How do I use the Graham Number to find undervalued bank stocks?

Calculate the Graham Number for each bank and compare it to the current share price. If the share price is below the Graham Number, the stock may be undervalued, but you need to verify that the earnings and book value behind the calculation are sustainable before acting on it.

The Graham Number gives you a single number you can compare against a bank's share price to spot potential bargains. But the real skill isn't in the calculation itself. It's in knowing what to check after the number flags a candidate.

Setting Up the Screen

Start by calculating the Graham Number for each bank in your target universe. The formula produces an estimated fair value based on earnings per share (EPS) and book value per share (BVPS). Any bank whose share price falls below its Graham Number is trading below Graham's conservative estimate of intrinsic value.

The gap between the Graham Number and the share price tells you how large the discount is. If a bank's Graham Number comes out to $50 and the stock trades at $38, that's roughly a 24% discount. Many value investors look for at least a 15-20% discount before digging deeper, since a narrow gap can disappear quickly with normal price fluctuations.

Suppose you're screening a list of 50 community and regional banks. You calculate the Graham Number for each one and find that 12 are trading below the formula's output. Seven of those show discounts of 15% or more. Those seven are your starting candidates, not your buy list.

Checking Whether Earnings Are Real

The Graham Number is only as reliable as the EPS figure you feed into it. For banks, earnings can be distorted by several factors that inflate or deflate the number temporarily.

One-time gains are a frequent culprit. If a bank sold a branch or booked a large recovery on a previously charged-off loan, that windfall flows through EPS but won't repeat. Abnormally low loan loss provisions have the same effect. During periods of strong credit quality, banks sometimes provision well below their historical average, which boosts reported earnings above what the bank can sustain through a full credit cycle.

Compare the trailing twelve-month EPS to the bank's three-to-five year average. If recent EPS is significantly higher than the historical norm, recalculate the Graham Number using the average instead. A bank that looks 25% undervalued using peak EPS might look fairly valued using normalized earnings.

Checking Whether Book Value Is Solid

Book value per share is the other input, and it needs the same scrutiny. Two issues come up frequently with bank stocks.

The first is goodwill from acquisitions. When a bank buys another bank at a premium, the excess purchase price sits on the balance sheet as goodwill. This inflates BVPS without adding tangible asset value. If goodwill makes up a meaningful portion of total equity, recalculate the Graham Number using tangible book value per share (TBVPS) instead. The result will be lower, sometimes substantially so.

The second is unrealized losses in securities portfolios. Banks that hold large portfolios of bonds purchased when interest rates were lower may carry significant unrealized losses. Under accounting rules, held-to-maturity securities appear at amortized cost rather than market value, so reported BVPS can overstate the economic value of equity. Check the bank's AOCI (accumulated other comprehensive income) adjustments and any disclosures about unrealized losses in the HTM portfolio.

Checking Asset Quality and Capital

A bank can trade below its Graham Number for perfectly good reasons. If credit quality is deteriorating, the discount is the market pricing in future losses that haven't hit earnings yet.

Before concluding that a stock is a bargain, review these indicators:

Non-performing loan ratio: rising NPLs signal that borrowers are struggling to repay, which will eventually reduce EPS through higher provisions and charge-offs

Net charge-off trends: are actual loan losses increasing quarter over quarter? A rising trend is a warning sign, even if the NPL ratio hasn't moved much yet

Reserve coverage ratio (allowance for loan losses divided by non-performing loans): a ratio below 100% means the bank hasn't set aside enough reserves to cover its current problem loans, let alone any additional deterioration

Equity-to-assets ratio: thin capitalization limits a bank's ability to absorb losses and may attract regulatory scrutiny. Banks with equity-to-assets below 8% are operating with less of a cushion

Any single red flag doesn't necessarily disqualify a candidate. But if you see NPLs climbing, charge-offs accelerating, and thin capital simultaneously, the Graham Number discount is probably reflecting genuine risk rather than market oversight.

Cross-Checking with Other Valuation Methods

The Graham Number offers one perspective on value. Confidence increases when other approaches point in the same direction.

Compare your Graham Number findings against price-to-book (P/B) analysis. A bank trading below its Graham Number and below the P/B ratio justified by its return on equity (ROE) gives you two independent signals saying the stock is cheap. If the Graham Number says undervalued but the justified P/B framework suggests fair pricing, the disagreement itself is worth investigating. The Graham Number may be reflecting temporarily elevated inputs, or the P/B model may be using an ROE assumption that's too conservative.

The dividend discount model provides yet another cross-reference. If three different methods all produce a fair value above the current share price, the case for undervaluation is much stronger than any single model can make on its own.

Value Traps to Watch For

Not every cheap bank stock is a good investment. Some banks trade persistently below their Graham Number because the market recognizes problems that the formula can't capture.

Banks with concentrated loan portfolios present one common value trap. A community bank with 60% of its loans in commercial real estate in a single metro area faces concentration risk that generic metrics won't fully reflect. If that local market weakens, the entire portfolio comes under pressure at once.

Pending regulatory actions are another source of persistent discounts. Banks operating under consent orders, memorandums of understanding, or other supervisory actions often trade well below calculated fair value. The Graham Number has no way to account for the uncertainty of regulatory outcomes or the operational restrictions that come with enforcement actions.

Management quality matters too. A bank with solid current numbers but a history of aggressive growth, poor underwriting decisions, or high executive turnover may deserve its discount. The Graham Number measures where the bank stands today, not where it's heading.

How Bank Type Affects the Approach

The screening process works differently depending on what type of bank you're evaluating.

For community banks (under $10 billion in assets), the Graham Number often works well because these banks tend to have simpler balance sheets, more predictable earnings, and book values that closely approximate tangible asset value. The main adjustment to watch for is goodwill from local acquisitions.

Regional banks ($10-100 billion) introduce more complexity. These banks are more likely to have significant securities portfolios, fee income streams that add earnings volatility, and goodwill from roll-up acquisition strategies. The gap between BVPS and TBVPS tends to be wider, so always check both versions of the calculation.

For the largest banks (above $100 billion), the Graham Number becomes less useful as a primary screening tool. These institutions have complex balance sheets, significant trading operations, and earnings driven by sources like investment banking, trading, and wealth management that don't fit neatly into Graham's framework. The formula was designed for simpler businesses.

Putting It All Together

A practical workflow using the Graham Number follows this general progression:

Calculate the Graham Number for each bank in your target universe and flag any trading at a 15-20% or greater discount to the formula's output

Filter out banks with obvious quality issues: ROE below 8%, return on average assets (ROAA) below 0.80%, or efficiency ratio above 70%

For remaining candidates, normalize EPS by comparing to the three-to-five year average and recalculate the Graham Number if recent earnings look unusually strong

Check book value quality, recalculating with TBVPS if goodwill is significant

Review asset quality indicators for signs of deteriorating credit

Cross-check against at least one other valuation method to see whether the undervaluation signal holds up

Perform individual due diligence on the final candidates, including reading recent earnings calls and regulatory filings

The banks that survive this entire funnel are the ones where the Graham Number discount is supported by solid fundamentals rather than explained away by hidden risks.