The S&P 500 trades around a 20x forward P/E as of late summer 2026, but that single number hides a huge amount of variation underneath it. Information technology sits near 19x forward earnings and closer to 28x on a trailing basis, while utilities trade around 17x forward and 18x trailing — a gap of roughly ten points between two sectors in the same index. Screen the whole market with one P/E cutoff and you’ll either wipe out every tech stock or let in every overpriced utility, and neither outcome is what you actually want.
The same ratio, two different businesses
A P/E ratio is a price paid for a dollar of earnings, but what that dollar is worth depends entirely on how fast it’s expected to grow and how reliably it shows up. Technology companies at roughly 28x trailing earnings are being priced for double-digit annual growth — the market is willing to pay more today because it expects meaningfully more tomorrow. Utilities at around 18x are priced for the opposite: slow, regulated, highly predictable growth in the low single digits, tied to rate cases and infrastructure spending rather than product cycles.
Neither multiple is objectively “right” or “wrong.” A 28x tech stock can be cheap relative to its own growth rate, and an 18x utility can be expensive relative to its own. The comparison that matters isn’t tech versus utilities — it’s a stock against the peers it actually competes with for capital.
Why a flat threshold backfires
Set a screener filter for “P/E under 20” and you’ll get a list that’s heavily tilted toward financials, energy, and utilities — not because those are the market’s best opportunities right now, but because those sectors simply carry lower multiples structurally. Meanwhile, a well-run software or semiconductor company trading at a discount to its own sector average never makes the list at all, even though it may be the better relative bargain. The reverse problem shows up with growth-oriented filters too: set the bar too high to capture reasonably priced tech names, and you’ll also wave through richly priced staples or utilities that have no business trading at that multiple. A single absolute threshold, applied market-wide, optimizes for sector composition rather than valuation quality.
Building a sector-relative screen
The fix isn’t complicated, just an extra step. Rather than screening the entire market on one P/E cutoff, run the screen sector by sector and compare each result to that sector’s own typical range. Filter by sector first, isolating one group — technology, healthcare, industrials — so you’re comparing like against like before any valuation filter is applied. Then sort by P/E within that sector: the cheapest quartile of a sector is a far more useful signal than an arbitrary market-wide number, since it accounts for that sector’s normal growth and margin profile.
Pairing valuation with quality strengthens the screen further — a below-average sector P/E combined with above-average ROE or margins filters out stocks that are cheap because the business is deteriorating, not because it’s mispriced. And the PEG ratio, which divides P/E by expected growth, works as a useful cross-check since it partly normalizes for the growth differences between sectors and can catch cases where a “high” P/E is actually well justified.