How to Use a Stock Screener: The Filters That Work and the Ones That Waste Your Time

Most people use a stock screener badly in the same way: they add filters until the result list feels impressive, then trade whatever's at the top. A good screen does the opposite — it removes almost everything, and what survives is a list you can genuinely work through before the open.

The target is 10 to 25 names. Fewer and you've over-fitted; more and you'll skim.

Layer 1: tradability (never skip this)

These filters aren't about opportunity. They're about whether you can get in and out without the spread eating your edge.

• Average daily volume — a hard floor. Below roughly 500k shares/day, your stop is a suggestion and your exit is a negotiation. Position size, not conviction, should set this number: if your typical position is a meaningful slice of a day's volume, you're the liquidity.
• Minimum price — sub-$2 names have structurally different behaviour (wider relative spreads, dilution risk, halt risk). Excluding them isn't snobbery; it's excluding a different game.
• Market cap floor — filters out shells and names where a single order sets the price.

Run these three and you've cut the universe by 80% without any opinion about the market.

Layer 2: the "something is happening" filter

This is where a screen earns its keep.

• Relative volume (RVOL) — today's volume vs the stock's own 20-day average, the most useful single field in any screener. RVOL > 2 means participation is unusual, which is the precondition for almost every intraday move worth taking. Absolute volume tells you if you can trade it; relative volume tells you if you should look.
• Gap percentage — for morning trading, combined with pre-market volume. Covered in depth in the pre-market screener guide.
• Distance from a key level — near 52-week highs, near the 200-day, breaking a multi-week range. Structure creates the decision point.
• ATR or ATR% — average true range tells you whether the stock moves enough to pay for the risk. A name with a 0.8% ATR will not deliver a 3% intraday target no matter how good the setup looks.

Layer 3: context (use sparingly)

Fundamentals belong in a screen when your holding period justifies them. For a day trade, a P/E ratio is decoration. For a multi-week swing, earnings date proximity is critical — being long into an unknown binary event isn't a strategy.

The one context filter almost everyone should use: exclude names with earnings in the next N days, where N matches your holding period. It prevents the most common avoidable loss in swing trading.

Filters that mostly waste time

• P/E for short-term trading. The names that move intraday are frequently the ones with no earnings at all.
• Dividend yield in a momentum screen. Different universe, different behaviour.
• Long moving-average crossovers as a primary filter. By the time a 50/200 crossover appears in a screen, thousands of systems have already traded it.
• "Analyst rating: buy." Nearly everything is rated buy.
• Stacking six technical indicators. They correlate heavily; you get the illusion of confirmation from what's essentially one signal counted six times. Two or three uncorrelated conditions beat six correlated ones.

Build your screen backwards

The reliable method: take your last 20–30 trades, keep only the ones that worked, and ask what those names had in common the morning you found them. That common set is your screen. Building forwards — picking filters that sound sensible — produces screens that match nothing you actually trade well.

Then, crucially: save it and stop editing it. A screen edited after every losing day is a mood ring. Give it 30 trading days before judging it, and change one variable at a time.

Where the AI layer changes things

A traditional screener answers a question you already knew how to ask. That's a real limit: you can only filter for patterns you've thought of. The AI layer's job is to work the other direction — scan everything continuously, surface the setups that qualify, and explain why each one qualified, including combinations you wouldn't have queried.

The honest caveat is that this only helps if the reasoning comes attached. A ranked list with no explanation is just someone else's screen, and you can't learn from it or reject it intelligently. Judge any AI screener on whether it tells you why — and on whether it publishes what happened afterwards.

A starter screen for intraday momentum

Average volume       > 1,000,000
Price                > $5
Relative volume      > 2
ATR%                 > 2%
Gap                  > 3% (morning) OR near 20-day high
Earnings             not today

That's six lines and it'll typically return a workable list. Adjust the numbers to your market and your position size — the structure matters more than my thresholds.

Bottom line

Filter for tradability first, unusual participation second, and context only if your holding period justifies it. Aim for a list you can actually read, build the screen from trades that worked, and leave it alone long enough to learn something from it.

Not financial advice. Screens find candidates, not trades.