Cyclicals and normalised earnings

Every so often you will find a company that looks almost too good to be true. Its profits are at a record. Its P/E ratio — the price you pay for each rupee of the company's yearly profit — is a tiny 5 or 6, when the market average is 22. On every screen and every filter, it shouts cheap.

Very often, it is the most dangerous stock on the page.

The company is a cyclical: a steel maker, a cement plant, a sugar mill, a shipping line, a car maker. Its profits do not grow in a steady line. They swing — up hard in a boom, down hard in a bust — driven by a cycle it does not control. And the cruel trick of a cyclical is that its valuation signals run backwards. It looks cheapest exactly when it is most dangerous, and dearest exactly when it may be a bargain.

This module is about seeing through that trick. The tool is simple to name and harder to do well: stop valuing the company on this year's profit, and start valuing it on what it earns in an average year, across the whole cycle.

A cyclical business is one whose profits swing widely with an outside force — the price of a commodity, the credit cycle, the housing cycle, global demand — rather than staying roughly steady year to year. When steel prices are high, a steel maker's profit balloons, because most of its costs are fixed and every extra rupee of price drops almost straight to the bottom line. When steel prices fall, the same fixed costs crush the profit just as fast. The company barely changed; the cycle did the work.

Now put that swinging profit under a P/E. The P/E is price divided by this year's earnings. At the top of the cycle, the earnings — the number on the bottom of the fraction — are swollen. Divide today's price by a swollen number and you get a small multiple. The stock looks cheap. But the market is not stupid: it knows the peak profit will not last, so it refuses to pay a high multiple for earnings it expects to shrink. The low P/E is not a discount. It is a warning, written in the one place beginners are trained to read as good news.

At the bottom of the cycle, the reverse happens. Earnings have collapsed to almost nothing. Divide the price by a tiny number and the P/E explodes — 40×, 60×, sometimes the company makes a loss and there is no P/E at all. The stock looks frighteningly expensive. Yet this can be the moment of greatest value, because the market is already looking past the trough to the recovery it expects. A high multiple on depressed earnings can be cheaper than a low multiple on peak earnings.

This is the inversion at the heart of the module: for a cyclical, a low P/E often signals danger and a high P/E often signals opportunity. The reflex that serves you well for a stable company points you the wrong way here. That is why cyclicals get their own module in this part on special cases — the standard tools do not just weaken, they reverse.

The repair is to build your valuation on normalised earnings — an estimate of what the company earns in an average year across a full boom-and-bust cycle, rather than at either extreme. You are asking a simple question: strip out the luck of where we happen to sit in the cycle today, and what does this business earn through the middle of it?

There are two honest ways to get there, and using both as a cross-check is wise.

Average the past. Take the company's profit — or better, its return on capital, or its operating margin — over a full cycle, ideally seven to ten years so that at least one boom and one bust are inside the window. Average them. That average is a rough mid-cycle margin: the margin the company earns in a normal year, between its peak and its trough. Apply that mid-cycle margin to today's sales, and you have a normalised profit.

Reason from the middle. Ask what a fair, sustainable price for the commodity is — its long-run cost of production plus a normal profit for the industry. Ask what margin the company earns at that price. That is another route to the same mid-cycle number, and it catches the case where the whole past decade was itself an unusual boom.

Here is the arithmetic, worked simply. illustrative

Take a steel maker. This year, with steel prices well above their ten-year average, it earned a record profit of ₹1,200 crore. But over the last full cycle its return on capital averaged out to roughly half of this year's — some years it earned ₹1,500 crore, several it earned ₹300 crore, and one bad year it lost money. A fair estimate of its normalised, mid-cycle profit is around ₹600 crore, not ₹1,200 crore.

Now watch the two multiples. At today's price the market values the company at ₹6,000 crore.

• On this year's ₹1,200 crore profit, that is a P/E of 5× — dazzlingly cheap.
• On normalised ₹600 crore, it is a P/E of 10× — ordinary, and honest.

The 5× was an illusion created entirely by dividing into a peak. The 10× is what you are actually paying for the average earning power of the business. When you hear that "you should buy cyclicals at a high P/E and sell them at a low P/E," this is what it means: the high P/E appears at the trough, when normalised earnings say the business is cheap, and the low P/E appears at the peak, when normalised earnings say it is dear.

Walk one composite valuation the normalised way. illustrative

A cement maker sells 20 million tonnes a year. Right now, in a construction boom, it earns an operating profit of ₹1,000 per tonne — a fat margin, because cement prices are high and its plants are running full. This year's operating profit: ₹2,000 crore. On today's market value of ₹10,000 crore, the stock trades at 5× operating profit. The screen says bargain.

Do the normalising work instead. Over the last full cycle — a boom, a slump when new plants flooded the market, and the recovery since — this company's operating profit per tonne averaged about ₹550, not ₹1,000. In the worst year it was ₹200; in the best, ₹1,050. The ₹1,000 it earns today is near the top of that range, and it is high for a reason every cycle repeats: fat margins have already tempted rivals to announce new capacity, and when that capacity arrives, prices soften. This is competition fading returns to the middle — the single force that turns every peak into a trough.

So value the average year. Normalised operating profit is 20 million tonnes × ₹550 = ₹1,100 crore. On the ₹10,000 crore market value, that is roughly 9× normalised operating profit — a fair, unremarkable price, not a bargain. The 5× was the cycle flattering the company, not the company being cheap.

Then discount, gently, what that normalised stream is worth. A steady ₹1,100 crore of mid-cycle profit, taxed and turned into cash, is what you are really buying — and a business is worth the cash it produces over its life, not the cash it happens to produce in its best twelve months. The honest verdict is not "cheap" or "dear" but fairly priced for a cyclical at the top of its cycle — which is a very different, and far more useful, thing to know than "5×, buy."

Normalising is a defence against the peak-earnings trap. It is not a crystal ball, and it hides its own dangers if you trust it too far.

It cannot tell you when the cycle turns. You may correctly judge that a stock is priced at its peak and still watch the boom run for two more years before it breaks. Normalising tells you where you are in the cycle, never how long the current phase will last. Anyone who claims to time the turn is guessing with confidence.

It cannot handle a structural change hiding inside a cyclical one. Sometimes what looks like a temporary peak is a genuine, permanent improvement — a cost breakthrough, a shift to higher-value products, a rival gone bankrupt for good. Then the old average understates the new normal, and mechanically averaging the past would make you too bearish. The reverse also bites: a company in permanent decline whose past average flatters a future that will never return. The average of history is only a guide if the future rhymes with it.

It cannot rescue a bad balance sheet through the trough. A cyclical with heavy debt can go bankrupt in the bust before the recovery you normalised for ever arrives. Mid-cycle earning power means nothing to a company that does not survive the down-leg. For cyclicals, the balance sheet is not a footnote — it is whether your normalised value ever gets a chance to be right.

And it cannot give you false precision. "Normalised profit is ₹600 crore" is an estimate wrapped around a wide range — the real answer is "somewhere around ₹500 to ₹700 crore." Better to be roughly right about the middle of the cycle than precisely right about a peak that is about to end.

The cyclical is one of the great graveyards of beginner money, and it catches people in the same handful of ways.

1. Buying the low P/E at the top. The screener flags a 5× P/E; the buyer sees a bargain and never asks what the denominator is doing. They have bought peak earnings priced to fall.

2. Selling the high P/E at the bottom. The same reader, holding through the bust, panics when the P/E hits 50× on collapsed earnings and sells near the trough — exactly when normalised value said hold. The multiple frightened them out at the worst moment.

3. Extrapolating the peak forever. The forecast simply takes this year's fat margin and rolls it forward for a decade. That single assumption — peak margins as the base case — quietly does all the damage, and it ignores the competition that every fat margin summons.

4. Ignoring the balance sheet. Focusing on mid-cycle earning power while the company carries debt it cannot service through the trough. The value was real; the company did not live to see it.

5. Mistaking a structural break for a cycle (and the reverse). Averaging the past blindly, without asking whether the world has actually changed. Sometimes the peak is the new normal; sometimes the average is a dying business's memory of better days.