In July this site published four bank earnings analyses within days of each other. Each quoted its company's CET1 capital ratio for the quarter ended 2026-06-30. Each attached an adjective. None of the four named another bank's number.
Put them in one column and the adjectives run backwards:
| Bank | CET1 | What our article called it |
|---|---|---|
| SOFI | 18.7% | "Capital keeps piling up" — and the only one raising "idle capital" |
| JPM | 14.1% | "CET1 sits at a fortress-like 14.1%" |
| GS | 12.9% | "returning more capital and building its buffer simultaneously" |
| BAC | 11.2% | "CET1 at 11.2% ($202B) leaving real headroom" |
The highest capital ratio on the site drew the most sceptical description. The lowest drew the most reassuring one.
That looks like four writers contradicting each other. It is not. All four adjectives are correct, and understanding why is the single most useful thing a reader can take from a regulatory ratio: the number alone does not tell you whether it is comfortable. Only the number set against that bank's own requirement does.
A note on the column. These are Standardized CET1 ratios. Where a bank reports two bases they diverge — Goldman's Advanced ratio is 13.7% against its Standardized 12.9%, JPMorgan's is 14.2% against 14.1% — and mixing bases inside one comparison would manufacture a ranking out of methodology. SoFi and Bank of America each report a single figure.
The one requirement we can source, and the three we cannot
SoFi's capture states its requirement outright: 7.0%. At 18.7% it holds 2.7x its minimum — that division is the entire reason the sceptical adjective is fair.
For JPMorgan, Goldman and Bank of America, this repository does not hold the equivalent number. I searched all three captures for requirement, minimum, SCB, stress capital, GSIB and surcharge: zero matches in each, against the SoFi capture where the same search does hit. Those thresholds are public in each bank's own regulatory filings. They are not here, and this piece will not supply them from memory.
That absence is the finding rather than a hole in it. A global systemically important bank carries capital surcharges and a stress capital buffer that lift its floor well above 7%, so JPMorgan at 14.1% is operating far closer to its own bar than SoFi is to hers — even though SoFi's number is 4.6 points higher. The adjectives track headroom, which inverts against the raw ratio. Both halves of that sentence are true at once.
Which means a reader of three of those four articles was shown a number and given no way to tell whether it was comfortable. That is not a criticism of the articles; each described its own bank accurately against a bar its readers could not see. It is a statement about what a ratio needs printed beside it.
And the ratio is not the only ranking available. Bank of America's 11.2% is the lowest percentage in the table and the largest absolute capital position in it: $202B. Ratios and absolutes rank differently, and a reader handed only the first will draw a conclusion the second contradicts.
Why SoFi's ratio rose while it was growing fastest
Worth one paragraph, because growth normally consumes capital and SoFi's ratio went the other way — 18.7%, up from 14.3% a year earlier — during a quarter of record volume.
From its capture: originations hit a record $14.8B, up 69%, while deposits rose $5.3B in the quarter to $45.5B, more than 90% of average total liabilities. The rate paid on those deposits ran 156bps below its warehouse facilities, worth roughly $712.6M annualised. Growth funded by cheap deposits rather than by capital lets the balance sheet expand while the capital ratio climbs.
One label that matters: SoFi's figure is the company's own estimate, marked "(June 30, 2026, estimated)" in the release. The other three are reported.
The same lesson, one metric over
The site tells readers what to use instead of the Rule of 40 for banks. Our Rule of 40 explainer recommends net interest margin, return on tangible common equity, the efficiency ratio and credit quality. Return on tangible common equity has the identical problem, and JPMorgan's own release demonstrates it inside a single line.
JPMorgan reports ROTCE of 29% — and, in the same release, 23% excluding significant items. The gap is two one-off gains: a $4.6B net gain on Visa shares worth $1.27 of EPS, and $1.0B of gains on equity investments worth $0.29. Together they take reported EPS of $7.70 down to $6.14 ex-items. The capture's own note is blunt about it: the reported $7.70 "is not clean run-rate."
Goldman reports ROTE of 25.5%.
So on the reported figure JPMorgan leads Goldman by three and a half points. On the ex-items figure it trails by two and a half. The ranking inverts on the metric we recommend, and which answer you get depends entirely on a basis nobody states.
Three reasons not to rank those two numbers anyway
Each of these is a caveat that survives the arithmetic, and together they are this piece's thesis arriving a third time.
- They are not the same measure. Goldman labels its figure ROTE; JPMorgan and Bank of America label theirs ROTCE (Bank of America's is 17.03%). Near-identical, not identical, and a straight ranking across the two names hides that.
- Only JPMorgan publishes an ex-items version. Goldman and Bank of America disclose no such adjustment. That is not evidence their figures are clean — only that no adjusted figure exists to compare. "Goldman's 25.5% is the clean number" is the easiest wrong sentence available here, and this repository cannot support it.
- The efficiency ratio repeats it one level down. Goldman's is 57.4%, from 63.4%; Bank of America's is 59%, from 63% — comparable. JPMorgan reports an "overhead ratio" of 48% reported / 47% managed: the same concept under a different name and a different perimeter. Three banks, three namings.
What this piece does not claim
- That any of the four articles was wrong. They were not. Each described its own bank correctly, and the language tracks headroom even where the raw numbers rank the other way.
- That any of these banks is under- or over-capitalised. Three of the four requirements are not in this repository, so no such judgement is available from it — that is the whole point.
- That SoFi should or should not deploy its capital. Its capture leaves the question open and our July analysis already put both readings on the page.
- Anything about share prices. This repository stores no previous close for any company.
A capital ratio, a return on tangible equity, an efficiency ratio: three numbers a reader meets constantly, and none of them means anything alone. The first needs its regulatory floor. The second needs its basis — reported or ex-items. The third needs to be the same metric at both companies before it is a comparison at all. Four articles on this site printed the first without its floor, and the reason none of them looked wrong is that separately, none of them was.
Every reported figure is from this site's captures of the four Q2 2026 releases, all for the quarter ended 2026-06-30. SoFi: CET1 18.7% ($9.11B), explicitly marked estimated, against 14.3% a year earlier and a stated 7.0% requirement; record originations $14.8B (+69%); deposits $45.5B, up $5.3B in the quarter and over 90% of average total liabilities; deposit rates 156bps below warehouse facilities, ~$712.6M annualised. JPMorgan: Standardized CET1 14.1% (Advanced 14.2%), ROTCE 29% and 23% ex-significant-items, overhead ratio 48% reported / 47% managed, the $4.6B Visa gain (+$1.27 EPS) and $1.0B equity-investment gains (+$0.29 EPS), and EPS of $7.70 reported against $6.14 ex-items. Goldman: Standardized CET1 12.9% (from 12.5% QoQ, Advanced 13.7%), ROTE 25.5%, efficiency ratio 57.4% from 63.4%. Bank of America: CET1 11.2% Standardized ($202B), ROTCE 17.03%, efficiency ratio 59% from 63%. The four quoted adjectives are verbatim from this site's own July analyses of each bank, linked above and in our July coverage. Derived, and the only derivation here: SoFi's 2.7x coverage of its requirement (18.7 ÷ 7.0). The capital requirements for JPMorgan, Goldman and Bank of America are stated as absent rather than estimated — searching all three captures for "requirement", "minimum", "SCB", "stress capital", "GSIB" and "surcharge" returns no matches in any of them, while the same search does match the SoFi capture. No ex-items figure is supplied for Goldman or Bank of America, because neither release discloses one. No share price, price move or market-capitalisation figure appears here.