-looking-statement language and not a single financial figure — the exhibit itself is not in the text (Section 8). There is no Q2'26 10-Q in the set. The most recent full quarterly filing is Q1'26, balance sheet as of 2026-03-31. Every Q2'26 number in this dive is a vendor figure derived from the earnings release, and is treated as less reliable than the 10-Q and 10-K figures throughout. Second, the $7.366B is a quarter-on-quarter step, so it necessarily bundles the Visa gain with whatever seasonal and Markets variation also occurred. The attribution is strongly supported, not filing-confirmed; the confirming document will be the Q2'26 10-Q, which we do not have.

Three consequences that flow straight into the verdict. (a) The FY2026 consensus of $24.64 contains roughly $2.05 of gain, so the clean FY2026 base is about $22.58 and the true forward multiple is ~15.8x, not 14.5x. (b) The street already knows: its Q3'26 estimate is $5.83, a deliberate step down from $7.59. (c) The 8.6-point three-month outperformance against the S&P is a re-rating on a quarter that, cleaned, was in line.

3. Growth — quality high, rate ordinary

Firmwide, filed: FY2025 net income $57.0B, down 2%, EPS $20.02, on managed net revenue of $182.4B, up 3%. Q1'26 net income $16.5B, up 13%, EPS $5.94, on net revenue of $49.8B, up 10%.

Net interest income and the guide. FY2025 NII was $95.4B, up 3%; NII excluding Markets was $92.6B, flat. Management's FY2026 outlook, given 2026-01-13 and reaffirmed 2026-04-14, is NII of approximately $103B (+8%) and NII excluding Markets of approximately $95B (+2.6%). The entire guided acceleration is Markets net interest income — the most volatile and least durable NII line in the firm, driven by trading-book funding rather than the deposit franchise. That distinction is management framing and is half-weighted here; it is also the single most useful number in the guide.

The balance-sheet engine is growing. Average loans $1.4 trillion FY2025 (+6%), rising to $1.5 trillion in Q1'26 (+11%). Average deposits $2.5 trillion FY2025 (+5%), rising to $2.6 trillion in Q1'26 (+7%), with inflows in Payments and Securities Services, new-account growth in CCB and AWM, and growth in the international consumer initiatives.

Consensus forward EPS (epsAvg, the one estimate row the data contract certifies as reliable):

The +1.4% FY2026-to-FY2027 line is an artifact of the one-off sitting in the FY2026 base. Cleaned, consensus models about 10.7% then 8.3% — mid-to-high single digits into low double digits. That is the honest growth rate of this company, and it is why the growth score is a 5.

Tangible book value per share — the compounding measure that matters for a bank. Filed: $107.56 at 2025-12-31, up 10.5%; $108.87 at 2026-03-31, up 8% year-on-year. That is the rate at which intrinsic value accretes, and it is the primary driver of the return bridge in Section 6b.

4. Balance sheet, capital, and why enterprise value does not exist here

A bank is not an industrial, and the vendor's balance-sheet apparatus fails on it completely. This section states the rejection and then does the work properly.

The rejection, by name. The vendor reports enterpriseValueTTM of $1,886,038,415,200 — $1.886 TRILLION — against a $958.0B market capitalisation, i.e. it adds roughly $928B of "net debt." Deposits are a bank's raw material, not leverage. A bank funds itself with customer deposits by design; treating that funding as debt to be added to equity value is a category error, and there is no repair that makes it meaningful. Note also that the vendor's implied net debt of ~$928B does not even reconcile to its own FY2025 balance sheet, which reports totalDebt of $942.378B and netDebt of $599.040B, and that JPMorgan's $2.6 trillion of deposits do not appear in either figure — if they did, "EV" would exceed $3.5 trillion. The number is wrong twice over and unusable in any case.

Rejected explicitly and by name: enterpriseValueTTM, evToSalesTTM (6.337x), evToEBITDATTM (20.762x), enterpriseValueMultipleTTM (20.762x), netDebtToEBITDATTM (10.216x), evToOperatingCashFlowTTM and evToFreeCashFlowTTM (13.381x) — all conceptually void for a depository. Also rejected: currentRatioTTM and quickRatioTTM (both 0.8404), because banks do not present classified balance sheets and the vendor has mapped $2.729 trillion of deposits into "other current liabilities"; workingCapitalTTM (−$601.2B), investedCapitalTTM (−$499.2B), netCurrentAssetValueTTM (−$1,475.5B), grahamNetNetTTM (−$577.73), daysOfSalesOutstandingTTM and cashConversionCycleTTM (665.2 days) — every one an artifact of applying manufacturing arithmetic to a balance sheet of loans and deposits. And interestCoverageRatioTTM of 0.854, which for a company whose interest expense is its cost of goods reads like distress and means nothing.

What to use instead — the regulatory capital stack, from the filings:

CET1 at 14.3% against an 11.5% requirement is roughly $57B of excess CET1 at the Q1'26 risk-weighted asset base — the source of the buyback. The stress capital buffer is 2.5%, in effect through 2027-09-30.

Deriving tangible book value per share ourselves, as the method requires, from the FY2025 balance sheet:

• Total stockholders' equity $362.438B
• less preferred stock $20.045B → common equity $342.393B
• less goodwill $52.731B and other intangibles $11.727B ($64.458B) → tangible common equity $277.935B
• ÷ approximately 2,602M period-end common shares → ~$106.8

The Firm's filed figure is $107.56, the ~0.7% difference being the deferred-tax-liability add-back the Firm's own definition applies ("less goodwill and identifiable intangible assets other than MSRs, net of related deferred tax liabilities"). The derivation cross-checks the filing.

And it exposes a vendor error that must be flagged. bookValuePerShareTTM of $134.327 and tangibleBookValuePerShareTTM of $111.268 both fail to deduct the $20.045B of preferred stock. The test is arithmetic: tangibleAssetValueTTM is $310.294B, which equals $111.2683 × 2,788,700,000 exactly — and $310.294B is total equity less goodwill and intangibles, preferred included. The same series divides by a weighted-average share count rather than period-end common shares, which the Firm's own TBVPS definition requires. Both are errors; they happen to run in opposite directions and partly offset. Correcting on the vendor's own equity base — $310.294B less $20.045B of preferred = $290.25B of tangible common equity — and dividing by roughly 2,570-2,600M period-end shares gives a derived Q2'26 TBVPS of about $112, against the last filed figure of $108.87. This dive anchors on $108.87 (filed, hard) and ~$112 (derived, labelled) and uses the vendor's $111.268 for nothing.

5. Cash flow — the section that does not apply

For a bank, operating cash flow is not a measure of anything distributable, and the vendor's cash-flow-derived ratios must be rejected outright. The evidence is in the series itself:

Operating cash flow swung from −$42.0B to +$100.9B in a single year while net income moved from $58.5B to $57.0B. The swing is trading-book and working-capital movement — securities lending, repo, receivables — not earnings. A metric that moves $143 billion while earnings move $1.5 billion is measuring balance-sheet mechanics, not profit.

Accordingly freeCashFlowYieldTTM (14.71%), priceToFreeCashFlowRatioTTM (6.80x), priceToOperatingCashFlowRatioTTM (7.07x), freeCashFlowToEquityTTM ($315.6B), freeCashFlowToFirmTTM ($217.1B), incomeQualityTTM (2.404) and operatingCashFlowPerShareTTM ($50.54) are all rejected. A 14.7% "free cash flow yield" on a bank is not a bargain signal; it is noise. Note also that capitalExpenditure is $0 in every year, so "free cash flow" is definitionally identical to operating cash flow — JPMorgan does spend on premises and technology, but the vendor does not populate the line.

What is genuinely distributable is capital return, and that is knowable. FY2025: $34.591B of repurchases (3.61% of today's market capitalisation) and $16.625B of common dividends (1.74%) — a combined shareholder yield of roughly 5.3%. The Board declared a $1.50 quarterly dividend on 2026-03-17, $6.00 annualised, a 1.68% yield at $357.52. A $50 billion repurchase authorisation was announced effective 2025-07-01, replacing a $30B programme. FY2025 diluted shares averaged 2,793.7M against 2,879.0M in FY2024 — a 3.0% reduction.

One caution on the buyback that follows directly from Section 4: every dollar repurchased above tangible book value dilutes tangible book value per share. At 3.2x tangible book, $34.6B of repurchases retires about $10.8B of tangible common equity and $34.6B of capital — the arithmetic is far less friendly than it was for MKL buying at 1.11x book the same day. The buyback here supports EPS; it does not build tangible book.

6. Valuation — priced in or room?

At $357.52 (market cap $958.0B, ~2.68B shares):

The residual-income frame, worked. Justified price-to-tangible-book = (ROTCE − g) ÷ (cost of equity − g). At a 10% cost of equity and 5% long-run growth:

Invert it against the actual price. At 3.19x derived tangible book, the same frame implies a sustained ROTCE of 20.95%. Filed FY2025 ROTCE was 20%; Q1'26 was 23%; the ex-Visa Q2'26 run-rate is roughly 21%. The multiple and the return are at equilibrium. This is the entire verdict.

Estimate coverage is adequate: 9, 12 and 8 analysts on the EPS lines and 8, 9 and 5 on revenue — far better than MKL's 2-5. But the est revenue line runs on a different basis from the vendor's actuals: FY2025E revenue of $183.544B matches the 10-K's managed net revenue of $182.4B, while inc_a.revenue for FY2025 is $279.745B (gross). Forward revenue estimates are on the net basis and must never be compared to the vendor's actual revenue line. The est EBITDA/EBIT rows are corrupt in the FY2021-FY2023 rows exactly as the data contract warns — FY2023 shows EBIT of $117.651B against EBITDA of $64.698B, which is impossible — and EBITDA is meaningless for a bank in any case (Section 8). All forward valuation here runs on epsAvg and on tangible book.

Peer context. The vendor peer set is genuinely relevant — Bank of America ($446.4B), Citigroup ($234.7B), Wells Fargo ($267.3B), HSBC ($364.4B), Royal Bank of Canada ($289.8B), Toronto-Dominion ($201.7B) and three other Canadians, plus Nu Holdings ($69.4B) as the digital-bank outlier. JPMorgan at $958.0B is more than twice the next-largest US peer. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn — which matters, because whether 3.2x tangible book is rich depends heavily on what BAC, WFC and C trade at, and this data set cannot answer that.

6a. What today's price assumes (the inversion)

At $357.52 — 3.19x derived tangible book and 14.3x FY2027 consensus EPS — the price embeds the following falsifiable claims:

• Return on tangible common equity is sustained at roughly 21%, indefinitely. (Derived: residual income at a 10% cost of equity and 5% growth gives an implied ROTCE of 20.95% at a 3.19x multiple.) Filed FY2025 ROTCE was 20%; Q1'26 was 23%. This is the most fragile assumption in the price and it is named as such. It is fragile not because 21% is implausible today but because it embeds no cycle: at 17% the justified multiple is 2.40x and fair value is $276.
• FY2026 EPS of $24.64 is delivered, and the market does not discount the roughly $2.05 of it that is the Visa exchange gain. (Consensus, 9 analysts; the $2.05 is our number, derived in Section 2.) Falsifier: the Q3'26 print on 2026-10-13 against the $5.83 estimate. The street is already honest here — $5.83 is a deliberate step down from $7.59 — so the risk is that price, not consensus, has extrapolated.
• Net interest income reaches approximately $103 billion in 2026. (Management guidance, 2026-01-13 and 2026-04-14 — half-weighted as management framing.) That is +8% on FY2025's $95.4B, but NII excluding Markets is guided to only ~$95B, +2.6%. The price assumes Markets NII — the least durable line — carries the year.
• Credit stays benign through 2027: Card Services net charge-offs around 3.4%, allowance coverage stable near 1.82%, no reserve-build cycle. (Guidance plus filed metrics.) Against this: FY2025 provision $14.2B vs $10.7B, net charge-offs $9.8B (up $1.2B), a $4.4B allowance build, NPAs up 11%. Credit is normalising, not deteriorating — but the price has no room for it to deteriorate.
• Capital return continues at roughly $34B a year of buyback with no regulatory setback. (FY2025 actual, against the $50B authorisation.) Falsifier: the Basel III Re-Proposal and GSIB Surcharge Re-Proposal, which the Firm's own 10-Q estimates would together raise required CET1 capital by approximately 4-6%.

6b. The return bridge (why the multiple moves)

Expected return over 12-24 months decomposes as tangible-book-value-per-share growth (+8-10%, filed rate) + multiple drift (modest COMPRESSION assumed, 3.19x → ~3.05x, roughly −4%) + dividend yield (1.68%) — call it 5-7% a year total.

We assume the multiple compresses slightly, and the reasoning is explicit. 3.19x prices a 21% ROTCE as permanent. The 21% is being earned at a point when capital-markets revenue is exceptional (CIB net income +30% in Q1'26, advisory fees +82%), when credit costs are benign, and before the Basel re-proposals bite. We are not forecasting a collapse — we are declining to capitalise a cycle peak into perpetuity. A 3.05x exit on 12-month-forward tangible book of about $122 gives $372.

Note what this bridge does NOT contain: any re-rating. Almost the whole expected return is tangible-book compounding. If the multiple simply held at 3.19x against $122 of forward tangible book, the price would be $389 — 8.8% above spot — so the base case is actually conservative relative to a flat multiple. We take the conservative leg deliberately, because the buyback runs against us here. At 3.2x tangible book, $34.6B a year of repurchases retires capital at more than three times its tangible carrying value; that supports EPS but suppresses tangible-book growth, which is why the observed 8-10.5% compounding is slower than a 20% ROTCE alone would produce.

The bull case at $422 requires the multiple to expand to 3.40x — i.e. the market capitalising a 22% ROTCE. That leg is fragile and is not needed for the base. The bear case at $276 requires only that ROTCE normalise to 17%, with no crisis and no credit event. That asymmetry — a bull that needs a re-rating and a bear that needs only mean reversion — is the honest shape of the risk here.

6c. Variant perception (where we differ, what would surprise)

On the valuation itself, we have essentially NO variant perception, and per the expectations addendum that is a verdict input rather than an omission. Our $372 base sits between the street's $367.83 consensus and $370 median, inside a $305-$420 range, against a 1/31/27/2 rating distribution. Two independent methods — a residual-income multiple on tangible book and a P/E on FY2027 consensus — converge on 3.05x and 14.9x, which is where the crowd already is. When our number and consensus agree to within 1%, there is no edge, and a no-edge name defaults to Hold.

Where we do differ is on quality of information, not on level:

• We differ on what the 36% beat means. Every screen in the market now shows JPMorgan with a 36% earnings surprise and a 15.4x trailing multiple. Cleaned, the quarter earned about $5.50-5.60 against a $5.59 estimate, and the trailing multiple is ~16.9x. The 8.6-point three-month outperformance against the S&P is a re-rating on an in-line quarter. Watchable number: Q3'26 EPS on 2026-10-13 against $5.83 — if it lands near consensus without another one-off, the market will have paid for a beat that was not there.
• We differ on the forward multiple. Consensus FY2026 EPS of $24.64 embeds roughly $2.05 of Visa gain, making the honest forward multiple ~15.8x, not 14.5x. Anyone comparing JPMorgan to peers on 14.5x is comparing a gain-inflated figure to clean ones. Watchable number: whether analysts strip the item from their FY2026 bases when they publish FY2027 revisions after the Q3 print.
• We differ on the whole vendor valuation apparatus. Screens show this bank with a $1.886 trillion enterprise value, a 10.2x net-debt-to-EBITDA and a 14.7% free-cash-flow yield. All three are void (Sections 4, 5, 8). This is not a correction to a number; it is a rejection of a frame — and it means quantitative screens rank this name on quantities that do not exist.
• We differ on what the knowledge-base lane says. A naive read gives 63 claims at 42 bullish to 11 bearish. Only 21 are about the equity at all, 34 are monetary-system arguments that merely name the firm, and 8 are false positives (Section 7). The corrected lane is thin, momentum-derived and 113 days stale.
• Positive surprise that would force a re-rate: a materially softer final Basel III / GSIB outcome than the Firm's own approximately 4-6% CET1 estimate, which would free capital, lift both ROTCE and the buyback, and justify 3.4x-plus. Also: NII excluding Markets beating the ~$95B guide, which would show the deposit franchise re-pricing rather than the trading book carrying the year. Watchable numbers: the CET1 requirement and NII ex-Markets.
• Negative surprise that would break the thesis: credit normalising faster than guided — Card Services net charge-offs above 3.4%, allowance coverage rising materially off 1.82%, or a provision quarter above $4B. Or a capital-markets slowdown, since the CIB is now 48.8% of firm net income and its Q1'26 growth was 30%. Either takes ROTCE toward 17% and the multiple toward 2.4x.

Synthos fair values

All three anchors are multiples of tangible book value per share, cross-checked against FY2027 consensus EPS of $24.99, and each is stated with the ROTCE it implies.

• Bear ~$276 — 2.40x tangible book of ~$115 (slower compounding in a downturn). Implies ROTCE of 17% at a 10% cost of equity and 5% growth. Cross-check: 11.5x a recessionary FY2027 EPS of ~$24. The scenario: no crisis, simply mean reversion — capital-markets revenue normalises from an exceptional 2025-26, Card net charge-offs run above 3.4%, the allowance is built, and Basel re-proposals land at the top of the 4-6% CET1 range. Note the 52-week closing low is $282.84, so this level was trading within the last year. −22.8%.
• Base ~$372 — 3.05x 12-month-forward tangible book of ~$122 (8-10% compounding off the derived $112). Implies ROTCE of ~20.3%. Cross-check: 14.9x FY2027E EPS of $24.99, essentially today's forward multiple. The scenario: the bank keeps earning 20%, tangible book compounds at its filed rate, the multiple gives back a little as the cycle peak is recognised. +4.0%, plus a 1.68% dividend. Note a flat 3.19x multiple on the same $122 would give $389 (+8.8%) — the base is deliberately the conservative leg.
• Bull ~$422 — 3.40x forward tangible book of ~$124. Implies ROTCE of 22% sustained. Cross-check: 16.9x FY2027E EPS, and — a useful external check — the street's own high target is $420. The scenario: Basel relief frees capital, ROTCE holds above 22%, AWM and Payments keep shifting the mix toward fee revenue, and the market pays a durable-franchise premium. +17.9%.

Base is 4.0% above spot; asymmetry is 0.78:1 — negative (17.9% up against 22.8% down). A great business, a fair price, and a skew that does not pay you to be early. Hold.

7. Knowledge base — 63 claims, and why only 21 of them count

Raw hits naming JPMorgan: 63, across 21 distinct sources, spanning 2019-08-29 to 2026-04-13. Quarantined (quarantine_misattributed): 0. Discarded as false positives: 8. Excluded from conviction as monetary-system rather than equity claims: 34. Claims that actually bear on the equity: 21.

Aliases searched: JPM, JPMorgan, JP Morgan, JPMorgan Chase, entity-exact and case-insensitive, across all 51,928 records.

The first thing to say is that the lane has been SILENT for 113 days. The most recent claim of any kind is 2026-04-13. Nobody we track has said anything about this company through the Q2 print, the run to a 52-week high, or the Visa gain. A conviction lane that has not spoken during the only interesting quarter of the year is not conviction; it is an archive.

The second is the composition problem the raw count hides. The largest contributor is arthur_hayes at 18 claims — 29% of the lane — and on reading them, 16 of the 18 are monetary-system or crypto-market arguments that name JPMorgan as an actor in the plumbing, not as an investment. Representative:

> 2025-12-01 · arthur_hayes · bullish · conviction 65

> "Credit creation shifts from Fed to commercial banks in 2026—industrial-policy-backed lending (e.g. JPM's $1.5T) becomes the new liquidity source."

The same idea recurs on 2025-10-15, 2025-11-27, 2025-12-06 and 2026-04-10 — five near-identical restatements counted as five claims. It is an argument about system liquidity flowing to crypto. It is not a view on owning the stock, and it is not counted as conviction on the equity. The same applies to the stablecoin-rail cluster from forward_guidance (6 claims), natalie_brunell (6) and raoul_pal (2). 34 claims in total fall in this bucket.

Eight claims are outright false positives — JPMorgan appears only as the publisher of a research call, not as a company: luke_gromen on "JP Morgan's $8,500 [gold] target" (2026-02-02), natalie_brunell on "JP Morgan calls Bitcoin undervalued with ~$165k fair value" (2025-10-14), lex_fridman on JPMorgan's Argentina growth projections (2024-11-19), andreas_steno citing a JPMorgan macro note (2023-09-03), odd_lots on the zero-DTE "doom loop" research (2023-03-24), forward_guidance on silver-manipulation folklore (2025-04-04), and two arthur_hayes claims about a MicroStrategy index-inclusion debate where the only JPMorgan content is a remark by James Dimon (2025-12-04, 2025-12-19). All eight discarded.

The 21 equity claims, verbatim. The bulls first:

> 2026-01-13 · compound_and_friends · bullish · conviction 85 · entities: JPMorgan, JPM

> "JPMorgan is extraordinarily profitable ($150M/day, $57.5B FY2025) and dominant; a long-term shareholder shrugs off a boring call and post-earnings dip."

The strongest and most specific claim in the lane, and it checks out against the filing: the 10-K reports FY2025 net income of $57.0B against the claim's $57.5B.

> 2023-07-27 · odd_lots · bullish · conviction 75 · entity: JPMorgan

> "Mega banks win on scale — diversified fee revenue, less spread-dependence, and a widening tech/AI/talent gap; they were the deposit-flight beneficiaries and can outspend community banks."

> 2026-04-08 · invest_like_the_best · bullish · conviction 70 · entities: JPMorgan, Jamie Dimon

> "Factory-model firms need elite risk management; imitators chasing Jamie Dimon's scale rarely match his risk-management discipline."

> 2023-03-24 · warren_pies · bullish · conviction 75 · entity: JPMorgan

> "JPMorgan is the strongest institution in the world — proven by absorbing Bear Stearns; strongest banks can backstop failing peers."

> 2023-04-03 · arthur_hayes · bullish · conviction 75 · entity: JP Morgan

> "JP Morgan focused on shorter-duration assets, is basically unaffected, and benefits from deposit flight into big banks."

> 2025-07-06 · jordi_visser · bullish · conviction 70 · entities: BKX, JPM

> "Banks are the most important sector to watch; rate cuts with no recession are extremely bullish for financials, plus JP Morgan's AI-driven efficiency push. Watching JPM."

> 2026-01-20 · compound_and_friends · bullish · conviction 72 · entities: Bank of America, JP Morgan

> "Bank earnings are the economic source of truth; net charge-offs declining and credit-card stress absent — the awaited consumer rollover isn't here yet."

Then the concentration problem. jordi_visser supplies 7 of the 21 equity claims, and 5 of those are one idea restated across three days:

> 2025-11-16 · jordi_visser · bullish · conviction 75 · entities: JPM, GS

> "JPMorgan making all-time highs Wednesday; banks, especially JPM and Goldman, benefit most from M&A, IPOs; financials a major AI margin beneficiary next year."

with near-duplicates on 2025-11-16 (conviction 72), 2025-11-16 again (75), 2025-11-15 (65) and 2025-11-17 (80), plus 2025-12-07 (60): "As long as banks — Goldman and JP Morgan in particular — sit near all-time highs, credit/bubble fears are unfounded." These are momentum claims. The stated reason to be bullish is that the stock is making all-time highs. That is a circular signal at best, and today — with the stock again at a 52-week high — it is precisely the configuration the bearish claims name.

The bears, and they are the ones describing today:

> 2025-09-14 · brent_johnson · bearish · conviction 62 · entity: JP Morgan

> "Financials in general will struggle in Q4; JP Morgan is at its high and wouldn't be surprising to see it roll over."

> 2025-08-03 · brent_johnson · bearish · conviction 55 · entity: JPMorgan

> "The overbought rollover isn't just tech; JPMorgan showed the same relentless move higher then a severe late selloff."

> 2025-09-15 · all_in · bearish · conviction 65 · entities: JPMorgan, Visa, Mastercard

> "Incumbents have regulatory muscle and scale but are slow to adopt tech and can't hire top talent, losing to nimble fintechs."

> 2023-05-19 · jim_bianco · bearish · conviction 72 · entities: Western Alliance, JPMorgan

> "A 'bank walk' (deposit flight to money-market funds) pinches bank net-interest margins and could cumulatively trigger a credit crunch by year-end."

And the most interesting claim in the lane, which is tagged bullish but is bearish on this equity in content:

> 2025-10-11 · arthur_hayes · tagged bullish · conviction 85 · entities: Tether, USDT, USDC, Circle, JPMorgan

> "Stablecoins are a new open banking license disrupting banks; a 150-person Tether worth half a trillion vs JPMorgan's 315,000 employees shows the coming valuation collapse."

This is a stance-tagging error and it is worth naming. The claim explicitly forecasts a valuation collapse for JPMorgan-type incumbents and is stored as bullish. The same content, from a promoter-class voice at low weight, appears as anthony_pompliano 2026-01-20 (neutral, conviction 38): "JP Morgan's ~300,000 employees could be increasingly displaced by AI agents/algorithms." This is the strongest long-horizon bear argument available on the name and the raw stance counts hide it. A second tagging inversion runs the other way: arthur_hayes 2026-03-05 is tagged bearish while its content — "deposits flee to government-guaranteed giants like JPMorgan; big banks are fine" — is bullish for this equity.

What the corrected lane establishes. Equity claims run 16 bullish / 4 bearish / 1 neutral across 11 independent voices (compound_and_friends, odd_lots, invest_like_the_best, warren_pies, arthur_hayes, jordi_visser, brent_johnson, all_in, jim_bianco, michael_howell, anthony_pompliano). No management or _mgmt voices appear, so the lane is fully independent — that is a genuine strength relative to names dominated by their own chief executive. But it is shallow (the durable structural bull case rests on two claims, from 2023), momentum-derived (a third of the bullish weight is one voice saying "it is at all-time highs"), contradicted at the top of conviction (the 85-conviction claim is a disruption thesis), and 113 days stale. Mixed, and rated as such.

8. Data integrity — what we rejected from the vendor file and why

Eleven material items. This is the longest such list in the batch, because a bank breaks more of the vendor's apparatus than any other business type.

1. The 8-K of 2026-07-14 is NOT the earnings release, and there is no Q2'26 10-Q — the single most important gap. The brief anticipated this filing would carry the Q2 numbers. It does not. It is an Item 7.01 Regulation FD filing whose entire substance is: "On July 14, 2026, JPMorganChase held an investor presentation to review 2026 second quarter earnings. Exhibit 99 is a copy of slides furnished for, and posted on the Firm's website." The extracted text contains no financial figure of any kind — the exhibit itself is not in the file. The most recent full quarterly filing in the set is the Q1'26 10-Q, balance sheet as of 2026-03-31. Every Q2'26 figure in this dive therefore comes from the vendor payload, is treated as release-grade rather than review-grade, and is labelled as such wherever used. data_as_of.fundamentals is set to 2026-03-31, not 2026-06-30, for this reason.

2. The 8-K of 2026-07-23 is a by-laws amendment, not a capital action. The brief hypothesised a dividend declaration or post-stress-test capital action. It is neither. Item 5.03: the Board adopted an amendment to Section 2.03 of the by-laws, effective 2026-07-21, providing that a Lead Independent Director, if any, shall be appointed by the non-management directors. Governance housekeeping with no financial content. Separately, _index.json lists two distinct 8-K accessions both dated 2026-07-23 (0001193125-26-314128 and 0000019617-26-000288) written to the same file path — only the second survives on disk. One 8-K in this set was overwritten and is unavailable. Flagged, not worked around.

3. Enterprise value and every EV-derived multiple — REJECTED ON CONCEPT, not corrected. enterpriseValueTTM is $1,886,038,415,200, implying ~$928B of net debt on a $958.0B market capitalisation. Deposits are a bank's raw material, not leverage; enterprise value has no meaning for a depository. The figure is also internally inconsistent: the FY2025 balance sheet reports totalDebt $942.378B and netDebt $599.040B, neither of which produces $928B, and JPMorgan's $2.6 trillion of deposits appear in none of them. Rejected: enterpriseValueTTM, evToSalesTTM (6.337x), evToEBITDATTM (20.762x), enterpriseValueMultipleTTM (20.762x), netDebtToEBITDATTM (10.216x), evToOperatingCashFlowTTM and evToFreeCashFlowTTM (13.381x). No corrected EV is offered because none exists.

4. bookValuePerShareTTM ($134.327) and tangibleBookValuePerShareTTM ($111.268) do not deduct preferred stock — REJECTED, filing used instead. JPMorgan carries $20.045B of preferred stock (FY2025 balance sheet), and both per-share figures include it. The proof is arithmetic: tangibleAssetValueTTM is $310.294B, exactly $111.2683 × 2,788,700,000, and $310.294B is total equity less goodwill and intangibles. Both also divide by a weighted-average share count rather than period-end common shares, contrary to the Firm's own stated TBVPS definition. The 10-K reports tangible book value per share of $107.56 at 2025-12-31 and the 10-Q reports $108.87 at 2026-03-31; the vendor implies $111.27. We use the filings. Correcting the vendor's own base (remove $20.045B of preferred → $290.25B of tangible common equity) gives a derived Q2'26 figure near $112, used only where a post-Q1 estimate is required and always labelled. priceToBookRatioTTM (2.662x) and shareholdersEquityPerShareTTM ($134.327) are rejected for the same defect.

5. currentRatioTTM and quickRatioTTM (both 0.8404) — REJECTED as meaningless. Banks do not present classified balance sheets. The vendor has invented one, mapping $2.729 trillion of liabilities (overwhelmingly deposits) into otherCurrentLiabilities and producing totalCurrentLiabilities of $3.585 trillion. A "current ratio" for an institution whose liabilities are demand deposits and whose assets are multi-year loans measures nothing. Rejected alongside workingCapitalTTM (−$601.210B), investedCapitalTTM (−$499.205B), netCurrentAssetValueTTM (−$1,475.497B), grahamNetNetTTM (−$577.73), daysOfSalesOutstandingTTM / operatingCycleTTM / cashConversionCycleTTM (all 665.2 days) and receivablesTurnoverTTM (0.549).

6. All free-cash-flow derivatives — REJECTED. netCashProvidedByOperatingActivities reads −$42.012B (FY2024) then +$100.867B (FY2025) while net income moved from $58.471B to $57.048B. A $143 billion swing against a $1.4 billion earnings move is balance-sheet mechanics, not profit. capitalExpenditure is $0 in every year, so freeCashFlow is identical to operating cash flow by construction. Rejected: freeCashFlowYieldTTM (14.71%), priceToFreeCashFlowRatioTTM (6.797x), priceToOperatingCashFlowRatioTTM (7.074x), freeCashFlowToEquityTTM ($315.614B), freeCashFlowToFirmTTM ($217.123B), incomeQualityTTM (2.404), operatingCashFlowPerShareTTM ($50.541). The real distribution facts are the FY2025 buyback of $34.591B and common dividends of $16.625B.

7. interestCoverageRatioTTM (0.854) — REJECTED. For a bank, interest expense is the cost of the product. A sub-1.0 "coverage ratio" reads as distress on a screen and means nothing here. Same for debtServiceCoverageRatioTTM (0.082), solvencyRatioTTM (0.0155) and debtToEquityRatioTTM (3.305) — the last is not wrong so much as uninformative next to a 14.3% CET1 ratio against an 11.5% requirement.

8. The Q2'26 quarterly record is internally inconsistent and is release-derived — flagged. Three defects in that one row: (a) depreciationAndAmortization is $0 for Q2'26 against $2.364B in Q1'26 and $2.312B in Q4'25, which forces ebitda = ebit = incomeBeforeTax = $27.516B; (b) netIncomeDeductions is $0 and bottomLineNetIncome equals netIncome, where every other quarter carries a preferred-dividend deduction ($335M in Q1'26, $335M in Q4'25, $350M in Q3'25) — JPMorgan pays preferred dividends every quarter; (c) the weighted-average diluted share count is frozen at exactly 2,793,700,000 for five consecutive quarters (Q2'25 through Q2'26), which is the FY2025 annual weighted average back-filled into each quarter. The true Q1'26 diluted count implied by the filed EPS of $5.94 is about 2,720M, so the vendor count is ~2.7% too high. No per-share figure in this dive is recomputed from that share count; reported EPS is used as reported.

9. est EBITDA and EBIT rows are corrupt for FY2021-FY2023, and EBITDA is void for a bank regardless — REJECTED. FY2023 shows ebitAvg of $117.651B against ebitdaAvg of $64.698B; FY2022 shows $106.956B against $58.816B. EBIT exceeding EBITDA is impossible. The FY2024-FY2028 rows are correctly ordered but EBITDA is not a meaningful measure for a bank — depreciation and amortisation are trivial and interest is revenue and cost — so no EBITDA-based figure appears anywhere in this dive. All forward valuation runs on epsAvg, which the data contract certifies and which checks out: FY2025E $20.198 against the filed actual of $20.02.

10. est revenue and inc_a revenue are on DIFFERENT BASES — a trap, flagged rather than reconciled. est.revenueAvg for FY2025 is $183.544B, matching the 10-K's managed net revenue of $182.4B. inc_a.revenue for FY2025 is $279.745B — gross interest income plus noninterest revenue. These are not comparable and no growth rate may be computed across them. Consequently priceToSalesRatioTTM (3.219x) and revenuePerShareTTM ($106.728), both built on the gross figure, are rejected. Note also that seg_prod sums to the NET basis ($182.447B for FY2025), agreeing with the filing — so the segment table and the income-statement revenue line in the same payload disagree by $97 billion by construction.

11. Minor discrepancies between vendor and filing, filing preferred in each case. (a) FY2025 EPS: the vendor reports $20.05 diluted; the 10-K states $20.02. We use $20.02. (b) FY2025 net income: inc_a reports $57.048B, cf_a reports $56.780B, the 10-K states $57.0B — a $268M internal vendor inconsistency; we use the filing. (c) cf_a.preferredDividendsPaid is $0 in every year while netDividendsPaid equals commonDividendsPaid — JPMorgan pays roughly $1.3B of preferred dividends annually, so the split is wrong even though the total appears right. (d) seg_prod FY2023 and earlier use the pre-2024 four-segment structure and are not comparable to FY2024+. (e) The insider block contains a null record (Marianne Lake, 2026-06-25, zero shares, zero price, zero holding, blank transaction type) — discarded.

12. The vendor rating block applies industrial thresholds to a bank — sub-scores rejected. The composite is B, overall score 3, with returnOnEquityScore 5 and discountedCashFlowScore 5 but priceToBookScore 1 and debtToEquityScore 1. A price-to-book score of 1 for a bank earning 20% on tangible equity, and a debt-to-equity score of 1 for an institution funded by insured deposits at a 14.3% CET1 ratio, are the model mis-specifying the industry. The composite is noted; the sub-scores are not used.

Not defects, correctly reported: seg_prod agrees with the 10-K on FY2025 segment revenue to within a rounding error — a genuine and unusual vendor success, and worth recording as such. seg_geo is populated across all years, unlike most names in this batch. The price-target set shows real dispersion ($420 high, $305 low, $370 median, $367.83 consensus) across 61 rated analysts — a genuine consensus, not the single-source artifact seen elsewhere. beta of 0.982 and dividendYieldTTM of 1.678% both reconcile ($6.00 ÷ $357.52). stockBasedCompensationToRevenueTTM of 0 reflects an unpopulated line, not an absence of stock compensation.

9. Technicals — a 52-week high on below-average volume

• Price $357.52, up 1.38% today. This is the 52-week CLOSING high: the computed 52-week closing high equals the last price exactly and the drawdown from peak is 0.0%. The quote's $363.00 52-week high is today's intraday print (the session's high), so the "1.5% below the 52-week high" framing describes a level the stock set itself this afternoon.
• +26.4% above the 52-week closing low of $282.84 (quote basis: $279.10). The stock is at the very top of its annual range, not near it.
• Above both moving averages and well clear: +8.4% over a 50-DMA of $329.79 and +14.5% over a 200-DMA of $312.37, both rising. There is no technical support anywhere near spot; the first is 8.4% down.
• RSI 60.2 — firm, not overbought. MACD +6.81 — positive and widening.
• Volume 8.39M against a 9.43M average. A new closing high on 11% below-average volume is not a conviction breakout.
• Relative performance, and this is the important part: 12-month +23.6% against SPY +24.3% and QQQ +30.8% — the stock has MATCHED the index over a full year and lagged the Nasdaq by 7 points. But 3-month is +16.2% against SPY +7.6%, an 8.6-point outperformance, and 6-month is +16.0% against +11.1%. Essentially the entire year's relative gain was produced in the three months containing the 2026-07-14 print.
• Practical read: this is a momentum configuration built on a quarter that, cleaned of the Visa exchange gain, was in line with consensus. Buying here is chasing the reaction to a beat that was not operating. The knowledge base's own bearish claims describe exactly this setup — "JP Morgan is at its high and wouldn't be surprising to see it roll over" (brent_johnson, 2025-09-14) — as, circularly, do its bullish ones. There is no entry advantage at spot. A pullback to the 50-DMA at ~$330 would put the stock at roughly 2.95x derived tangible book, at which the residual-income frame implies a required ROTCE of about 19.8% against a filed 20% — a genuinely different proposition.

10. Insiders — awards and gifts, no open-market activity

The reading: this block carries no signal in either direction, and should not be presented as if it did. The four director entries dated 2026-06-30 at an identical $327.33 are quarterly deferred-compensation share credits, not discretionary purchases — fractional share counts at a single common price are the signature. Jennifer Piepszak's 60,214 shares at a $0 price is an equity grant, and her post-transaction holding equalling the grant exactly suggests a new or newly-reported position rather than accumulation. The two July entries are gifts, which are dispositions in form but carry no view on value.

What matters is what is absent: there is not a single open-market purchase and not a single open-market sale in the file. For a company with 320,560 employees and a large executive team, an eight-row window showing only awards and gifts tells us the file is a narrow slice, not that insiders are neutral. Contrast MKL the same day, where directors were accumulating at the bottom of the range and nobody sold. Here we simply have no information. One record — Marianne Lake, 2026-06-25, all fields zero or blank — is corrupt and discarded (Section 8).

11. Verdict, kill-criteria and flip conditions

Hold.

This is a genuinely excellent institution trading at a genuinely fair price, and the discipline of the dive is to say so rather than to manufacture a view.

What the business is: the largest bank in the United States, earning a 20% return on tangible common equity (filed, FY2025) and 23% in Q1'26, with CET1 at 14.3% against an 11.5% requirement, $1.5 trillion of liquidity sources, $2.6 trillion of average deposits growing 7%, the number-one global investment-banking franchise by fees, an asset-management arm compounding net income at 20%, and tangible book value per share compounding 8-10.5% while $34.6B of stock is repurchased annually. There is no quality objection to make.

What the price is: 3.19x derived tangible book value per share of ~$112 (3.28x the last filed figure of $108.87), which on a residual-income frame — justified P/TBV = (ROTCE − g) ÷ (COE − g) — already requires a sustained ROTCE of 20.95% at a 10% cost of equity and 5% growth. The Firm earns 20-21%. The multiple and the return meet exactly. Our $372 base is 4.0% above spot; the street is at $367.83. Two methods and the crowd converge on the same number.

What the quarter was: the 36% beat that drove the last three months of outperformance was, on the arithmetic in Section 2, almost entirely a Visa Class B-2 exchange gain the Firm disclosed in its Q1 10-Q on 2026-05-01 — net interest income flat, provision flat, tax rate up, noninterest revenue up $7.366B in one quarter. Cleaned, the quarter earned about $5.50-5.60 against a $5.59 estimate. The street already knows: its Q3 estimate is $5.83, a step down from $7.59. The price may not.

Why Hold and not Buy: no discount, no variant perception, and a negative skew — 17.9% to the bull against 22.8% to the bear, because the bull needs a re-rating and the bear needs only mean reversion in ROTCE from a cycle peak. Why Hold and not Avoid or Watch: nothing here is broken. A 20% ROTCE compounder with a fortress capital position, bought at fair value, still returns tangible-book compounding plus a 1.68% dividend — 5-7% a year — and does so with a 0.982 beta. If you own it, own it. If you do not, this is not the price at which to start.

Pre-registered KILL criteria — what would take this to Avoid:

• ROTCE printing below 17% on a trailing basis, which takes the justified multiple to 2.40x and fair value to $276, 23% below spot. This is the single most important line.
• A provision quarter above $4.0B, or allowance coverage rising above ~1.95% from the current 1.82%, or the Card Services net charge-off rate exceeding the guided 3.4% — the credit cycle turning while the price embeds it staying benign.
• Net interest income excluding Markets missing the ~$95B guide, which would mean the deposit franchise is not carrying the guided growth and Markets NII is masking a flat core.
• A Basel III / GSIB outcome at or above the Firm's own ~6% CET1 estimate combined with a suspension or material slowdown of the $50B repurchase programme.
• A second one-off gain being treated as recurring — the Q1'26 10-Q notes Visa may extend further exchange offers on timing outside the Firm's control. If FY2027 consensus rises on another such item, the forward multiple is being computed on a fiction.
• Commercial & Investment Bank net income falling more than 15% year-on-year — at 48.8% of firm earnings, a capital-markets downturn is now the dominant single-segment risk.

Pre-registered FLIP TO BUY (initiate or upsize):

• A price below ~$330 (the 50-DMA), at which the derived multiple is ~2.95x and the required ROTCE falls to about 19.8% against a filed 20% — the first level where the arithmetic pays you rather than merely being fair. Below ~$300 (~2.68x, required ROTCE ~18.4%) the case becomes emphatic.
• A Basel III / GSIB final rule materially softer than the ~4-6% CET1 increase the Firm estimates, which would free capital, lift both ROTCE and buyback capacity, and justify 3.4x.
• Two consecutive quarters of ROTCE above 22% with the provision stable and NO one-off items — evidence that the current return is structural rather than cyclical.
• NII excluding Markets beating the ~$95B guide, showing the deposit franchise re-pricing.
• A clean Q3'26 print near or above $5.83 with no gain, which would validate the underlying run-rate and remove the principal information risk in this dive.

Where JPM fits in the Synthos Framework Portfolio. The low-beta financials / quality-compounder sleeve, as a 0-1% position at this price — a hold for existing holders and not an initiation. Beta 0.982 means it offers no defensive discount to the index either. On overlap within the batch: JPM and MKL (also covered 2026-08-04) are both financials and are not complementary at these prices — they are a direct choice. MKL trades at 1.11x book on a 12.4% return on equity against a residual-income-justified 1.64x, with zero street buy ratings, at the bottom of its range, with a six-claim conviction lane and zero insider selling. JPM trades at 3.19x tangible book on a 20% ROTCE against a residual-income-justified 3.00-3.20x, at a 52-week high, with a 32-of-61 buy consensus and a knowledge-base lane that has been silent for 113 days. The businesses are not comparable; the setups are, and one of them offers a discount while the other offers a fair price. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $357.52.

Single biggest risk: that a 20-21% return on tangible common equity is a cycle peak rather than a through-cycle number. Every element of the valuation rests on it — the 3.00-3.20x justified multiple, the 8-10% tangible-book compounding, the buyback capacity. It is being earned in a period of exceptional capital-markets revenue (CIB net income +30% in Q1'26, advisory fees +82%), benign credit (provision flat at ~$2.5B a quarter), and pre-Basel-reform capital requirements. Take away any one of those three and 17% is a perfectly ordinary outcome for this bank — at which the stock is worth $276. Nothing in the filings says it is a peak. Nothing says it is not. That symmetry, at a price that assumes one answer, is why this is a Hold.

Most fragile assumption in the price: that the market has correctly distinguished a $7.59 quarter from a $5.55 quarter plus a Visa share exchange. The street's forward estimates suggest it has; the three-month price action suggests it may not have. The Q3'26 print on 2026-10-13 against a $5.83 estimate is the test, and it is 70 days away.

Provenance & disclosures

• Traceability: 63 tagged knowledge-base claims name JPMorgan across 21 distinct sources, spanning 2019-08-29 to 2026-04-13, searched on the aliases JPM, JPMorgan, JP Morgan and JPMorgan Chase (entity-exact, case-insensitive) across all 51,928 records. Zero quarantine_misattributed claims — none excluded on that basis. Of the 63: 8 discarded as false positives where JPMorgan appears only as the publisher of a research call (gold target, Bitcoin fair value, Argentina GDP, a macro note, zero-DTE research, silver folklore, and two MicroStrategy index-inclusion claims); 34 excluded from conviction as banking-system, monetary or crypto-rail arguments naming the firm as an actor rather than as an investment — including 16 of arthur_hayes's 18 claims, five of which are the same "$1.5T lending programme" idea restated between 2025-10-15 and 2026-04-10; 21 counted as claims about the equity, running 16 bullish / 4 bearish / 1 neutral across 11 independent voices. No management or _mgmt voices appear in the lane, so no half-weighting was required; anthony_pompliano (1 claim, conviction 38) is treated as promoter-class and low-weighted. Two stance-tagging errors are flagged in Section 7: the 85-conviction arthur_hayes claim of 2025-10-11 is tagged bullish but forecasts a "coming valuation collapse" for incumbent banks, and the 2026-03-05 claim is tagged bearish while its content ("big banks are fine") is bullish for this equity. The lane has been silent for 113 days. All quotes are verbatim from the stored claim text.
• Data as-of: fundamentals — 2026-03-31, the balance-sheet date of the most recent full quarterly filing (10-Q filed 2026-05-01); annual statements 2025-12-31 (10-K filed 2026-02-13). Q2'26 figures come from the vendor payload only and are release-grade, not review-grade · estimates 2026-08-04 · prices 2026-08-04 ($357.52; 50-DMA $329.79; 200-DMA $312.37; RSI 60.2; MACD +6.81) · knowledge-base claims 2026-08-04. All figures come from the Synthos vendor data file for JPM or from the four SEC filings on disk; no figure comes from memory, recall or external retrieval.
• The Q2'26 filing gap, stated plainly: there is no Q2'26 10-Q in this set. The 8-K of 2026-07-14 is an Item 7.01 Regulation FD filing furnishing investor-presentation slides — its extracted text contains no financial figures, and Exhibit 99 is not on disk. The 8-K of 2026-07-23 is an Item 5.03 by-laws amendment (Lead Independent Director appointment), not a dividend or capital action; a second 8-K accession of the same date was overwritten at the same file path and is unavailable. No Q2'26 number in this dive is presented as audited or reviewed filing detail.
• Data rejections (detailed in Section 8): enterprise value rejected on concept, not corrected — deposits are a bank's raw material, and the vendor's $1.886 trillion EV is both conceptually void and internally inconsistent with its own $599.040B netDebt; all EV multiples rejected (evToSales 6.337x, evToEBITDA 20.762x, enterpriseValueMultiple 20.762x, netDebtToEBITDA 10.216x, evToOperatingCashFlow / evToFreeCashFlow 13.381x). bookValuePerShareTTM ($134.327), tangibleBookValuePerShareTTM ($111.268), shareholdersEquityPerShareTTM and priceToBookRatioTTM (2.662x) rejected for failing to deduct $20.045B of preferred stock and for dividing by a weighted-average rather than period-end share count — the 10-K's $107.56 and the 10-Q's $108.87 used instead. currentRatio and quickRatio (both 0.8404), workingCapital (−$601.210B), investedCapital (−$499.205B), netCurrentAssetValue (−$1,475.497B), grahamNetNet (−$577.73), daysOfSalesOutstanding / cashConversionCycle (665.2 days) all rejected as classified-balance-sheet artifacts. All free-cash-flow derivatives rejected (freeCashFlowYield 14.71%, priceToFreeCashFlow 6.797x, priceToOperatingCashFlow 7.074x, freeCashFlowToEquity $315.614B, freeCashFlowToFirm $217.123B, incomeQuality 2.404) — operating cash flow swung from −$42.012B to +$100.867B while net income moved $1.4B. interestCoverageRatio (0.854), debtServiceCoverageRatio (0.082) and solvencyRatio (0.0155) rejected as meaningless where interest expense is cost of goods. priceToSalesRatio (3.219x) and revenuePerShare ($106.728) rejected as built on the gross revenue line ($279.745B) rather than managed net revenue ($182.4B). est EBITDA/EBIT rows FY2021-FY2023 rejected as inverted (FY2023 EBIT $117.651B > EBITDA $64.698B), and EBITDA rejected conceptually throughout. The Q2'26 quarterly record flagged for three internal defects: depreciationAndAmortization $0, netIncomeDeductions $0 with bottomLineNetIncome equal to netIncome despite quarterly preferred dividends, and a diluted share count frozen at exactly 2,793,700,000 for five consecutive quarters. cf_a.preferredDividendsPaid $0 in every year rejected as wrong. Vendor FY2025 EPS ($20.05) superseded by the 10-K's $20.02; inc_a net income ($57.048B) and cf_a net income ($56.780B) disagree by $268M, with the filing's $57.0B used. rating sub-scores priceToBookScore (1) and debtToEquityScore (1) rejected as industrial thresholds mis-applied to a bank. One null insider record discarded.
• Filing-versus-vendor conflicts, named: the vendor's tangible book value per share of $111.268 against the 10-Q's $108.87 at 2026-03-31 and the 10-K's $107.56 at 2025-12-31 — we use the filings. The vendor's FY2025 diluted EPS of $20.05 against the 10-K's $20.02 — we use $20.02. The vendor's FY2025 net income of $57.048B (inc_a) and $56.780B (cf_a) against the 10-K's $57.0B — we use the filing. Where the vendor and the filing AGREE, and this is worth recording: seg_prod FY2025 sums to $182.447B against the 10-K's $182.4B of managed net revenue; the derived Q1'26 net interest income of $25.366B matches the 10-Q's "$25.4 billion" and derived noninterest revenue of $24.470B matches "$24.5 billion"; the derived Q1'26 provision of $2.507B matches "$2.5 billion". That agreement is what licenses the Q2'26 derivation in Section 2.
• Basis note — the most important one in this dive: the Q2'26 EPS of $7.59 against a $5.59 estimate is, on the evidence assembled in Section 2, almost entirely a one-off gain on the exchange of 18.6 million Visa Class B-2 shares — disclosed in advance in the Q1'26 10-Q under Recent events and Note 2, where the shares are stated to be held "at their nominal carryover basis" and the replacement Visa Class C shares to be "recognized at fair value, which is expected to result in a gain that may be recorded as early as the second quarter of 2026." A structurally identical $7.9B Visa gain was recorded in Q2'24 and is cited in the 2025 Form 10-K. Quarter-on-quarter, net interest income moved +$145M, the derived provision moved +$8M, the effective tax rate ROSE from 19.5% to 23.1%, and derived noninterest revenue rose +$7.366B (+30.1%). Tax-effected, that is roughly $2.05 a share, leaving a clean quarter of about $5.50-5.60 against the $5.59 estimate. This attribution is strongly supported by the filings and the arithmetic but is NOT confirmed by any filing in this set, because the Q2'26 10-Q is not available. It is labelled as a derivation everywhere it is used, and the FY2026 consensus of $24.64 is treated as containing ~$2.05 of gain, making the honest forward multiple ~15.8x rather than 14.5x.
• Cash-flow note: for a bank, operating cash flow reflects trading-book and financing movements, not distributable earnings — it was −$42.012B in FY2024 and +$100.867B in FY2025 while net income moved $1.4B. The actual distributions are the FY2025 buyback of $34.591B (3.61% of market capitalisation) and common dividends of $16.625B (1.74%), a combined shareholder yield of roughly 5.3%, under a $50B repurchase authorisation effective 2025-07-01 and a $1.50 quarterly dividend declared 2026-03-17. Note that repurchases executed at 3.2x tangible book dilute tangible book value per share.
• Estimate coverage: adequate and materially better than most of this batch — 9 analysts on FY2026 EPS, 12 on FY2027, 8 on FY2028; 8, 9 and 5 on the revenue lines. epsAvg verified against actuals (FY2025E $20.198 versus the filed $20.02). The est revenue line is on a managed-NET-revenue basis and is not comparable to inc_a.revenue, which is gross. The EBITDA/EBIT estimate rows are rejected (Section 8) and no conclusion rests on them.
• Peer note: the vendor peer set is genuinely relevant — Bank of America ($446.4B), HSBC ($364.4B), Royal Bank of Canada ($289.8B), Wells Fargo ($267.3B), Citigroup ($234.7B), Toronto-Dominion ($201.7B), Bank of Montreal ($125.9B), Canadian Imperial ($109.3B), Bank of Nova Scotia ($106.2B) and Nu Holdings ($69.4B). JPMorgan at $958.0B is more than twice the next-largest US peer. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn — a real limitation, since whether 3.2x tangible book is rich depends on where BAC, WFC and C trade.
• Fair-value caveat: the $276 / $372 / $422 anchors are multiples of tangible book value per share (2.40x / 3.05x / 3.40x), each stated with the return on tangible common equity it implies under a residual-income frame at a 10% cost of equity and 5% long-run growth, and each cross-checked against FY2027 consensus EPS of $24.99 (11.5x / 14.9x / 16.9x). Stated arithmetic, not a discounted cash flow. The base assumes modest multiple COMPRESSION (3.19x → 3.05x) on the view that a 20-21% ROTCE is at or near a cycle peak; a flat 3.19x multiple on the same forward tangible book would give $389. The bull requires expansion and is the fragile leg. Tangible book value per share is anchored on the FILED $108.87 (2026-03-31), with a derived ~$112 for Q2'26 used only where a post-Q1 figure is required and labelled as derived.
• Earnings-call transcripts are not available on this plan. Management Q&A from the 2026-07-14 call is absent, and the 8-K furnishes the presentation slides rather than the call. No management quotation appears in this dive that is not drawn verbatim from the 10-K or 10-Q. All forward guidance cited (NII ~$103B, NII ex-Markets ~$95B, adjusted expense ~$105B, Card net charge-offs ~3.4%) is management framing and is half-weighted per house rule.
• Risk factors: the Q1'26 10-Q's Item 1A contains no new or amended risk factor — it incorporates pages 9-31 of the 2025 Form 10-K by reference. No changed risk factor to report.
• Timing: Q2'26 was reported 2026-07-14, 21 days before this dive. The next print is 2026-10-13, 70 days away, so the earnings-within-five-days rule is not triggered and no banner applies. Non-equity tripwire not triggered: JPM is the common stock (beta 0.982, a $279-$363 annual range, 8.4M average daily shares). Note that the Firm has many listed preferred depositary-share series (5.75% Series DD, 6.00% Series EE, 4.75% Series GG, 4.55% Series JJ, 4.625% Series LL, 4.20% Series MM) and several guaranteed note and ETN listings, all of which appear on the 8-K cover pages; none of those is this security and none should be valued as common equity.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.