Does a low PE ratio mean a stock is cheap?
We backtested our current scoring system across ten years of S&P 500 data. The short version: there's no single right way to read a stock — and a familiar rule of thumb is where it shows up first.
Short answer: there isn't one right way — and our own data shows why. Take the most familiar rule of thumb going: a low price-to-earnings (PE) ratio is supposed to mark a "cheap," good-value stock. When we backtested our current scoring system across the S&P 500 for a decade, that rule pointed the wrong way. The stocks our system rated as expensive outperformed the ones it rated as cheap. That isn't a verdict on PE ratios for all time — it's a record of what one specific decade did. And it points at something more useful than any single rule: how you read a stock depends on the market you're in and the kind of investor you are.
Here's what the data showed, and how we read it.
What we tested
Between 30 April 2016 and 28 February 2026 we ran the FIS score across the S&P 500. That came to 58,738 individual score calculations covering 648 different companies, taken from monthly snapshots — 119 of them. The score rates each company from 0 to 100 across nine fundamental checks: how profitable it is, how it's valued, how fast it's growing, how strong its balance sheet is, and what it pays in dividends.
Two things make the result trustworthy. First, at each point in history the test only counted companies that were actually in the index at that time — including ones that later went bust or got dropped. (That's why 648 companies show up, not 500.) Second, it only used information that was publicly available on each date, never figures that came out later. Those two choices stop a backtest from quietly flattering itself.
The question was simple: did a higher score line up with better returns afterwards?
What the headline numbers showed
Each month we sorted every scored company into ten equal groups — the top 10%, the next 10%, and so on down to the bottom 10% — and tracked how each group did next.
| Group | Average yearly return | What $1 grew to over the decade |
|---|---|---|
| Top 10% (highest scores) | 10.30% | $2.64 |
| The whole universe (equal-weighted S&P 500) | 9.35% | $2.43 |
| Bottom 10% (lowest scores) | 7.77% | $2.10 |
Over ten years, the gap between the top and bottom groups worked out to roughly 54% more wealth. On the face of it, a good result.
Two honest caveats belong right here, because they colour everything below.
First, the score didn't make things safer. The top group's worst drop from a high point to a low point (−29.0%) was actually a touch bigger than the market's (−27.7%). A higher score lined up with better returns, not a smoother ride.
Second, that comparison is against the equal-weighted S&P 500 — where every company counts the same. We use that as the fair, like-for-like yardstick, because our scoring method also treats every company equally, so the comparison shows whether the picking worked rather than mixing in a difference in how the index is built. The more familiar version of the S&P 500 is cap-weighted — bigger companies count for more — and over this exact decade that version returned about 14.4% a year, ahead of even our top group. The reason is that a handful of giant technology companies (the "Magnificent 7") drove an unusual share of the gains. Both things are true at once: our score sorted companies well within its own fair comparison, and over this particular decade a plain index fund tracking the bigger version would have finished ahead of our top group.
The "cheap" check that came back upside down
We also tested each of the nine checks on its own. Most behaved as you'd expect. One didn't.
The PE check — the one that rewards a low price-to-earnings ratio, the classic sign of a "cheap" stock — was backwards. The stocks that passed it (the "cheap" ones) underperformed the stocks that failed it (the "expensive" ones) by 1.24 percentage points a year, and by nearly 27 percentage points added up over the full ten years.
Put plainly: our system marked "expensive" stocks down, and "expensive" stocks won the decade.
Why? This was a stretch where the market kept rewarding pricier, faster-growing companies. A check built to prefer cheap-looking ones swam against that current for ten years. It's an important finding for how we evolve the system — but it's a record of what this decade did, not a law that a low PE is bad. In a decade where bargain stocks beat growth stocks, the same check could easily read the other way.
The same check can be right one year and wrong the next
The price-to-sales check makes the point even more clearly. Over the full ten years it was a useful positive signal (worth about +1.54 percentage points a year). But split it up by what kind of market we were in and it flips: clearly negative through the COVID crash and recovery, then clearly positive in the 2022 swing back toward value stocks. Same check, opposite results, depending only on which way the market was leaning at the time.
Against that, the checks that stayed steady no matter the market were the less glamorous ones — measures of quality and financial strength:
| Check | Edge per year (companies that passed vs. those that failed) | How it behaved |
|---|---|---|
| Current ratio (can it cover its short-term bills?) | +2.51 pp | Strong, steady in every market |
| Net margin (profit per dollar of sales) | +2.04 pp | Strong, steady |
| Revenue growth | +1.67 pp | Solid |
| Return on equity (profit vs. shareholder money) | +1.54 pp | Solid |
| Price-to-sales | +1.54 pp | Solid overall, but flips with the market |
| Price-to-earnings | −1.24 pp | Backwards over the period |
The pattern is hard to miss. The valuation checks (how a stock is priced) swung with the market's mood; the quality checks (how the business is actually doing) did the steady work in every kind of market.
It also depends on what you're trying to do
The market is only half the story. The dividend check shows the other half. Over the period, companies that paid a dividend returned 7.44% a year; companies that didn't returned 10.21%.
To a growth-focused investor, that makes the dividend check look like a drag on returns. To an income-focused investor, the lower total return is simply the accepted trade-off for the income that's the whole point of the strategy. Same numbers, opposite readings — and both are right, for different people.
That's the heart of it. A value investor, an income investor, and someone who'd rather just hold a broad ETF aren't looking at the same company in the same way, and ten years of data doesn't crown one of them the winner. The individual checks inside a single fixed score behave differently for different goals and different conditions — which is exactly why one fixed formula can't serve everyone in every market.
The toughest test — and our least flattering result
There's a harder way to test a ranking than the groups above. Imagine stripping out the market's overall rise entirely, so you can see whether the ranking itself adds anything on its own. It's done by betting on the highest-scoring stocks and against the lowest-scoring ones at the same time, in equal amounts — the market's up-and-down cancels out, and what's left is the pure signal of the ranking. It's the standard test for whether something is a real, dependable edge.
On that test, the signal was weak. It returned about 1.25% a year, won in only four of eleven years, and went through a five-year losing run from 2018 to 2022. One common way to score this is "how much return you got for the bumpiness endured" — and by that measure it scored 0.20, well below what professionals would call a genuine edge. The honest reading is that most of the good result in the groups above probably came from simply being in a rising market, not from a special, market-proof signal in the ranking.
We're telling you this because honest research shows the unflattering result next to the flattering one. It reframes what the score is for: historically it's worked as a structured way to judge quality and spot weakness when you're choosing between individual stocks — and it was strongest, in fact, at flagging the weakest companies — rather than a machine for beating the market.
So what does a decade of data tell you about reading a stock?
That no single number is the whole truth.
A low PE didn't mean cheap-and-good this decade. A pricing check that helped one year hurt the next. A dividend was a plus or a minus depending entirely on what you wanted from your money. Even the checks that held up only sorted clearly at the very top and very bottom, and blurred in the middle of the pack.
The conclusion we draw from our own research is the opposite of a hot tip. No single data point means much on its own; the numbers, the market conditions, and an investor's own goal only make sense together. That's the thinking behind the FIS approach — to put the clearest possible picture of a company, across every factor that matters and in plain language you don't need a finance degree to follow, in front of the person making the decision, and to leave the decision itself where it belongs.
A flat buy-or-sell call assumes there's one right answer for every reader. A decade of our own data says there isn't.
This piece describes historical backtesting research conducted by Financial Intelligence Service. Backtested results are hypothetical and do not represent the returns of any real portfolio. Past performance does not indicate future results. Nothing here is financial product advice or a recommendation to buy, sell, or hold any security; it does not account for any individual's circumstances or objectives.