US Banking After the Print

Part 1 asked whether the first hike of the cycle would turn the banking system's paper loss on its bond books into a real one. Twelve prints later, on the banks' own documents, the forced-sale branch did not open beyond one sale of size and the capital branch ran as computed: the books were run off and hedged rather than sold, the mark took 10 to 38 basis points of CET1 where its reach is printed and the two ranges Part 1 said it would breach held, deposits stayed, and the managements that gave a deposit beta — the share of a rate change they pass on to depositors — gave last cycle's. Their story moved the other way, to two or three more hikes and more net interest income, and the one thing that changed in behaviour was the pace of capital return, slower at three of the twelve for reasons each gave itself. What the calls worried about was not the bond book but the assets the banks finance: six managements named a risk in the AI build-out and none sized it, and the loan category with the largest dollar increase in the system went unmentioned at the three banks Part 1's test tracks. The market sold nine of the twelve on the day; tangible book per share fell at six of nine, the route by which the mark reaches a share price whether or not it reaches a ratio. Of Part 1's thirteen tests two fired, three held, Part 1 missed twice on its own rules, and the rest wait for the 10-Qs and the FDIC.

TON618 Capital Research · As of 18 October 2026 · Download PDF

Part 1 of this note, published on 10 October, set out what the twelve were going into the quarter and froze thirteen tripwires, six watch items and a reversal block, so that the week could be scored rather than narrated. This is what the week taught, with the score as its receipt. It is written on the banks' own documents only: the releases, supplements and presentations filed to EDGAR on 13, 14 and 15 October, the cautionary section of each release read in full, Schwab's September activity figures, and the twelve calls archived with their provenance — the company's transcript where one is posted, independent transcriptions where it is not. The Q3 10-Qs, the Call Reports and the FDIC's quarterly profile post-date this note; every gauge that needs them is scored "pending" exactly as Part 1 froze it, and a November addendum closes them. Nothing from Part 1 is re-cut.

Terms used throughout, defined once. AOCI: accumulated other comprehensive income, the equity line where unrealized gains and losses on available-for-sale (AFS) bonds sit after tax, alongside cash-flow hedges, pension items and currency translation; held-to-maturity (HTM) bonds are carried at cost and their loss is a footnote. The mark: the unrealized gain or loss on a bond book. CET1: common equity tier 1, the core regulatory capital ratio. Tier 1 leverage: Tier 1 capital over average assets, no risk weights; Schwab's "adjusted" version puts AOCI back in. ROTCE / P/TBV: return on tangible common equity, and price to tangible book value, the two axes of the line in §5. Deposit beta: the share of a change in the policy rate that a bank passes through to the rate it pays on deposits, quoted cumulatively over a hiking or cutting cycle and usually on interest-bearing deposits; at a 50% beta a 25-basis-point hike raises the deposit rate about 12.5 basis points once repricing is complete. "Last cycle's" betas are the pass-through the managements saw over the last hiking and cutting cycle, which they now cite as the path for this one: "about a 50% or so" at U.S. Bancorp, "in the mid-50s" at PNC. NDFI: nondepository financial institutions — private-credit funds, mortgage companies, broker-dealers, insurers. Fired / held / disclosed / none / pending: a tripwire's condition was met; was tested and not met; a management stated the thing asked about; the documents were silent, scored as the silence branch Part 1 froze; cannot be tested until a later document.

One. The bond book was run off and hedged, not sold, and the mark reached capital by about the amount Part 1 computed. Part 1's question was whether the hike would turn the system's paper loss into a real one. The managements answered it with what they are doing with the books. Bank of America "funded that growth by reducing securities balances" — $34 billion of loan growth — and runs its held-to-maturity book off at about $9 billion a quarter; U.S. Bancorp's portfolio "has been coming down, and we have been doing more in cash"; BNY has "roughly about $5 billion of securities rolling off per quarter" reinvested 150–200 basis points higher; PNC lengthened its duration to 4.1 years from 3.6 with $5 billion of 2.65% paper rolling off in the fourth quarter; Schwab still has "$190 billion roughly of securities that are yielding sub 2%" and is letting them roll into lending; State Street's available-for-sale duration is "around one year". The self-heal Part 1 sized at $21 billion a quarter for the system is the earnings story each of them told: higher reinvestment yields on a shrinking book. Five banks sold at a loss and one in size — JPMorgan's $658 million, which its release calls "discretionary" with no reason given — and no one disclosed a transfer into or a sale out of held-to-maturity, so the branch of Part 1's chain that runs from crystallization to liquidity did not open. Where the mark is printed and the book is unhedged, the quarter moved it 0.90 to 1.15 times Part 1's projection (Bank of America 1.15, JPMorgan's securities line 1.09, State Street 0.99, PNC 0.90); BNY's hedged book moved 0.55 times the unhedged projection, and Schwab's whole AOCI moved $164 million against $1.4 billion projected, the one number in the week Part 1 got wrong in size. Where the mark's reach into capital is printed it matched: about 36 basis points of CET1 at JPMorgan against 29 computed, 38 at Wells Fargo against 36, about 10 at Bank of America against 10. Three banks print only a total AOCI line — Wells Fargo's fell $8.2 billion, 1.6 times the after-tax projection, U.S. Bancorp's and Citigroup's 1.4 times — and two print none, and the two that print both lines show why the totals are not scored: at Bank of America "a little less than half" of the fall was cash-flow hedges that never reach capital. The bank-by-bank scoring, the totals and the loss sales are in §4.

Two. Deposits stayed, and the cost of money rose by less than the hike and more than the average funds rate. The average rate paid on interest-bearing deposits rose at every bank that prints it — JPMorgan 2.18% from 2.11%, Bank of America 2.03% from 1.98%, Citigroup 2.76% from 2.71%, Wells Fargo 2.05% from 1.99%, BNY 2.43% from 2.33% — and period-end deposits rose at eight of the eleven with bond books, 2.5% at JPMorgan, 7.2% at Morgan Stanley, 3.7% at PNC; they fell 2.5% at State Street and 2.2% at Schwab's bank, and BNY's quarterly average fell 0.9%. None reached the −3% that, with a loss sale, fires the thirteenth tripwire. The hike came on 16 September, fourteen days before the quarter closed, and added 25 basis points to the target; on the quarter's average the effective funds rate rose 3.6 basis points, from 3.631% to 3.667%. Against the hike, deposit costs rose a fifth to two-sevenths as much, two-fifths at BNY; against the average they rose more, which the managements that explained it put down to mix rather than to the policy rate — PNC to brokered certificates and commercial balances, U.S. Bancorp to about 2 basis points from the hike itself, with no reason given for the rest. Part 1 wrote "rising more than the funds rate did" and did not say which funds rate. The house reads it as the hike, because the observable was written about the hike's pass-through and a 25-basis-point step cannot show in a 3.6-basis-point average; on that reading it did not fire, on the other it fired, and both are scored in §6 so that the fourth quarter, the first full quarter at 3.75–4.00%, settles it with a reading rule that is no longer open. The cheapest money held its share at JPMorgan (24.0% of average deposits from 23.7%) and Bank of America (26.5% from 26.4%), slipped at Wells Fargo (24.3% of period-end deposits from 24.7%) and fell at State Street (13.4% from 14.5%); the uninsured shares, the second half of the funding test, are 10-Q and Call Report figures and are pending.

The betas, as stated. None of the money centers gave a figure. JPMorgan's chief financial officer said consumer deposit betas have run "lower-than-modeled" in what he called "a different cycle", so the firm is "a little bit more asset-sensitive than the EaR shows", and expected the low-beta consumer money to "behave roughly the way it has been behaving"; Citigroup's said betas are "performing in line with our expectations"; Wells Fargo expects this cycle's to match the last cycle's component for component — "not … that different versus what we saw last go around" — and "slightly higher" in aggregate because of the commercial mix; Bank of America's chief financial officer, asked to unpack the betas behind a 2027 outlook that carries three more hikes, said "we came from 4.5% down to 3.75% – 3.5%. Now we're retracing steps, I would anticipate it will look the same going back up as it did coming down". The regionals gave numbers: PNC expects a cumulative beta "in the mid-50s", its 7-basis-point rise to 1.98% "primarily driven by mix"; U.S. Bancorp assumes "about a 50% or so" on the hikes ahead and puts "approximately 2 basis points of impact from the September Fed rate hike" in its 7. BNY cited "deposit margin compression" and gave no figure; State Street said it would be "very attentive to deposit betas and repricing"; Schwab's sensitivity table assumes a zero deposit beta and its chief financial officer said the September cash move "is not cash sorting in any way"; Goldman Sachs was not asked and did not say; Morgan Stanley gave no figure. Nothing in the week contradicts Part 1's reading that the betas the July calls leaned on are intact; nothing in it tests them either, because the hike had two weeks.

Schwab's cash. Transactional sweep cash ended September at $492.8 billion, $7.1 billion above June's $485.7 billion and $19.8 billion above the $473 billion that fires the tripwire. The release attributes the rise to "continued demand for long/short strategies and organic growth, partially offset by client asset allocation decisions", and footnote 7 defines the measure to include "short credits related to certain client long/short strategies", which Schwab's deck puts at $63.3 billion after a $19.6 billion build in the quarter. On the balance sheet, bank deposits fell to $244.1 billion from $249.7 billion (−2.2%) and Federal Home Loan Bank borrowings rose to $2.0 billion from $0.5 billion; the chief financial officer told the call that "even with those higher rates, this is not cash sorting in any way". The tripwire, as frozen, is held and the deposit question stays closed this quarter; the steelman's reading, that the measure rose on short credits while the bank's deposits fell, is correct on the balance sheet. There was no stand-alone September activity release; the figures are the release's own monthly table, and the October report is the next read.

Three. The earnings story has moved onto more hikes, and the guides that moved, moved up. Part 1 recorded that no July outlook assumed more than one hike, late in the year. The October outlooks assume two or three — JPMorgan and Bank of America three, PNC two, U.S. Bancorp "a couple", Citigroup and Wells Fargo no stated count — and the guides that moved, moved up. JPMorgan lifted its full-year guide for net interest income excluding Markets to about $97.2 billion from July's roughly $96.5 billion and put a preliminary 2027 central case at about $107 billion, with one hike in December and two in the first half of 2027 as the forward curve stood; Citigroup raised the same line to the "mid-to-high 6% range" of growth from the roughly 5–6% it set at the start of the year and calls higher-for-longer "net positive" on its non-dollar asset sensitivity; State Street raised its net interest income growth guide to 15–16% from July's 14–15% and calls the rate path "a tailwind"; U.S. Bancorp raised full-year revenue growth to 8–9% from 7–9% on a like-for-like basis, assuming "a couple of hikes". Bank of America, Wells Fargo and PNC held — PNC at 15.5%, the top of July's range, agreeing with a 2027 figure about $1.0–1.1 billion above 2026; Bank of America expects 2027 "to be 8% or more higher than 2026" on three quarter-point hikes, called rates "a small component" of that, and said its Markets net interest income is liability-sensitive and may fall, as Wells Fargo's is. The exception ran the other way at Schwab, whose revenue and margin guide moved to the lower end of July's ranges and whose "doesn't impact the financials for 2026" was not repeated. PNC printed its baseline in the release's own cautionary statement — "the Federal Reserve to raise rates twice more, at the December 2026 and March 2027 FOMC meetings", to 4.25–4.50% — and now calls short rates "pretty neutral" for 2026. The direction is the one Part 1 flagged as the first sector tell: the hike that July's calls treated as one and late is now two or three and near, and the guides that moved, moved up.

Four. Capital return slowed for three different stated reasons, and the firm that kept buying most says the final rule sets its pace from here. The quarter's rate move cost CET1 30 basis points at JPMorgan (to 13.9%: $7.9 billion of securities losses, $51 billion more risk-weighted assets and distributions), 37 at Bank of America (to 10.9%, 90 basis points over its minimum, with a step-up to 10.50% due on 1 January), 30 at Wells Fargo (to 10.0%, the bottom of its 10–10.5% range, 38 of it the AOCI decrease) and 50 at Goldman Sachs (to 12.4%, on $26 billion more risk-weighted assets, 100 basis points over its requirement); it left Citigroup and Morgan Stanley flat, and lifted State Street 20 on lower risk-weighted assets. The responses were not of a kind. JPMorgan cut its buyback to $664 million net from a cadence an analyst put at $7 billion, its chairman giving the reason as redeploying "probably $30 billion or something of excess capital" at better returns than buying the stock at "a 6% or whatever the earnings yield would be", and saying the cut had "nothing to do with AOCI or guessing about the environment or even Basel III"; U.S. Bancorp paused repurchases after $12 billion of risk-weighted-asset growth and a 40-basis-point AOCI hit to its AOCI-inclusive ratio, and expects to resume "over the next quarter or two"; Schwab halved its buyback to $0.5 billion and "allowed capital to build"; Wells Fargo will size repurchases quarter by quarter and "continue to consider the interest rate environment and the impact on accumulated other comprehensive income when determining how much stock we repurchase"; State Street issued $496 million of preferred, held its buyback at $400 million and guided it up to about $700 million; Bank of America returned $8.3 billion and called the pace "business as usual … with one exception. We need to wait for the final rules"; Goldman returned $4.49 billion and said it is hopeful the endgame and surcharge proposals are finalized by year-end; Morgan Stanley bought back $1.5 billion on a 300-basis-point buffer and looks to "upcoming transparency as our regulators finalize the bank capital regulatory framework"; PNC held at $0.6 billion with the fourth quarter guided the same; BlackRock at $550 million. Three managements dated the final rules to the fourth quarter or year-end (Goldman Sachs, Bank of America, U.S. Bancorp), three more referred to their finalization without a date (Wells Fargo, Morgan Stanley, PNC), and at JPMorgan and Citigroup the date came from the analysts' questions; Wells Fargo put the relief at about 7% of risk-weighted assets, U.S. Bancorp at about 5%, and U.S. Bancorp said it is "fully ready to transition into" Category II regardless and is "not counting on any benefit". The finding has two halves that the documents keep apart. The three that cut or paused gave reasons of their own: JPMorgan redeployment, U.S. Bancorp risk-weighted-asset growth and the mark, with regulation named among the drivers of its restart, Schwab a capital build against volatile rates. Of the firms that kept buying, Bank of America said the final rules set its pace from here, Wells Fargo named rates and AOCI, Goldman gave a hope of year-end finalization and no pace, and Morgan Stanley no pace at all. The two ranges Part 1 watched are the sharpest case; §4 bridges both. What separates the two readings is whether the buybacks resume when the rule lands with the marks unchanged, which the reversal block in §6 sets as the test.

Five. The worry has moved to the assets the banks finance: six managements named a risk in the AI build-out, none sized it, and the deposit side of it was dismissed, discounted or conceded as not-yet at every call that asked. The AI cycle reached the calls from two directions, and the managements answered them differently. As the thing the banks finance, it came up on nine of the twelve calls — at eight because an analyst asked, Goldman Sachs taking three questions and PNC two, and at Wells Fargo from both sides, an analyst asking and the chief executive raising it himself: "The scale of investment here increasingly resembles prior infrastructure booms, creating both significant opportunities but also risks given the significant leverage used to finance the buildout". Six managements named a risk in it. JPMorgan's chief financial officer, asked about concentration risk, said the theme "is so broadly diffused into the marketplace, it's not just traditional construction lending. Now, increasingly, it's working capital revolvers", that the bank runs stresses against "a meaningful change in the overall theme across the entire enterprise", and that "our exposures are still quite modest in the context of our overall wholesale exposure, but that's not the same as saying that we would be immune to a major change in sentiment in the AI space. I think I would be surprised if anyone frankly was immune to that at this point." Goldman's chief executive, asked whether financing is the choke point: "I don't know and nobody knows", then "there is risk. There is risk to pressure on, you know, crowding out the size, the scale. Not every project's going to be received as well", and higher rates "will have an impact on the pace of build-out and what projects get done". Morgan Stanley's: "I have no doubt there will be times when it'll feel like it's going too fast, and there is a sense of overbuilding in some respects", and on lending into it, "I imagine there are going to be cases where we are going to make mistakes." Wells Fargo's chief financial officer lists "AI related disruption" among the credit risks it monitors and keeps "very hard limits in terms of what we're willing to do" on the financing. Bank of America's and PNC's chief executives drew the same ripple: if the growth that is "due to AI activity" were "to slow down", it "could have ripple effects through construction, staffing" (Bank of America); and at PNC, "not large in the data center finance space", asked how much of its loan growth is AI-related — "Maybe the negative of it, Mike, is the more worrisome thing, right? So if the CapEx broadly slows down related to AI, do we think even though we don't lend directly to that, that it would slow our loan growth? Yes" — a knock-on effect, its chief financial officer agreed. Citigroup dissented — "I don't think that higher rates is going to trump the imperative to invest there … we're not anticipating a lot of shift in behavior", citing an investment-grade-heavy client base and saying it keeps "an eye on sentiment" — while taking lead roles on the SK hynix, Intel and SoftBank financings; U.S. Bancorp is looking for clues to how much of its loan growth and credit is AI-related and finds the growth broader than the adjacent sectors; BlackRock is raising the money and says the market still lacks the data and ratings to trade the asset. In writing, the cycle appears as a macro risk in one release, PNC's: "a reversal in AI-related sentiment, which would have knock-on effects on both capex and wealth-driven consumer spending". The AI language in the four other cautionary sections that carry it — Wells Fargo, Bank of America, BlackRock, State Street — concerns the firm's own adoption, management and cyber exposure, not the cycle, and Wells Fargo's release counts "the investment in AI infrastructure" among the "strong positive forces" in the economy. The second direction is the deposit base. Whether AI agents will move money out of banks was asked at six calls — JPMorgan, Citigroup, Bank of America, Wells Fargo, U.S. Bancorp and PNC — and dismissed at four, discounted with a caution at JPMorgan, and conceded as a not-yet at PNC: "completely overdone, especially for the time period we can see" (Wells Fargo), "I am not worried about deposit flight risk in Services" (Citigroup), "we're not worried about it … We believe strongly the human in the loop" (Bank of America), "we haven't seen any AI cash sorting" (U.S. Bancorp), and at JPMorgan "a little bit less persuaded that you really have cliff effects there", with the caution that a bank is "probably not supposed to be hoping that your deposits are stable because the money is difficult to move"; PNC's chief executive: "there's no question that deposit movement is going to become more efficient … you're going to see more competition in deposits", though "not showing up in the moment", and the agentic applications that try to "transact on your behalf inside of your accounts" are "incredibly unsafe". Two firms are keeping outside agents out for now: JPMorgan is "not going to allow just any outside agents to come in inside the company", and Schwab is "blocking unauthorized agents who try to log in on behalf of clients". What the two directions share is where the managements put the risk: in the assets they finance, not yet in the deposits they hold. The theme JPMorgan describes as reaching "working capital revolvers" and "kind of everywhere" is an exposure none of the twelve sized on the day; the 10-Qs are the next document that could.

Six. Credit is benign on the measures that lag and ticked on the ones that lead, and the nonbank book went unspoken where it is largest. Citigroup's branded-card and retail-services net credit losses were $1.7 billion in the quarter with a $208 million reserve build it attributed to the macroeconomic outlook; it cut its full-year U.S. card loss expectation to about 4%, the low end of its prior range, and says 86% of balances carry FICO scores of 660 or better. JPMorgan held its full-year card charge-off expectation at about 3.2%. Wells Fargo described card charge-offs as down on the quarter and the year within a 70-basis-point consumer loss rate, with early-stage delinquencies at historic lows. Capital One, the card peer Part 1 used as the reference, reports after this note and is in the pending ledger.

The fifth tripwire tracks JPMorgan, Wells Fargo and Morgan Stanley and fires on any markdown or reserve build against lending to nondepository financial institutions; a first-time breakout of balances scores "disclosed"; silence scores as silence, on the banks' own documents only. At JPMorgan and Wells Fargo the releases, supplements and presentations contain no occurrence of "nondepository", "NDFI", "nonbank financial" or "private credit"; the calls — twenty-three questions from twelve analysts at JPMorgan, nineteen from ten at Wells Fargo — contain no prepared remark and no question on it. At Morgan Stanley the release and supplement carry no such line — the supplement's one "non-bank" footnote is $2 billion of securities-based lending to non-bank entities in Wealth Management, collateral lending rather than credit to nondepository lenders — and the call's three independent transcriptions contain no mention of lending to the category. All three are scored silent, the branch Part 1 froze. Outside the set, Bank of America's release is the only one of the twelve whose cautionary statement names the category — "a deterioration in private credit markets", the standing line Part 1 recorded as forward-looking language — while the three tracked names' cautionary sections cross-refer to their 10-Ks or list rate, deposit and securities risks without it; State Street told a loan-growth question it sees "healthy opportunities in supporting our clients in the subscription finance space and in AAA CLOs and … also even in fund finance", Bank of America's chief executive mentioned only "lending to warehouse mortgage lenders that go up and come down", and the one bank that treated the category as a credit matter was U.S. Bancorp: its deck carries a page titled "NDFI Portfolio — Well Diversified, Strong Credit Quality", $50 billion of loans to nondepository financial institutions against $109 billion of core commercial and industrial loans, with an internal rating equivalent of A on average and a criticized share under 1%; and its chief financial officer said commercial balances were flat because of "some runoff in some balances related to more in that NDFI space" where "new deals coming to market didn't hurdle" its return threshold — a lender turning the dial down, not a markdown; Goldman Sachs mentioned private credit once, as a fundraising tailwind for its own alternatives platform "despite the noise in the media", and attributed its $368 million provision to loan growth and "a couple" of idiosyncratic corporate impairments. The steelman is right that the visibility gap is intact; the tripwire is also right that nothing in the banks' own documents opened it, and the two statements are the same fact.

The industry reading is the mirror of Part 1's: the loan category that grew fastest in the system is the one its largest lenders did not describe, and where a lender did describe it, it reported runoff where new deals did not meet its return threshold. The lagging measures are benign: charge-offs ran 28 basis points at PNC (including $35 million on a student-loan transfer), 44 at Bank of America, 51 at U.S. Bancorp, and every card lender's losses fell or held. The leading ones ticked: PNC's nonperforming loans rose $167 million, 8%, on the quarter, "driven by higher commercial nonperforming loans"; Citigroup's firm-wide reserve build was $369 million, "driven by changes in the macroeconomic outlook and portfolio growth", of which the card piece was $208 million; Goldman's provision was $368 million against $102 million in the second quarter and "primarily reflected impairments related to wholesale loans"; Bank of America's provision fell. None of it is a credit event; all of it is the direction the steelman's case needs, and the fourth-quarter prints read it next.

JPMorgan Chase would rather redeploy its excess capital than buy its stock at this price, and took the week's one securities loss of size. The buyback was cut to $664 million net while the chairman named about $30 billion of excess capital and a redeployment plan ahead of repurchases at a 6% earnings yield, and the $658 million of securities losses are described as "discretionary", with no reason given. Earnings were $5.57 a share, $5.75 before the losses, against the $5.91 Part 1 archived — a miss on both bases — with a 20% return on tangible equity; CET1 13.9%, down 30 basis points on $7.9 billion of securities losses in the capital roll, $51 billion more standardized risk-weighted assets and distributions. The 2027 central case of about $107 billion of net interest income excluding Markets assumes three hikes; consumer deposit betas are running "lower-than-modeled"; card losses are held at about 3.2%; nothing on lending to nonbanks in twenty-three questions. Tripwires three and eight fired here and nowhere else in size. The stock fell 0.9% on the day.

Goldman Sachs is growing its balance sheet into its capital buffer while it buys back stock, and its growth story is the alternatives platform. Standardized CET1 fell 50 basis points to 12.4% on $26 billion more risk-weighted assets, leaving 100 basis points over its requirement, the top of its 50–100 target buffer, with $3.00 billion of buybacks and a dividend raised to $5 a share; the platform's fundraising target doubled to $150 billion and private credit was mentioned once, as a fundraising tailwind "despite the noise in the media". $17.01 against $13.79, an 18.7% return on equity. Goldman does not guide net interest income and attached no rate path; it said markets price about 75 basis points of further tightening. The stock fell 0.2% on the day; twenty-four analyst questions, none on the securities book.

Citigroup returned capital through the mark, and the market punished it hardest for a beat. The chief financial officer twice described the quarter's loss in other comprehensive income as something to "absorb" while returning $4.1 billion ($3 billion of buybacks, $13.3 billion year to date against a $30 billion programme, the dividend up 12%), with CET1 flat at 12.8% and total AOCI down $2.3 billion. $2.85 against $2.63, a return on tangible equity of 11.5% from 13.0%, and the quarter's clearest raised guide: net interest income excluding Markets now growing in the "mid-to-high 6% range" against the roughly 5–6% set at the start of the year, with higher-for-longer "net positive" on non-dollar asset sensitivity. Card losses were cut to about 4% for the year, the low end of the range, with a $208 million macro reserve build; four analysts raised AOCI or the capital rules, none the nonbank book. The stock fell 5.6% on the day and 7.3% over two sessions, the week's largest reaction to a beat.

Wells Fargo is at the bottom of its own capital range, and the mark is now weighed in sizing its buyback. CET1 fell to 10.0%, the bottom of its stated 10–10.5% range, 38 basis points of it the AOCI decrease by the company's own attribution as total AOCI fell $8.2 billion, and repurchases — $2.5 billion in the quarter — will now be sized with "the impact on accumulated other comprehensive income" in view. $1.87 against $1.85, a 16.3% return on tangible equity, net interest income held at $50 billion plus or minus. Betas are expected to match last cycle's component for component, "slightly higher" in aggregate on commercial mix; the dividend rose to $0.50 from $0.45; deposits were flat and the noninterest-bearing share slipped to 24.3% from 24.7%; tangible book per share fell 3.6%, the most of the twelve; $29 million of debt-securities losses; nothing on nonbank lending in nineteen questions; the Basel endgame is "tailwinds", about 7% of risk-weighted assets. The stock fell 3.1% on the day.

Bank of America split what the rate move did to its capital into the part that reached CET1 and the part that did not, and said the final rules set its pace from here. Of the $3.24 billion fall in AOCI, "a $1.8 billion portion of AOCI of AFS securities" reached CET1 and "a little less than half" was cash-flow hedges that did not; the available-for-sale mark moved 1.15 times Part 1's projection; CET1 fell 37 basis points to 10.9%, 90 over a minimum that steps up to 10.50% on 1 January; $6 billion was repurchased and the pace is "business as usual … with one exception. We need to wait for the final rules." $0.91 reported and $1.08 excluding a $1.2 billion regulatory settlement charge, against $1.12 — a miss on both bases and the week's second-largest fall, 4.8%. Net interest income rose 8% to $16.6 billion with the yield at 2.16%, the 2027 guide is 8% or more on three hikes, the total deposit rate rose 3 basis points to 1.49%, and betas will "look the same going back up as it did coming down". Card losses 3.22% and falling; the only nonbank reference was warehouse lines to mortgage lenders.

Morgan Stanley is choosing to be more of a bank. Wealth Management deposits rose by nearly $30 billion in the quarter to $465 billion as "a strategic choice" to grow the bank (firm-wide, $478 billion, up 7%), the segment's net interest income fell modestly as the mix shifted, and the funding benefit is promised for next year. $3.03 against $2.95 and a 22.6% return on tangible equity from 26.6%; CET1 held at 14.9% with a 300-basis-point buffer the chief executive described as the firm's flexibility, and $1.5 billion was repurchased. Nothing on the securities book, AOCI or lending to nonbanks in seventeen questions; the capital framework was raised only by management, looking to its finalization. The stock fell 2.7% on the day.

BlackRock gave the week's fullest account of private-markets money, and it expects long rates to stay high. $68 billion raised in private markets over twelve months, $25 billion of private credit in HPS's first year, $15 billion of private-markets inflows in the quarter, $31 billion into cash management, and a management that expects "slightly higher rates, especially in the longer end" with the 10-year above 5% and real rates near 3%. $15.93 as adjusted ($10.57 reported) against $14.15, and the week's one two-session gain of size, 5.1%. Buybacks $550 million, with at least $550 million planned for the fourth quarter.

State Street holds its leverage range by issuing, not by slowing the buyback. Tier 1 leverage rose to 5.5% from 5.3% "primarily reflecting the preferred share issuance completed during the quarter" ($496 million) with the buyback unchanged at $400 million and guided to about $700 million toward an 80% payout. $3.69 against $3.66, a 25.1% return on tangible equity ex-notables, net interest income up 21% on the year with the margin at 112 basis points, and the guide raised to 15–16% growth from 14–15%; the available-for-sale mark moved 0.99 times Part 1's projection on a book whose duration is "around one year", and CET1 rose to 11.0% on lower risk-weighted assets. Period-end deposits fell 2.5%, the noninterest-bearing share to 13.4% from 14.5%; it sees "healthy opportunities" in subscription finance, AAA CLOs and fund finance. The stock fell 4.1% on the day.

Charles Schwab's capital did not feel the hike; its revenue and margin guide moved to the lower end of its ranges for a funding reason, loan growth outpacing deposit growth. AOCI moved $164 million against the $1.4 billion Part 1 projected — the week's one number Part 1 got wrong in size — so the adjusted Tier 1 leverage ratio printed 7.0%, the top of the 6.75–7.00% objective, as the company "allowed capital to build" with the buyback halved to $0.5 billion; revenue and margin were guided to the lower end of the summer ranges, and $190 billion of securities still yield under 2%. $1.69 as adjusted ($1.62 reported) against $1.66, a 3.04% net interest margin. Transactional sweep cash ended September at $492.8 billion, up $7.1 billion on the quarter with $19.6 billion of the build in short credits on long/short strategies, while bank deposits fell 2.2% and Home Loan Bank borrowings rose to $2.0 billion from $0.5 billion; "this is not cash sorting in any way". Eleven questions, none on the marks or the capital proposal. The stock fell 2.7% on the day and last traded 2.3% lower on the 16th.

PNC printed its own rate baseline, and its cost of money rose, it says, on mix. Its cautionary statement assumes two more hikes, December and March, to 4.25–4.50%, with "a reversal in AI-related sentiment" and "any further sharp rise in bond yields" as its named risks; its cost of money rose 7 basis points to 1.98%, "primarily driven by mix", $7 billion of brokered certificates among them, with a cumulative beta expected "in the mid-50s". $4.97 reported and $5.09 as adjusted against $4.91, an 18.3% return on tangible equity, net interest income guided up 15.5% for the year, the top of July's range, and about $1.0–1.1 billion higher again in 2027 on a figure its chief financial officer "would agree with". The available-for-sale mark moved 0.90 times the projection, the securities leg outran the $83 billion swap book, duration lengthened to 4.1 years, tangible book fell 3% "entirely driven by the decline in AOCI", and CET1 printed 9.8% against a 10% target, 8.8% including AOCI — a measure its chief financial officer called "stale" because the coming rules include it. Buybacks $0.6 billion, the fourth quarter the same; $1.4 billion of student loans moved to held for sale; nonperforming loans up $167 million on commercial credits. The stock fell 0.4% on the day.

U.S. Bancorp gave numbers the others did not: the hike's share of its deposit cost, and its nonbank book. Deposit costs rose 7 basis points with "approximately 2 basis points of impact from the September Fed rate hike" and a beta of "about a 50% or so" assumed ahead; buybacks were paused as $12 billion of risk-weighted-asset growth (heavy loan demand and a co-brand card onboarding) and a 40-basis-point AOCI hit took the AOCI-inclusive CET1 ratio to 9.1% from 9.4%, to resume "over the next quarter or two"; and the nonbank book was both printed — a deck page of $50 billion of NDFI loans against $109 billion of core commercial and industrial, rated A on average with a criticized share under 1% — and allowed to run off in part, because "new deals coming to market didn't hurdle". $1.34 against $1.32, an 18.5% return on tangible equity, record revenue. It is "fully ready to transition into" Category II regardless of the rule's timing, expects about 5% relief on risk-weighted assets, and is shifting from securities into cash. The stock rose 2.0% on the day.

BNY has a book it calls immunized to rates and a deposit rate that rose the most of the twelve. "Most of our peers and ourselves took a bit of a mark-to-market in the AOCI book", and its answer is about $5 billion of securities rolling off each quarter reinvested 150–200 basis points higher, a book "largely immunized" to a hundred basis points either way, and the hedged available-for-sale mark moving 0.55 times Part 1's unhedged projection; the rate on interest-bearing deposits rose 10 basis points to 2.43%, under "deposit margin compression". $2.34 reported and $2.37 as adjusted against $2.27, a 29.0% return on tangible equity from 31.3%, net interest income up 15% on the year and guided to the same for 2026, and the week's largest gain on the day, 3.2%. Tier 1 leverage fell 24 basis points to 5.7%, the upper end of its 5.5–6% range; $937 million was repurchased at an 85% payout; average deposits fell 1% on seasonality.

The written sections. The cautionary section of each of the twelve releases was read in full. Four — JPMorgan, Morgan Stanley, Schwab and BNY — cross-refer to the 10-K and name no risk of their own. What the other eight name, in order of how many name it:

Against Part 1's ranking. Part 1 ranked the twelve managements' own risks from their July calls, 2026 letters and FY2025 10-Ks as geopolitics, the AI cycle both ways, asset prices, private credit, and deposit repricing, in that order. The October documents keep the first two where they were: Morgan Stanley's chief executive opened on "the headwinds of renewed tensions in the Middle East and higher interest rates globally", Goldman's cautionary note names "the continuation or worsening of the conflict in the Middle East", and the AI cycle is the economy's engine at PNC and BlackRock, a risk named by six managements on the calls (finding five) and a written macro risk at one. Asset prices moved from letters to a rate: BlackRock's "10-year Treasury yields more than 5%, real rates are close to 3%", PNC's "sharp rise in bond yields", Goldman's chief executive on a 5% 10-year that "doesn't scare" him. Private credit stayed where Part 1 put it — in Bank of America's standing cautionary language, in Goldman's and BlackRock's fundraising, in U.S. Bancorp's one dedicated page, and nowhere in the three money-center calls. Deposit repricing, Part 1's fifth, was named in two cautionary statements — Wells Fargo's, on "the effect of changes in the level of checking or savings account deposits on our funding costs", and Bank of America's, on "an inability to access capital markets or maintain deposits or borrowing costs" — and by no management on a call as a risk; it was answered, bank by bank, with last cycle's beta. What rose was not on Part 1's list because it is not a risk the managements name but one they are waiting out: the capital rule. Six of the twelve calls raised its finalization, three dating it to year-end, and the one exchange that touched Part 1's own question — the phase-in of AOCI into the Category III banks' capital — produced a "stale measure" at PNC and a Category II transition at U.S. Bancorp, both of which say the same thing: the mark will count, and both say they have planned for it.

Thirteen tests, each with the document it was read from. "Held" means the condition was tested and not met; "disclosed" means a management stated the thing the test asked about; "none" means the documents were silent and the silence branch Part 1 wrote applies; "pending" means the document that scores it is not yet filed.

The two ranges, bridged. Schwab. Part 1's arithmetic was this: 6.8% at June with $162 million of headroom; the quarter's available-for-sale move worth $1.4 billion after tax on the fitted duration and $0.9–1.1 billion on Schwab's own stated 1.9–2.3 years, taking the ratio to 6.49–6.6% static before a quarter's earnings worth 50–60 basis points. The print is 7.0%, the top of the objective. The bridge is in the release's own reconciliation. Adjusted Tier 1 capital rose $1,867 million, from $32,526 million to $34,393 million: earnings to common of $2,811 million, less a buyback cut to $0.5 billion from $1.0 billion in the second quarter, less dividends and preferred costs, and an AOCI change of $164 million. Adjusted average assets rose 2.6%, to $491.8 billion. Had AOCI moved the $1.4 billion Part 1 projected, with everything else as printed, the ratio would have been 6.74% — a basis point below the objective, after earnings and after the halved buyback. The range held because the mark did not move, and the release does not say why: the available-for-sale book shrank to $59.0 billion from $62.5 billion, the held-to-maturity book to $126.6 billion from $130.6 billion, the AOCI on the securities Schwab moved to held-to-maturity accretes back each quarter, and Schwab's "with derivatives" duration is the hedged one. Which of those did the work is the 10-Q's to say. What the print settles is narrower and exact: the quarter's rate move did not reach Schwab's capital objective and Schwab bought back half as much stock as in the prior quarter. On the call the chief financial officer said the ratio "ended at 7% as we allowed capital to build to support continued strong client activity against a backdrop of volatile rates where yields moved meaningfully higher", that "we still have $190 billion roughly of securities that are yielding sub 2%", and that full-year revenue and margin would land "towards the lower end of the ranges that we provided at the summer business update"; no one on the call asked about AOCI, the mark or the proposal, and July's one-December-hike assumption was not restated.

State Street. Part 1 had 5.3% at June against a 5.25–5.75% assumption, 13 basis points of available-for-sale move, 5.18% static before a quarter's earnings worth about 30. The print is 5.5%. State Street's own attribution for the quarter: "mainly driven by capital generated from earnings, higher preferred equity, and lower average balance sheet levels, partially offset by continued capital return", the preferred having risen to $4,055 million from $3,559 million, $496 million of new issuance. The buyback was $400 million, the same as the second quarter, and the call guided it up to about $700 million in the fourth quarter toward an 80% payout for the year; AOCI fell $471 million, and the one analyst who asked called it "not a whole lot of impact"; standardized CET1 rose to 11.0% from 10.8%, which the call attributed to lower risk-weighted assets. The range held on earnings and new preferred with the buyback unchanged, which is the "held, by what" Part 1 asked for and the second of the steelman's one-offs: the ratio was supported by issuing capital, not by the mark staying put, and State Street's available-for-sale mark moved exactly as projected.

PNC's swaps. PNC says the AOCI decline "reflected the impact of higher interest rates on securities and swaps, partially offset by the continued accretion of unrealized losses", and gives the securities leg: net unrealized losses on available-for-sale securities of $4.0 billion from $2.1 billion, about $1.46 billion after tax at 23%, of a $2.35 billion fall in AOCI. The remainder, about $0.89 billion, is the swap book and the pension items. The securities leg is the larger, so the twelfth tripwire does not fire; the exact split is in the 10-Q. PNC's own capital arithmetic now shows CET1 including AOCI at 8.8% against the 9.8% reported, and it guides fourth-quarter repurchases "to approximate" the third quarter's $0.6 billion. U.S. Bancorp, the third of Part 1's Category III names, bought back no stock in the quarter: its CET1 including AOCI fell to 9.1% from 9.4%, 40 basis points of it from the rate move, and it expects to resume repurchases "over the next quarter or two".

The marks, bank by bank. Part 1 projected the quarter's change in each company's available-for-sale mark from the lead bank's realized duration and the 5-year's 90-basis-point rise, and froze a band of 0.6 to 1.4 times around it. Six of the eleven print something that can be scored: four the available-for-sale mark itself, JPMorgan the securities component of its AOCI change, Schwab its AOCI. JPMorgan's supplement, in the CET1 capital rollforward, puts "unrealized gains/(losses) on investment securities" at a $7,893 million reduction for the quarter, after tax, against a $320 million addition in the second quarter; against the $7,210 million after-tax projection that is 1.09 times. Bank of America's net unrealized loss on available-for-sale debt securities went from $2.37 billion to $4.81 billion, a move of $2.44 billion against $2.12 billion projected, 1.15 times. PNC's net unrealized losses on available-for-sale securities went from $2.1 billion to $4.0 billion, 0.90 times, and its portfolio duration lengthened to 4.1 years from 3.6. State Street's available-for-sale portfolio went from a $56 million net gain to a $557 million net loss, $613 million against $620 million, 0.99 times. BNY's pre-tax loss on available-for-sale securities net of hedges went from $703 million to $1,934 million, $1.23 billion against $2.22 billion, 0.55 times: Part 1 projected the unhedged book and said in the same sentence that the hedges cover 56% of it, and the print says the hedges did roughly what that implied. Schwab gives no securities line but reports AOCI in its adjusted leverage reconciliation; Part 1 read Schwab's AOCI from its 10-Q as the available-for-sale mark plus the amortizing loss on the securities moved to held-to-maturity, so the total is scored as the securities mark and labelled so: $10.39 billion at September against $10.23 billion at June, a move of $164 million after tax against $1.40 billion projected, 0.12 times. The held-to-maturity marks that the three custody and consumer banks print moved 0.85 times the projection at Bank of America ($21.8 billion, to $103.9 billion), 0.92 at State Street and 0.72 at BNY.

What a total AOCI line hides. Three of the eleven — Citigroup, Wells Fargo and U.S. Bancorp — print total AOCI and nothing underneath it, and two, Goldman Sachs and Morgan Stanley, print no AOCI line at all. The two that print both lines are the caution. JPMorgan's total AOCI fell $12.1 billion in the quarter, 1.7 times the after-tax projection, while the securities line inside it fell $7.9 billion, 1.09 times; Bank of America's total fell $3.24 billion after tax while its securities mark fell $1.88 billion, so two-fifths of the move was cash-flow hedges and other items. On that evidence Wells Fargo's $8.2 billion fall (1.6 times the after-tax projection), U.S. Bancorp's $2.5 billion (1.4) and Citigroup's $2.3 billion (1.4, with currency translation inside it) are scored pending, not larger, and the hatched bars in Exhibit 1 are drawn for scale. Two fragments exist: Citigroup's regulatory capital page shows its cash-flow-hedge reserve going from a $10 million gain to a $116 million loss, and U.S. Bancorp's deck shows the AOCI-related adjustments to its capital at $8.85 billion from $6.79 billion, a $2.06 billion move or 1.17 times, on a line that also carries pension items. Three managements attributed the move. Wells Fargo's deck says the AOCI decrease took 38 basis points off CET1 — against the 36 Part 1 computed for Wells Fargo Bank's available-for-sale leg — and left the ratio at 10.0%, the bottom of its stated 10–10.5% range. Bank of America's chief financial officer said "a $1.8 billion portion of AOCI of AFS securities" reached CET1, about 10 basis points of risk-weighted assets against Part 1's 10, that hedging "continued to limit the AOCI effect of higher long-term rates on much of the AFS securities portfolio", and that "a little less than half of the total Q3 AOCI decline related to cash flow hedges and did not affect CET1". The reach test is held at both, and Bank of America's sentence is the direct evidence for scoring the total-AOCI banks pending. JPMorgan's CET1 fell 30 basis points to 13.9% on, in the chief financial officer's words, "OCI losses, higher RWA and capital distributions", which the supplement's roll does separate — $7.9 billion of securities losses, standardized risk-weighted assets up $51 billion, $4.4 billion of dividends and $1.0 billion of net share purchases — and the securities line alone is about 36 basis points of standardized risk-weighted assets against Part 1's 29 at the lead bank, which scores the reach test as held at a third firm (§4); Bank of America's fell 37 to 10.9%; Goldman's 50 to 12.4% on risk-weighted assets up $26 billion; Citigroup's and Morgan Stanley's were flat; State Street's rose 20 to 11.0%.

Crystallization. The test fires on any sale at a loss, and five banks sold at a loss. JPMorgan's Corporate segment booked $658 million of net investment securities losses, $498 million after tax and $0.18 of earnings per share, against $395 million in the second quarter — the line Part 1's eighth tripwire watched, negative again and larger. The release calls the losses "discretionary"; neither it nor the call uses the word repositioning, and whether the 10-Q does is the second leg of Part 1's eighth tripwire, read in November. Wells Fargo booked $29 million of net losses on debt securities against a $1 million gain in the second quarter; U.S. Bancorp $65 million against $49 million; BNY $25 million, the same as the second quarter; PNC $1 million, which it calls immaterial, against $139 million. Citigroup realized $64 million of gains on sales of investments, State Street none, and no release at Goldman Sachs, Morgan Stanley or Schwab carries a securities-gains line. No one disclosed a transfer into or a sale out of held-to-maturity, so the reading Part 1 attached to that branch — the reach moving upward — does not follow; Bank of America's held-to-maturity book ran off by $9.4 billion of amortized cost, Wells Fargo's by $6.7 billion, Schwab's by $4.0 billion, on schedule rather than by sale.

The test. Part 1 fitted one line through the eleven with bank books and ten regional peers — P/TBV = −0.76 + 0.165 × ROTCE, R² 0.88 — and froze a test: if the prints moved the residuals from that line by more than 0.3 times tangible book at three or more of the twelve without a matching move in their returns, the group is priced on something other than returns. Part 2 does not refit. Each company's price to tangible book is re-struck on the 15 October settled close and its 30 September tangible book, rolled forward from Part 1's June figure by the company's own reported change; its return is the nine months to September annualized on Part 1's definition; the fitted value is read off Part 1's line at that return; and the residual is compared with the one Part 1 printed. A residual can move because the price moved or because the return moved the fitted value; the test asks for the first, so each change is read with the price-to-book move beside it.

Price to tangible book is on Part 1's tangible-equity definition, rolled forward by each company's own reported change, so the test is computed on one definition throughout; at six names Part 1's June figure did not deduct preferred stock, so price divided by the printed tangible book per share gives a higher multiple (JPMorgan 2.98×, Bank of America 1.77×, State Street 2.81×), and at BNY, where Part 1's figure sat below the company's, a lower one (4.41× against 4.82×). Goldman and Schwab do not print tangible book per share.

What moved and what did not. Two names crossed the threshold on the frozen basis (the 15 October settled close, Part 1's published line), both downward toward it: State Street, from +0.25 to −0.06 times, and Schwab, from +0.64 to +0.01. The two names Part 1 handed to the valuations as residuals to explain, JPMorgan at −0.34 and Goldman at −0.36 times, moved toward the line by 0.26 and 0.12, below the threshold. The decomposition is the same at the three largest movers, and it is printed for all three: a residual change is the multiple's move less the fitted value's. At Schwab the multiple fell 0.57 (a 2.6% lower price on a tangible equity up about 6% on the quarter's earnings) against a fitted value up 0.06 on a flat return, so the change is almost all price. At State Street the multiple fell 0.13 and the fitted value rose 0.19 as the nine-month return rose to 19.3%, so the change is a little more return than price. At JPMorgan the multiple rose 0.04 and the fitted value fell 0.22 as the return normalized to 22.0% from the half-year's 23.4%, so the change is almost all return. The test counts crossings, not causes, and found two; the house keeps reading the group as priced on one line. It notes, for the next re-strike, that a 0.3× threshold is 5–6% of the multiple at Schwab and BNY, 11% at JPMorgan and Goldman, and 16–25% at the other money centers, that both crossings came at the high-multiple end, and that on the 16 October last trade — not the strike — JPMorgan's change reaches 0.30 on Part 1's unrounded coefficients. Citigroup, BNY and Morgan Stanley did not move at all. Goldman's and Schwab's tangible book are rolled from book value per share and from total equity respectively and are marked as such in the data.

The reactions. Nine of the twelve closed below their prior close on their print day and eight underperformed the financials index the same day; the index itself rose 0.4% from the 12th to the 15th. Citigroup fell 5.6% on a beat ($2.85 against $2.63 consensus) and 7.3% over two sessions; Bank of America 4.8% on reported earnings of $0.91 that carried a regulatory settlement charge ($1.08 without it, against $1.12); State Street 4.1% on a beat; Wells Fargo 3.1% on a beat; Morgan Stanley 2.7% and Schwab 2.7%; JPMorgan 0.9% on $5.57 reported, $5.75 before the securities loss, against $5.91 — a miss on both. BlackRock (+0.7%, then +5.1% over two sessions on $15.93 as adjusted, $10.57 reported, against $14.15), U.S. Bancorp (+2.0%) and BNY (+3.2%, on $2.34 reported and $2.37 adjusted against $2.27) rose. Eight of the twelve beat the consensus Part 1 archived on 8 October on reported earnings; JPMorgan and Bank of America missed on both the reported and the adjusted figure, Schwab and BlackRock missed on the reported and beat on the adjusted. Tangible book per share fell on the quarter at six of the nine that print it — Wells Fargo by 3.6%, PNC by 3.1%, U.S. Bancorp by 2.3%, BNY by 1.8%, Bank of America by 0.8%, JPMorgan by 0.6% — which is the AOCI move arriving in the denominator the group is priced on, and the mechanism by which the mark reaches a share price without ever reaching a regulatory ratio.

Where Part 1 was wrong, in plain words. First, the two ranges. Part 1 said the quarter's available-for-sale move breached Schwab's and State Street's self-set ranges on paper, before earnings, and that the prints would test whether earnings carried them or buybacks paused. State Street's did what Part 1 said: the mark moved as projected, earnings and new preferred carried the ratio. Schwab's did not: the mark moved $164 million against $1.4 billion, and on the counterfactual in §4 that one line is the whole difference between a breach and the top of the range. The projection of Schwab's mark was wrong, by a factor of eight, and the house does not yet know why; the claim that the move breached two ranges on paper is retired. Second, BNY. Part 1 projected the unhedged book and flagged the hedges; the test scores the printed number, net of hedges, and it is 0.55 times. That is a miss by the test's own rule and the lesson is in the rule: a projection that knows it is gross should have been frozen net. Third, a change in kind rather than a miss: Part 1 recorded that on nine July calls no analyst raised the marks, the election or the proposal, and that neither Schwab's nor State Street's call mentioned AOCI at all. In October JPMorgan cut its buyback to $664 million net — the UBS analyst put it as "$1 billion buyback versus the normal cadence of around $7 billion" — and its chairman gave the reason as redeploying about $30 billion of excess capital at better returns than buying the stock at roughly a 6% earnings yield, and volunteered that the smaller buyback had "nothing to do with AOCI", the environment or the capital rules; Citigroup's chief financial officer twice described the quarter's rate-driven loss in other comprehensive income as something to "absorb", Wells Fargo said it will "continue to consider the interest rate environment and the impact on accumulated other comprehensive income when determining how much stock we repurchase", and analysts touched the marks, hedging or capital at nine of the twelve calls. The silence Part 1 described broke on the management side first.

Where Part 1 held. The projection method is close where the mark is printed and the book is unhedged (JPMorgan, Bank of America, PNC, State Street). The held-to-maturity marks moved 0.72–0.92 times the projection at the three that print them. The hike has not shown in the betas. Deposits grew. The nonbank book is as invisible as Part 1 said. The two names handed to the valuations as residuals narrowed toward the line as their one-off quarter rolled out of the returns, which is the direction Part 1's §5 implied. The pricing test did not fire. The crystallization test fired exactly where Part 1's eighth tripwire said it would, at JPMorgan, and nowhere that changes the reading from an earnings drag to a liquidity question: five loss sales, none with a deposit fall, none disclosed out of held-to-maturity.

The hand-off. JPMorgan, Goldman Sachs and BlackRock were valued as the three archetypes after their prints — the universal bank, the markets house and the asset manager — and each note carries its own model and its own triggers; the residuals in §5 are the inputs those notes were handed and this note does not preview their numbers. The insurer thematic follows on 21 October.

The reversal block's observables, scored. Deposits −3% with a loss sale, or Schwab's sweep −$10 billion: not fired. A nonbank build at JPMorgan, Wells Fargo or Morgan Stanley: none at any of the three. Interest-bearing deposit costs rising more than the funds rate at the money centers: split — not fired against the 25-basis-point hike, fired against the 3.6-basis-point quarter average, at all four; Part 1 did not specify which, the house reads it as the hike for the reason given in §1, and records the other reading as fired. Retiring the nonbank concern: not met, no balances disclosed. Retiring the funding concern: partly — noninterest-bearing held share at JPMorgan and Bank of America and slipped at Wells Fargo; uninsured shares pending. Retiring the capital concern: pending the FDIC. Stablecoin supply against bank deposits, in the quarterly form: dollar stablecoins grew $2.2 billion in the third quarter against $257 billion of H.8 deposit growth, so the first of the two consecutive quarters fails the condition and the clock restarts. Treasury's final rule on the yield prohibition: not due — the proposal was published on 18 August (Federal Register 2026-16796) and its comment period closes on 19 October. The consortium token: not due — the 21-bank company forms in the second half of 2026 and targets a dollar token in the first half of 2027. The pricing residuals: two of three, not fired. Revising the system number: pending the FDIC; on the twelve, smaller at two of the six scored, against the six of eleven that would revise it downward. Dropping the two breaches: met, both ranges held with Schwab's buyback continuing at half its pace.

The pending ledger. What closes the open items, and when.

The addendum that closes the tripwire rows is dated to the FDIC's release and is the last document of this two-part note.