P/E 16 — Cheap or Expensive? You Can't Know Without History

Open any financial news site and you'll find P/E ratios quoted as if they mean something in isolation. "Apple trades at a P/E of 28." Is that expensive? Cheap? The article usually doesn't say. Neither does the number itself.

A P/E ratio only becomes meaningful when compared to two things: (1) the same company's own history, and (2) the growth rate embedded in it. This article is about the first. The PEG ratio — which handles the second — is covered in the GARP track.

The same number, two very different situations

Same number. Opposite situations. This is why every valuation metric needs its own history behind it.

How percentile ranks work

A percentile rank answers a simple question: of all the monthly P/E readings for this stock over the last N years, what fraction were lower than today's reading? A rank of 10% means the stock has been this cheap or cheaper in only 10% of months — it's near a historically low valuation. A rank of 90% means it's near a historically high valuation.

The direction of the metric matters too. For P/E, P/B, and EV/EBITDA, a low percentile is typically favourable — it means you're paying less than usual. For dividend yield, a high percentile is favourable — it means the yield is higher than usual relative to the stock's own history, which often means the price is low.

A real example: Johnson & Johnson

Between 2014 and 2024, Johnson & Johnson's trailing P/E ranged from roughly 13× to 28×. A reading of 16× sits in approximately the 15th percentile of that range — meaning JNJ has only been this cheap, relative to its own earnings history, about 15% of the time over the decade. For a Dividend King with 62 consecutive years of dividend growth, a P/E in the bottom quintile of its own history is worth paying attention to.

What to watch out for

• Earnings can be manipulated. Always check whether EPS growth is real or driven by share buybacks and accounting adjustments.
• Cyclical businesses (energy, materials, financials) can look "cheap" at peak earnings. The E in P/E is at its highest exactly when the business is most cyclical — Graham called this the "earnings trap".
• A persistently falling P/E might signal a structurally declining business, not a buying opportunity. The percentile rank doesn't distinguish between a temporarily depressed stock and one in secular decline.
• Historical data only goes back so far. If a company went through a major transformation 5 years ago, its pre-transformation P/E history may not be relevant.

Graham's approach: normalised earnings

Graham used "earnings power" — an estimate of sustainable mid-cycle earnings — rather than trailing EPS, precisely to avoid the cyclicality trap. For practical purposes, using a 7–10 year history and looking at the percentile of the current P/E against that history achieves a similar result: you automatically capture several business cycles and get a more stable signal.