P/E ratio for cyclical stocks

This is the most expensive mistake retail investors make in this sector, and everything else here turns on it. Never value a cyclical on trailing P/E. The lowest trailing P/E of the cycle shows up at the peak, and it's a sell signal. It only looks like a buy.

Why does the P/E look lowest at the peak?

Why does it happen? Near the top of a cycle, earnings explode, because price is far above mid-cycle while costs lag. But the market has seen cycles before, and it refuses to capitalize peak earnings at a normal multiple. It caps the price, because it knows E is about to mean-revert. So P/E = huge price ÷ enormous E, which gives you a small number. At the trough you get the mirror image. E collapses toward zero or below, the market prices the recovery, and P/E = modest price ÷ tiny E comes out huge or meaningless. The P/E is measuring where you are in the cycle, read upside down.

It tells you nothing about cheapness.

The pattern has a name, the Molodovsky effect. Peter Lynch put it bluntly in One Up on Wall Street: buying a cyclical after several years of record earnings, when the P/E has hit its low point, is a proven method for losing half your money quickly.

Worked example: Deccan Steel through one cycle

Here's the full cycle, worked. Deccan Steel Ltd (synthetic, but calibrated to the real Indian steel cycle of roughly FY16 to FY24) has 5 mt of flat-steel volumes, ₹12,000 cr of net debt throughout, 100 cr shares, D&A of ₹1,800 cr, interest of ₹1,000 cr and a 25% tax rate. EBITDA = 5 mt × EBITDA/tonne, and PAT = (EBITDA − 2,800) × 0.75.

Check every EPS yourself once. For FY6, (13,000 − 2,800) × 0.75 = ₹7,650 cr, which over 100 cr shares is ₹76.5. Now read the table the way the market lived it:

• FY6, the peak: EPS is ₹76.5 and the stock is at ₹390, an apparently absurd 5.1× P/E while the Nifty trades at 20×+. Business channels run "cheapest large-cap in India" segments. Buy here on the P/E and you lose 58% by FY8, as the stock goes from 390 to 165, while having been "right" that the P/E was low. That low P/E was the market correctly refusing to pay for peak earnings.
• FY2, the trough: EPS is ₹1.5 and the stock is at ₹135, a terrifying 90× P/E. Had EPS gone negative, there'd be no P/E at all. Yet here the cycle-aware investor gets interested. The price is 40% below mid-cycle value, a figure you'll compute in the six-step valuation. High-cost global supply is shutting, and the capital cycle is turning. Buying here and selling at the FY6 peak returned +189%.
• Drill the heuristic that follows. In deep cyclicals a high or meaningless trailing P/E on depressed earnings means investigate. A low single-digit trailing P/E on record earnings means danger. There's still information in the P/E. You just have to read it as a cycle clock instead of a value gauge.
• The same inversion hits EV/EBITDA on trough EBITDA, which looks wildly expensive at the bottom. It hits the "record dividend yield" at the peak too, paid out of peak cash flows that are about to halve. Every earnings-anchored multiple has to be re-anchored to mid-cycle earnings before it means anything, and the six-step method shows you how.

Why does debt look safest at the peak?

Now look at the last column, a preview of the leverage lesson in the KPI panel. The same ₹12,000 cr of debt is a comfortable 0.9× EBITDA at the peak and a covenant-testing 4.0× at the trough, and the company hasn't borrowed a rupee more. In this sector, leverage ratios are functions of the cycle. Lenders extend credit at the top, when the ratios look great, and pull it at the bottom, when it's needed.