Normalized Earnings Power for Cyclicals: A Mid-Cycle Framework

If you value a cyclical stock on trailing earnings, you will overpay at the top and refuse to buy at the bottom. That is the entire problem. The fix is to estimate what the business earns on average across a full cycle — its normalized earnings power — and value it off that number instead of whatever the last twelve months happened to deliver.

This post walks through a repeatable framework: how to define the cycle, how to estimate through-cycle margins, how to back into mid-cycle EPS, and where the method quietly breaks.

Why trailing earnings mislead for cyclical stocks

A cyclical is any business whose volumes, prices, or margins swing meaningfully with the economy, commodity prices, or end-market capex. Think steel, semis equipment, autos, chemicals, homebuilders, freight, capital markets, ad-dependent media. At the peak, margins look structurally high, P/E ratios look cheap (because the E is inflated), and the stock often trades down. At the trough, margins look terrible, P/E ratios go to infinity or negative, and the stock is often a bargain.

The market understands this — that's why cyclicals routinely trade at low multiples on peak earnings and high multiples on trough earnings. Your job is to normalize the E so the P/E actually means something.

How to define a full cycle for through-cycle analysis

Before you can average anything, you need to pick a window that contains a real peak and a real trough. Some rules of thumb:

• Industrials and materials: 7-10 years usually captures one full capex/demand cycle.
• Semis and semis equipment: 4-6 years per cycle, but cycles have been getting longer and more amplitude-heavy. Use the last two cycles if possible.
• Housing and autos: 8-12 years. Older downturns may distort long averages — decide whether to include them as one-off events or as real troughs you want represented.
• Capital markets (banks, brokers, asset managers): Tie the cycle to rate and credit conditions, not just GDP. Different regimes may require separate analysis.

The practical test: does your window contain at least one year where the company clearly over-earned and one where it clearly under-earned? If not, extend it.

Estimating through-cycle margins

This is the core of the exercise. The goal is an EBIT margin (or EBITDA margin, or ROIC, pick your level) that you believe the business can earn on average over a future cycle.

Three methods, in increasing order of effort:

1. Simple historical average. Take 7-10 years of reported operating margin, average it. Quick, dirty, and often fine. Two adjustments worth making: strip out clearly non-recurring items (litigation, restructuring, COVID stimulus distortion), and weight more recent years slightly higher if the business mix has shifted.

2. Peak-trough midpoint. Identify the highest and lowest margin year in your window and take the midpoint. This is rougher but useful as a sanity check on method 1. If your historical average and your peak-trough midpoint disagree by more than ~200 bps, ask why — usually it means your window has more peak years than trough years (or vice versa).

3. Build it up from unit economics. Estimate mid-cycle volumes, mid-cycle pricing or spread, and a normalized cost structure. This is what sell-side analysts do for commodity producers — pick a mid-cycle steel price, oil price, DRAM price, freight rate. The advantage is you can stress-test the assumption directly. The disadvantage is you now own the commodity forecast.

Whichever method you use, ask: has the business structurally changed? A steelmaker that closed half its high-cost capacity should not be normalized to its pre-restructuring margins. Semis equipment companies post-EUV have a different mid-cycle than pre-EUV. Don't just average through real structural breaks.

Building the mid-cycle EPS number

Once you have a through-cycle margin, the rest is mechanical:

1. Pick mid-cycle revenue. Trend-line the top line through the cycle. For most industrials, this means taking the trailing revenue and adjusting toward the trend. If TTM revenue is well above the long-run trend, mid-cycle revenue is lower than today.
2. Apply the through-cycle margin to get mid-cycle EBIT.
3. Subtract a normalized interest expense and tax rate. Don't use today's interest expense if the balance sheet has changed materially.
4. Divide by current share count (or projected, if buybacks are a real part of the story) to get mid-cycle EPS.

Now you have something you can put a multiple on. The right multiple for a mid-cycle EPS is not the trough multiple or the peak multiple — it's something close to the market multiple, adjusted for the company's growth and quality. A high-quality cyclical with mid-single-digit through-cycle revenue growth probably deserves a market multiple on mid-cycle EPS. A low-quality, no-growth cyclical deserves a discount.

Where this framework breaks

A few honest warnings:

• Secular decline masquerading as cyclical weakness. If volumes never recover to the prior peak, you don't have a cycle — you have a melting ice cube. Print advertising, coal, certain chemicals. Averaging the last ten years overstates the future.
• Structurally higher margins. Sometimes consolidation, mix shift, or pricing discipline genuinely lifts the through-cycle margin. Be willing to use a higher number than history if you can defend it.
• The trough is deeper than you think. Mid-cycle EPS is the average, not the floor. A stock trading at a low multiple to mid-cycle EPS can still fall significantly on the way to a trough year. Mid-cycle valuation tells you what to own through the cycle, not what the stock will do next quarter.

What to watch next

• Pull 7-10 years of operating margins for one cyclical you own and compute the average. Compare it to the consensus next-twelve-month margin. The gap is the market's implicit cycle call.
• For commodity-linked names, identify the single price assumption that drives 80%+ of earnings variance and write down your mid-cycle number for it.
• Check whether the company's capacity, mix, or cost base has structurally changed in your window. If yes, weight recent years more heavily or rebuild the margin from the bottom up.
• Apply a market multiple to your mid-cycle EPS. If the stock trades well below that, you have a candidate. If it trades above, you're paying for the next peak.