Can a low multiple be the peak of the cycle in disguise?

Two more forces bend every multiple: where a business sits in its cycle, and how high its quality is. Miss these and a multiple can point you the exact wrong way — most famously with cyclical companies, where a low P/E can appear at the worst possible moment.

The cyclical trap

Cyclical businesses — miners, carmakers, steel, chemicals, airlines — have earnings that swing wildly with the economy. Here's the counter-intuitive part:

• At the top of the cycle, earnings are at their peak, so the P/E (price ÷ big earnings) looks low — deceptively cheap.
• At the bottom of the cycle, earnings are crushed or negative, so the P/E (price ÷ tiny earnings) looks high — deceptively expensive.

This inverts the usual instinct. For a deep cyclical, a low P/E can be a warning that earnings are peaking, and a sky-high P/E can appear right as things are about to recover. Professionals handle this by looking at earnings averaged across a full cycle (a "normalised" multiple) rather than trusting a single peak-or-trough year. Whenever you see a suspiciously low multiple on a cyclical name, ask first: where in the cycle are these earnings?

Quality bends multiples too

Not all earnings deserve the same multiple. A company with high quality earnings — durable competitive advantages, high returns on capital, low debt, predictable demand — rationally commands a higher multiple than a low-quality peer, because those earnings are more likely to persist and grow. So a quality company at a P/E of 25 and a shaky one at a P/E of 12 might both be fairly priced. The higher multiple is buying more reliable, more durable earnings.

This is why "buy the low multiple" is such a naive rule: it treats a dollar of fragile, cyclical, declining earnings as equal to a dollar of durable, growing, high-return earnings. The market usually doesn't — and the multiple gap between them is often that judgement, not a mistake.

The reframe

Put cycle and quality together and a multiple becomes a question about the earnings underneath it: how repeatable are they, and where in the cycle were they earned? A number can't answer that. You have to look through it to the business.

In the data

Normalising earnings across a cycle means reading every year of revenue and net income, not the newest one. Here are eleven fiscal years of a steelmaker, the swing one P/E compresses:

How far back that history reaches is not uniform, and the IPO date is only a hint at it. For companies listed long ago the annual history in this data starts at fiscal 1985 whatever the listing date (Apple, Intel and Nucor all begin there), while Rivian's begins at fiscal 2019, two years before its 2021 listing (measured 28 September 2026). A business that listed after the last downturn has no trough year in the data at all, so the "full cycle" you average over may be half a cycle.

Try it now

1. Take the steelmaker above and read its multiples. One number, one moment in a cycle:

2. Now find the swing that number compresses, in the eleven years under "In the data". Find the best and the worst net income and say how many times larger the first is. Then average the eleven and divide the latest year's net income by that average. Normalising earnings means averaging the history, not trusting its newest year. Multiply step 1's trailing P/E by that latest-to-average ratio for a rough normalised multiple (rough, because the trailing figure covers the last four quarters, not the fiscal year), and set the two side by side.
3. Check how far back the history actually reaches, because the answer bounds the exercise. The last row of the eleven-year table is the oldest fiscal year shown, not the oldest held. The IPO date is what people reach for:

The caption gives the real boundary, fiscal 1985, so for the steelmaker the history reaches past several downturns. For a company that listed recently it would not; write down how much of a cycle eleven years covers here, and what you would be averaging if the history began in 2021. 4. For the quality half, compare a widely-admired stable company against a struggling peer in the same sector:

Read the return on equity and the operating margin first, then note the multiple gap that goes with the quality gap. If one trailing P/E reads 0, use the forward multiple and the price-to-book instead, and say why. 5. For any low multiple you find, ask the two context questions out loud: cyclical peak? low quality? Answering them is the analysis.