How to read a stock screener without fooling yourself

What a screener is actually doing

A screener runs a query against a vendor's database of end-of-day prices and parsed financial statements. Three consequences follow, and they explain nearly every screening mistake:

• It only knows fields the vendor collects. Warrants outstanding, the existence of an at-the-market programme, and a convertible note's conversion formula are not screener fields. They are frequently the most important facts about a micro-cap.
• It reflects the vendor's parsing. Two screeners will disagree about the same company's share count because they pulled from different filings on different dates.
• It has no opinion. Results are a list of rows matching criteria. Any significance is supplied entirely by you.

Used well, a screener turns 6,000 tickers into 30 you can actually read about. That's all it does, and it's worth a lot — as long as you remember that the 30 are candidates for research, not conclusions.

Price, market cap and enterprise value

Price is the last close, and it means less than people assume. A $2 stock is not cheap and a $200 stock is not expensive; price is an artefact of how many shares exist. The only things price genuinely determines are eligibility rules — exchange listing minimums, some brokers' restrictions, margin eligibility — and, informally, which crowd trades it.

Market capitalisation is price × shares outstanding. In small caps two failure modes recur: the share count is from the last filing and may be badly out of date after an offering, and the figure ignores debt entirely.

Enterprise value — market cap + debt − cash — is usually the more honest size measure. A company with a $40M market cap, $30M of cash and no debt has an enterprise value of $10M: the market is valuing the operating business at a quarter of the headline. The reverse is more common and more dangerous: a $40M market cap with $35M of debt is a $75M enterprise, and the equity is a thin slice on top of a large obligation.

Volume and liquidity: the field most people get wrong

Nearly every screener offers "average volume," in shares. It is the wrong unit.

Consider two stocks each showing 2,000,000 average daily shares:

Same volume field, two entirely different instruments. Always filter on dollar volume — price × average volume. If your screener doesn't offer it directly, apply a minimum price alongside a minimum share volume, which approximates the same thing.

Two refinements worth making:

• Use a longer average. A 10-day average volume is badly distorted by a single spike day. A stock that traded 40 million shares once and 60,000 shares on every other day will show a 10-day average around 4 million. The 50-day or 90-day figure is closer to what you can actually rely on.
• Consider median rather than mean if your tool offers it. Small-cap volume distributions are so skewed that the mean describes a day that never happens.

Float, shares outstanding and short interest

These three fields carry the most staleness risk of anything on the screen.

Shares outstanding comes from the most recent filing the vendor parsed. A company that closed a 20-million-share offering three weeks ago may still show its pre-offering count — which means the market cap on your screen is wrong too.

Float is a vendor estimate, not a reported number, and vendors classify affiliate holdings differently. Disagreements of 20–30% between two sources on the same micro-cap are routine. Never build a thesis on a single float reading.

Short interest is reported twice monthly and published on a lag. The number on your screen describes a position from one to three weeks ago.

Our guide to float and short interest goes through the reporting mechanics in detail. For screening purposes the rule is simple: use these fields to sort and narrow, never to conclude.

Fundamental fields and how stale they are

Every fundamental field — revenue, earnings, cash, debt, book value — describes a period that has already ended, sometimes long ago.

The timeline for a domestic small-cap filer: a quarter ends, the 10-Q is due within 40 or 45 days depending on filer status, the vendor parses it over the following days, and your screener updates. Best case you are looking at data 45 to 60 days old. Worst case is considerably worse, because small caps file late — a Form 12b-25 (commonly called an NT 10-Q or NT 10-K) buys a short extension, and some companies become delinquent for months.

For a company burning cash, this matters enormously. A balance sheet showing $8 million of cash as of a quarter-end four months ago tells you very little about solvency today if the burn rate is $2 million a quarter. The cash field is a historical fact, not a current one.

Ratios that break on small caps

The general principle: a ratio with a negative denominator doesn't tell you a company is bad value. It tells you the ratio doesn't apply. Sorting ascending on such a column puts the most distressed companies at the top of your "cheapest" list.

Exchange and tier filters

Where a stock trades tells you which rules it lives under, and this is one of the higher-value filters available.

Nasdaq and NYSE (including NYSE American) listings must satisfy continued-listing standards — minimum bid price, stockholders' equity or market value tests, corporate-governance requirements including an audit committee, and timely filing of periodic reports. That's a real, if low, floor.

OTC markets are tiered, and the tiers differ enormously:

• OTCQX — the highest tier, with financial standards and a prohibition on shell companies
• OTCQB — venture stage; requires current reporting, a minimum bid price, and an annual verification
• Pink — subdivided by how much information the company provides, from current to limited to no information at all
• Expert Market — quotes are not publicly displayed and most retail brokers permit only unsolicited orders; effectively very hard to trade

If your screener offers an exchange filter, use it deliberately rather than leaving it at "all." A screen run across everything will include companies with no current financial information, where the fundamental fields you're filtering on are guesses or blank.

Technical fields and the split problem

Moving averages, 52-week ranges and percent-off-high are computed from price history — which raises the question of whether that history was adjusted for splits and dividends.

Reputable vendors adjust. But the adjustment creates its own confusion in small caps: after a 1-for-20 reverse split, the historical chart shows a stock that traded at $60 two years ago, when in fact it traded at $3.00. "Down 96% from its high" is arithmetically true and narratively misleading — that high was a different, pre-split security.

Similarly, screening for "near 52-week low" in a universe of serial reverse-splitters mostly returns companies in structural decline, which is either exactly what you want or exactly what you should avoid, depending on your strategy. Know which.

Biases hiding in your results

• Survivorship. Most screeners query currently-listed securities. Companies that were delisted, acquired, or went to zero have left the universe. Any intuition you form about "how these usually work out" is built on the survivors.
• Look-ahead, when backtesting. If a screener applies today's restated financials to a historical date, results reflect information nobody had at the time. Point-in-time data is a premium feature for a reason.
• Crowding. Popular default screens on popular platforms are being run by thousands of people simultaneously. In a stock that trades $180,000 a day, that matters — the screen itself can be the catalyst.
• Multiple comparisons. Test forty filter combinations and some will look excellent purely by chance. The fix is to form the hypothesis first and test it once.

Building a screen worth running

A workable structure is three layers: make the universe tradable, then apply your actual thesis, then verify by hand.

Layer 1 — tradability (identical on every screen you run)

• Average daily dollar volume above a floor you set from your own position sizes — if you take $10,000 positions, $500,000/day is a reasonable minimum
• Price above $1.00, to exclude names facing listing-compliance problems (drop this deliberately if sub-$1 is your strategy, not by accident)
• Exchange: Nasdaq, NYSE, NYSE American — or explicitly include OTCQX/OTCQB if that's your universe
• Market cap within a band you actually understand

Layer 2 — the thesis (this is the part that varies)

Whatever you're actually looking for: revenue growth above a threshold, positive operating cash flow, insider buying, a particular technical condition. Keep it to two or three conditions. Every additional filter shrinks the result set faster than it improves it, and a screen returning four names is usually over-fitted rather than selective.

Layer 3 — manual verification (never skip)

For each survivor, before it earns any more of your time:

1. Open the latest 10-Q. Check the cover-page share count against your screener's figure.
2. Check cash and quarterly burn. Compute runway.
3. Scan the last 12 months of filings for S-1/S-3 registrations, 424B supplements, 8-Ks announcing offerings, and reverse splits.
4. Count warrants and convertibles outstanding and note their strike or conversion terms.
5. Confirm the company is current in its filings.

If a name fails layer 3, it doesn't matter how good it looked in layer 2. Our guide to reading SEC filings covers exactly where each of these lives.

Frequently asked questions

Which screener should I use?

For most people, the free screener bundled with their broker is sufficient for layers 1 and 2, because the accuracy that matters is verified by hand in layer 3 anyway. Paid tools are worth it when you need point-in-time data for backtesting, real-time intraday scanning, or fields like short borrow that free tools don't carry.

Why do two screeners give different results for identical criteria?

Different source filings, different parsing dates, different definitions of derived fields like float and EBITDA, and different universes. This is expected. If a name matters to you, verify it against the filing rather than picking whichever vendor you prefer.

How many results should a good screen return?

Enough to work through in a sitting — typically 15 to 50. Hundreds means your filters aren't doing anything. Under five usually means you've fitted the screen to a handful of names you already had in mind.

Can I screen for catalysts?

Only indirectly. Screeners hold periodic financial data, not upcoming events. Earnings dates are sometimes available; regulatory decision dates, trial readouts, and contract awards generally are not. Those come from filings, company calendars and news sources.