When It Looks Cheapest

In 2005 the American homebuilders were the cheapest-looking stocks on the board. The country was putting up more than two million new homes a year, a record, and the big builders were earning money hand over fist. On those earnings the shares traded at a handful of times profit, the kind of low multiple a value investor’s screen lights up for. Buy great American companies at single-digit earnings multiples, in the middle of the biggest housing boom in living memory. What could be safer?

Pulte went from about $48 a share to roughly $8. An 83% fall. KB Home lost around 2.6 billion dollars between 2007 and 2010. The earnings that made the stocks look cheap did not merely dip; they turned into some of the largest losses in the sector’s history, and the “low multiple” evaporated because the E in the ratio was a mirage.

This is the trap that defines a whole class of stocks, and it is the reason Peter Lynch called them the most misunderstood group he ever wrote about. Cyclicals: autos, steel, airlines, chemicals, oil, memory chips. Their profits swing with the economy, sometimes violently, and the single most expensive mistake an investor can make with them is to read their valuation the way you’d read anyone else’s.

Here is the rule, and it is genuinely his: for a cyclical, a low price-to-earnings ratio is usually a warning, not an invitation. It means earnings are at a cyclical peak, the denominator is fat, and the good times are closer to their end than their beginning. A high P/E, or no meaningful P/E at all because the company is barely breaking even, often marks the bottom, the point of maximum safety. It is the exact inverse of how the multiple works for a stalwart or a slow grower, and Lynch’s point was that people lose money precisely because the inversion feels wrong. The homebuilder buyers of 2005 weren’t reckless. They were applying a sound value habit to the one place it detonates.

Lynch’s other main instrument was the inventory report. Goods piling up faster than they sell is demand softening before management will admit it. Inventories drawing down at a beaten-up company can be the first real sign of a turn. It’s a way to read the cycle from a company’s own filings instead of guessing at the macro.

That’s about as far as I want to take Lynch here, and not because he was wrong. It’s because cyclicals, unlike the fast growers he owned so well, are territory other people have mapped at least as carefully, and the more interesting stuff is in the disagreements.

The sharpest of these comes from the capital-cycle school, most associated with the fund manager Marathon and the writer Edward Chancellor, whose book Capital Returns is the readable version. Their claim is that you’re watching the wrong dashboard. Forget sentiment and quarterly inventories; watch the supply side. High returns in an industry attract capital, capital builds capacity, and the new capacity is what eventually crushes everyone’s margins two or three years later. A depressed industry that has spent years starving itself of investment is quietly setting up the next boom. So the real leading indicator isn’t how bleak it feels or how the inventory line wiggles; it’s the capital-spending announcements, the new plants, the flood of IPOs into a hot commodity sector. Shale did exactly this: booming oil in the early 2010s pulled in a torrent of drilling capital, and that oversupply, not any collapse in demand, broke the oil price in 2014.

You can read this as a friendly upgrade to Lynch or as a rebuke. It tells you where in the cycle you are with less guesswork than “buy when it feels terrible,” and it can tell a genuine trough apart from a value trap that stays cheap for a decade. Where the two genuinely part company is timing: Lynch’s inventories might be drawing down, flashing buy, while capacity and capex are still climbing, flashing sell. Both can be true at once, and nobody hands you the tiebreak.

And then there’s the flat refusal to play at all. A large camp of serious investors thinks cycle-timing is a mug’s game, that tops and bottoms are obvious only in the rear-view mirror, that the big institutions front-run each other into false bottoms that turn out to be ledges, and that the whipsaw eats whatever edge the theory promised. Their alternative is unglamorous: don’t time anything, just own the lowest-cost producer with the strongest balance sheet, the one that survives the trough while its leveraged rivals go bankrupt or dilute themselves to death, and let it compound market share across several cycles. Nucor in steel is the stock example. This has the great virtue of not requiring you to be a prophet.

I won’t referee that. All three can be right for different temperaments, and honestly the neatest practice layers them: use the capital cycle to spot a supply-starved trough, buy the strongest operator in it, and treat Lynch’s inventory and P/E signals as confirmation the operating data is actually turning. What none of them can do is tell you the future, and I’d distrust anyone who says otherwise, including me.

Some of the ground has shifted since Lynch’s era, and it’s worth being honest about which parts. The inventory edge an attentive amateur once had is thinner now: supply chains are global, so a US carmaker’s stockroom can look clean while components pile up unseen in Asia, and satellite counts of oil tanks are on a screen for anyone who wants them. Commodity futures move ahead of reported earnings in real time, so the “numbers are turning while the news is grim” window gets arbitraged faster. Passive flows are a genuinely new distortion: a big cyclical that sits in the major indexes gets bought mechanically on its index weight, which can keep it from getting as cheap at the trough or as dear at the peak as it used to, muting the very price signal these frameworks lean on. And after the long era of near-zero rates, the interest-rate cyclicals, utilities, some financials, capital-heavy industrials, are once again being driven around by the cost of money rather than by demand, which makes Lynch’s old aside about timing them to the rate cycle suddenly current.

If you want something concrete to hold, take two things. First, put a cyclical’s current P/E against its own ten-year range, the low, the high, and roughly where it sits, and then remember to read it upside down: near the bottom of its historical range is the danger zone, not the bargain bin. Second, keep a short watch-list of gauges that actually tell you where the industry sits without a crystal ball: capacity utilization (near flat-out means peak and coming oversupply; deeply slack means trough), inventory-to-sales, capital spending relative to cash flow, and whether the relevant commodity’s futures curve is priced above or below today’s spot. None of these is a buy button. Together they beat a hunch.

The homebuilders in 2005 had every one of those gauges screaming, and a low multiple whispering the opposite. The whisper won, because a low multiple is comforting and capacity utilization at the redline is not. That’s the whole difficulty in a sentence. In a cyclical, the number that soothes you is the one trying to hurt you, and the sick feeling you get buying something when the news is unbearable and nobody else will touch it is, more often than not, the feeling of being early.

Oil traded below zero for a day in April 2020. That was the feeling. Ask anyone who bought.