How the model values a bank: JPMorgan

Banks break most valuation tools. JPMorgan is the cleanest demonstration in the model of why - and of what happens when you use the right tools instead.

What this example teaches

• Why DCF, EV/EBITDA, and P/S are structurally wrong for banks
• Which methods the model substitutes, and why four of them converge
• How outlier filtering keeps one broken method from distorting a blend

Model snapshot - June 11, 2026. Price $309.10. Blended model fair value estimate $180.70, which sits 41.5% below the price. Check the live strip above for today's values.

Why the usual toolkit is switched off

For a bank, deposits are raw material, not debt in the corporate-finance sense - so "enterprise value" is ill-defined, and EV/EBITDA is not computed into the blend at all. "Free cash flow" is equally treacherous: loan growth consumes cash and looks like negative FCF; loan runoff releases cash and looks like a windfall. JPMorgan screens at a 12% trailing FCF yield, which sounds extraordinary until you realize bank FCF mostly measures balance-sheet motion. The model computes a DCF anyway ($410.35 on the snapshot date), and the blend gives it no sector weight - it is listed on the stock page as excluded.

The bank toolkit, and a striking agreement

What banks get instead: price-to-book and justified P/B (banks are levered spread businesses on a marked balance sheet), the dividend discount model (banks are mature payout machines), residual income (which prices the spread between return on equity and the cost of equity), and CAPE (ten-year average earnings, which smooths the credit cycle).

Method	Model fair value (6/11/26)	Note
CAPE (10-yr earnings)	$172.68	cycle-smoothed
DDM	$177.86	dividend stream
Residual income	$179.02	ROE vs. cost of equity
P/B (sector-calibrated)	$190.57	balance-sheet anchor
P/E (sector-calibrated)	$268.42	trailing earnings
PEG	$588.83	excluded as outlier-prone for banks

The first four are independent calculations - different inputs, different formulas - and they land within about 10% of one another, between $173 and $191. That convergence is the headline. When methods built for the sector agree this tightly, the blended estimate ($180.70) is on much firmer footing than any single formula. PEG, built for growth stocks, produces $589 on a bank and illustrates exactly why sector weights exist.

Reading a -41% gap honestly

On the snapshot date JPMorgan traded at 2.4x book with a trailing P/E of 15 - historically rich levels for a money-center bank, which is why the book-anchored methods sit far below the price. The model's reading: the price embeds expectations (sustained high ROE, benign credit) well above what trailing fundamentals alone support. That is not a statement about what the stock will do. Banks have traded far above book-anchored estimates for years at a time when returns on equity stay elevated. What the model contributes is the size of the premium and exactly which assumptions it rests on - both visible, line by line, in the stock page's "Why this fair value" panel.

A new tool arrives: P/TBV (added June 12, 2026)

One day after this walkthrough was published, the model gained a bank-specific method this page argued was missing: justified price-to-tangible-book. It values a bank off TANGIBLE book - equity minus goodwill and intangibles, the capital that actually absorbs credit losses and earns the spread - scaled by how far the bank's return on tangible equity (ROTE) exceeds a sustainable growth rate, relative to its cost of equity:

FV = TBVPS x (ROTE - g) / (CoE - g)

On its first day, the method read JPMorgan at $248 - above the $173-191 cluster the other bank methods formed. That gap is itself a lesson: methods built for the same sector can still disagree, because they credit different things. Plain P/B treats every dollar of book alike; P/TBV strips the goodwill out and then scales the multiple directly with how far the bank's ROTE exceeds its cost of equity. JPMorgan carries one of the higher ROTE readings among large banks in this dataset, so this method produces a higher estimate than the book-anchored methods that do not differentiate on returns. Whether that premium is durable is exactly the kind of assumption the formula makes visible instead of hiding.

Figures are model estimates computed from public fundamentals under stated sector assumptions, as of June 11, 2026 (P/TBV reading as of June 12, 2026). They are educational illustrations, not investment advice or a prediction of future prices.