What they are
Two of the simplest ratios in fundamental analysis:
P/E = Stock Price ÷ Earnings per Share (or Market Cap ÷ Net Income) P/S = Stock Price ÷ Sales per Share (or Market Cap ÷ Revenue)
P/E tells you how many years of current earnings the market is pricing in. A P/E of 20 means investors are paying $20 today for $1 of trailing earnings. P/S does the same against revenue.
Both come in "trailing" (TTM = trailing twelve months) and "forward" (analyst-estimated next-12-months) flavors. We use trailing in our grade card because forward is sell-side consensus, which has its own biases, and we'd rather anchor to what actually happened.
When to use P/E
P/E is the right ratio when net income is positive and not heavily distorted. That covers maybe 70% of the public equity universe. Mature businesses with stable earnings — consumer staples, industrials, large-cap tech, banks, established healthcare — all live in P/E land.
Rough P/E bands:
- Under 10: Either deep value or a value trap. Look at the trend and the balance sheet. Cyclicals near a cycle top routinely print P/Es of 5–8 right before earnings collapse.
- 10–15: Statistically cheap by historical standards. Most "value" picks live here.
- 15–22: Market-average territory. The S&P 500's long-run average P/E is roughly 17.
- 22–30: Premium quality. Justified if growth is real and durable.
- Above 30: Either hypergrowth that the market is pricing forward (NVIDIA in 2024), or a quality compounder commanding a premium (Costco). Worth a second look — pay this multiple only when the business is genuinely exceptional.
When to use P/S
P/S is the right ratio when net income is negative or barely positive. A company can't have a meaningful P/E if it doesn't make money — the ratio either prints as negative (which gets dropped from screeners) or infinitely high (when earnings round to zero).
P/S bands are wildly sector-dependent:
- Software / SaaS: 5–15 is normal during expansion phases. The high gross margins (80%+) mean revenue eventually becomes operating leverage.
- Mature tech, semiconductors: 3–6 typical.
- Consumer brands, retail: 1–3 is the band. P/S of 8 on a clothing retailer is a warning, not a signal of quality.
- Banks, insurers: P/S is largely meaningless — their "revenue" is net interest income, which isn't comparable to a software company's revenue. Use P/B (price-to-book) or P/E instead.
How the Bull Rankings model picks one
For its grade card, every name is sorted into growth, value, or speculative. A company lands in the speculative group when it is burning cash or has no positive P/E — biotech, pre-revenue tech, distressed names with optionality. For speculative names the card grades P/S. For growth and value names it grades P/E.
This is exposed in the metric card: a growth or value pick shows P/E, a speculative pick shows P/S. Each is graded against the stock's broad sector when at least 20 peers report the same multiple, and against fixed bands otherwise. On those fixed bands a P/S of 5 grades B, while a P/E of 5 grades A. The grade card is separate from the quality-growth score, which doesn't use this split.
Common gotchas
- Trough earnings. A cyclical at the bottom of its cycle has artificially low earnings. P/E looks elevated even though the stock might be cheap on normalized earnings. Steel mills, refiners, semiconductor capital-equipment makers all do this. P/S smooths it out — sales don't collapse as much as profits do at a cycle low.
- One-time charges. A massive impairment, lawsuit settlement, or restructuring write-off can take TTM earnings negative even for a fundamentally profitable business. P/E briefly reads as n/m. Look at three years of earnings, not one quarter, before drawing conclusions.
- Growth that justifies the multiple. A 50× P/E sounds insane until you realize the company is growing 40%+ and will compress that multiple to 20× organically in two years. This is what PEG (covered separately) is designed to capture.
- Banks and insurers. Don't use P/S on financials. Use P/B (price-to-book) and ROE together — that's the canonical valuation pair for the sector.
- Stock-based comp adjustments. Some screeners use GAAP earnings; others use "adjusted" or "non-GAAP." A growth-stage software company can have a GAAP P/E of 200 and a non-GAAP P/E of 30. We use GAAP because it's the legally-required, comparable number across every company — but we read it knowing this discrepancy exists.
Bottom line
The choice between P/E and P/S isn't artistic — it's structural. P/E breaks down when earnings round to zero or go negative; P/S is the fallback. Once you've picked the right ratio, the band you compare to is what makes the answer useful. A 50× multiple on a hypergrowth software name is reasonable; a 50× multiple on a regional bank is hallucinatory. Sector-aware bands aren't optional.
How we use both
Neither ratio is used against an absolute threshold in the score. P/E is scored as a percentile against the stock's broad sector, and the model's preference order is explicit.
A growth-adjusted P/E comes first — weight 1.0 — because a growth-adjusted price is the more informative read (in the score it's the trailing P/E divided by revenue growth; see the PEG guide). Sector P/E is the fallback at the same weight, used only when there is no growth to adjust by. P/S isn't one of the value pillar's reads. It stands in for P/E on the grade card instead. A company reporting a net loss takes the most expensive growth-adjusted read, and its other P/E-based reads drop out, so the value pillar can't rest on free cash flow yield alone.
That ordering encodes the view that a raw multiple can't distinguish 30× for a business growing 30% from 30× for one growing 5%, and that most of the time you'd rather know which of those you're looking at.
A caution about the P/E you see
Something worth knowing if you compare our figures with others: the displayed P/E is the vendor's trailing P/E, matching what you'd see on Google or Yahoo — deliberately so, because a site quietly showing different reference numbers than everywhere else invites the suspicion that it's massaging figures.
But our own audit checks that vendor P/E against market cap divided by our filing-sourced net income, and roughly 90 names disagree by more than 60% on any given night. When those cases were pulled and checked against the actual filings, the vendor was frequently the one that was wrong — one company's vendor P/E implied a net income that nothing in its 10-K came within five times of.
We publish the vendor figure anyway, and say so. Reference metrics matching the rest of the internet is worth more than being right about a number most readers skim — but the score doesn't take it on faith. When a company's filings show a loss behind a positive P/E, the score ignores that multiple and the PEG that goes with it, the same way the grade card does. If a P/E on this site looks strange against the fundamentals beside it, that gap is real and it's usually the multiple that's wrong.
When P/S is the only option
P/S is what remains when a company has no earnings, which makes it the natural ratio for young or loss-making businesses — and also the easiest to abuse, because every company has revenue and so P/S always produces a number.
The model carries an explicit sanity limit: a price-to-sales ratio beyond roughly 250× is treated as a broken record rather than an expensive stock. One company in the universe carried $325,000 of revenue against a $2.45bn market cap — a P/S of over 7,500×. The richest genuinely-valued name in the same book sat at 67×, and so did the 99th percentile. Nothing legitimate lives anywhere near that line, which is exactly why it can be drawn.
That check is set roughly four times above the richest real name deliberately, so it never becomes a backdoor valuation screen. It exists to catch data that isn't describing a business, not to express an opinion about expensive stocks.