Data as of 18 September 2026. P/E and estimates from Finviz; the accounts from 10-K and 10-Q filings made to the SEC. Prices move in real time.
There is a screening rule that sounds reasonable: a stock is cheap if what you pay per dollar of earnings is less than the rate at which those earnings are growing. P/E below growth. If the P/E is 12 and earnings grow 20%, you are paying twelve for something advancing at twenty.
We ran it across the entire Russell 2000 — 1,944 companies — requiring it to hold for two consecutive years rather than one. Then we asked the question almost nobody asks: where is that growth supposed to come from?
The funnel
Of 1,944 companies, 825 lose money. No earnings, no P/E, so they leave the screen before it starts. Another 91 have no analyst estimates. That leaves 1,028 measurable.
| Screen result | Companies |
|---|---|
| Pass both years | 115 |
| Year 1 only | 282 |
| Year 2 only | 183 |
| Neither year | 448 |
| Lose money | 825 |
| No estimates | 91 |
115 out of 1,944. That is 5.9%. Up to here it is an ordinary screen and the output looks like a bargain list.
Then we opened the accounts of 45 of them — the ones with complete enough data to audit — and it stopped looking like a bargain list.
The trap: growing from almost nothing
Energy Services of America (ESOA) shows up with first-year earnings growth of 2,896%. Kulicke & Soffa (KLIC), 1,775%. Ironwood Pharmaceuticals (IRWD), 715%.
None of the three has discovered anything. Look at where they start from:
| Company | Last reported EPS | Year 1 estimate | Growth |
|---|---|---|---|
| ESOA · Energy Services of America | $0.0199 | $0.5963 | +2,896% |
| KLIC · Kulicke & Soffa | $0.2059 | $3.8613 | +1,775% |
| IRWD · Ironwood Pharmaceuticals | $0.1504 | $1.2264 | +715% |
| LQDA · Liquidia | $0.5260 | $3.0694 | +483.5% |
| HCC · Warrior Met Coal | $1.0942 | $6.3023 | +476.0% |
ESOA started from two cents a share. Any recovery toward normal produces a four-digit percentage, and a four-digit percentage automatically clears any P/E. The rule is not detecting a cheap company. It is detecting a company that barely earned anything last year.
Of the 45 audited, 17 are rebounds from a near-zero earnings base. That does not make them bad companies — Warrior Met Coal (HCC) and Kulicke & Soffa (KLIC) are real businesses — it means the screen approves them for a reason that is not the one the screen thinks it is measuring.
What survives once the rebound is stripped out
After auditing all 45 against their own filings, the split is this:
| Verdict | Total | Of which rebounds | Clean |
|---|---|---|---|
| Possible | 30 | 10 | 20 |
| Borderline | 1 | 0 | 1 |
| Does not work | 11 | 5 | 6 |
| Impossible | 2 | 2 | 0 |
| Not judgeable | 1 | 0 | 1 |
Twenty. Out of 1,944 companies, twenty clear the screen, survive an audit of their accounts, and do not owe their pass to a statistical rebound. That is 1% of the index, and it is a far more honest number than the 115 the screen produced.
How you check whether the growth is possible
The arithmetic is simple and deliberately generous. Take the year-2 earnings estimate, run the tax back out of it (divide by 0.79), adjust for the share that goes to minority holders, add back interest, and you get what the company would have to earn pretax. Divide that by its best-ever operating margin and you get what it would have to bill.
Then compare the required margin against its record. Below 95%, possible. Between 95 and 105%, borderline. Above, it does not work.
It is generous because it hands every company the best margin it has ever achieved, and because it projects revenue at its most recent quarterly pace. Even so, thirteen of the forty-five fall short.
The ones that fall short, ranked by how far
| Company | Margin required | Its record | vs. record | Bills today | Would have to bill |
|---|---|---|---|---|---|
| PHR · Phreesia | 10.0% | 2.5% | 406% | $508M | $2,393M |
| MUX · McEwen | 62.9% | 25.2% | 250% · impossible | $248M | $887M |
| SWX · Southwest Gas | 14.8% | 6.0% | 247% | $5,210M | $9,904M |
| FLYW · Flywire | 11.3% | 5.5% | 205% | $714M | $2,095M |
| CENX · Century Aluminum | 41.2% | 27.3% | 151% · impossible | $2,667M | $5,278M |
| ARQT · Arcutis | 26.1% | 17.3% | 151% | $462M | $1,401M |
Phreesia (PHR) is the extreme case. To earn the $0.64 a share estimated for year 2 it would need a 10% operating margin. Its best-ever margin is 2.5%. It bills $508M and would have to bill $2,393M — nearly five times more, while currently growing at 10.4% a year.
Southwest Gas (SWX) deserves its own line because it is neither small nor a rebound: it is a utility with $5,210M of revenue, and consensus is asking it for a margin two and a half times its record.
What had to be corrected by hand
Three adjustments that change the answer and that no automated screen makes:
Minority interests. When part of the profit does not belong to the listed company's shareholders, the earnings-per-share figure demands far more revenue than it appears to. We found ACMR at 28%, DBI at 30%, TPC at 30% and CVI at 60%. Excelerate Energy (EE) was the worst case and had to be read by hand in its June 10-Q: 76% of the profit belongs to the other share class.
The phantom margin record. Only years in which the company billed at least half what it bills today are allowed to count. Without that filter, Caleres (CAL) came out with a 168% record — a margin from when it was a different company.
The company with a 1.4 P/E. Sabre (SABR) showed up as by far the cheapest name among those audited. That P/E comes from having sold a business; today it loses money. Discarded.
What this screen cannot judge
It excludes 38 financials — among them ENVA, ATLC, MCB, BWB, SEZL and TBBK — because operating margin measures nothing at a bank. Another 10 have net income above operating income, a sign that what they earn is not coming from operating. And 15 have incomplete data in their filings. Seven file no US accounts at all: AVEX, HMH, PAGS, PLGO, SFL, STNE and TOYO.
What we would do
Not buy a list. This is not a portfolio, it is a funnel that turns 1,944 names into twenty that deserve to have someone actually open their accounts. The work begins where the table ends.
And if we had to keep one practical idea from all of it: before trusting a growth percentage, look at the number it grew from. A 2,896% and a 15% can say the same thing about a company's future, and one of the two is built on two cents.
In one sentence
A screen demanding that P/E be lower than growth passes 115 of the Russell 2000's 1,944 companies, but once you audit the filings only twenty are growing for real and can deliver what is estimated of them: the rest are rebounds from almost nothing, or margins the company has never once achieved.