Premarket stock scanners: what to look for

What a scanner does

A premarket scanner is a loop. Every few seconds, or every few minutes on slower tools, it takes a universe of stocks, pulls fresh data for each, calculates the measures you care about, discards anything that fails your filters and shows what is left, usually sorted by percentage move. A movers list ranks; a scanner filters. The difference matters because a plain ranking favours small, thin stocks, as the gainers and losers guide shows, while filters let you decide what "tradable" means before you look.

A scanner's output is only as good as three things: the universe it covers, the filters you set and the data feeding both.

Filters that matter

Three of these deserve more than a line.

Dollar volume beats share volume. For example, 50,000 shares of a $2 stock is $100,000 of trading, while 50,000 shares of a $60 stock is $3 million. A share-count filter treats them the same.

RVOL depends on how it's measured. Relative volume compares today's volume with normal volume. A simple version divides today's volume by the average daily volume, so at 07:00 nearly all stocks look quiet, because the day has barely started. A time-of-day version compares with the volume usually traded by the same time, which is more informative early but needs more history. Know which one your scanner uses before you pick a threshold; relative volume in the premarket goes into the details.

Spread is a cost you pay on the way in and out. Spread as a percentage of the midpoint is a direct estimate of the round-trip cost. For example, a quote of $24.80 bid and $25.10 ask has a midpoint of $24.95 and a spread of about 1.2%. A momentum trade aiming for a few percent gives up a large part of it to a spread that wide. See spread and premarket liquidity and spreads.

Data freshness and quality

A filter applied to bad data still produces a confident-looking result. Four questions to ask about any scanner's feed:

1. Real-time or delayed? Free data is often delayed by 15 minutes. In the premarket, 15 minutes can be the entire move; a delayed scanner is fine for study and unusable for entries.
2. Consolidated or single-venue? A single-exchange feed sees only part of the trading and may start later. For example, Alpaca's free IEX feed shows premarket data only from 08:00 ET.
3. How old is each quote? A quote can come from a real-time feed and still be stale if nothing has updated for a while. Look for an exchange timestamp and its age in seconds.
4. Is the quote sane? A crossed quote, with the bid above the ask, signals a data problem or a very fast market. A trade printed far outside the bid and ask is usually an error or a late report, and shouldn't move the scanner's numbers.

Alerts versus automation

A scanner ends in either an alert or an action. An alert tells you something passed your filters; you then check the chart, the news and the quote, decide, and place an order, by which time the price may be elsewhere. What you gain is judgment: you can pass on the stock with a dubious catalyst. An automated setup acts on the rules immediately and identically every time. That removes hesitation, but it also removes the human check, so the data has to be gated far more strictly: a delayed or crossed quote that a person would shrug off can become an order. Automated momentum trading covers that trade-off in full.

How Momentum Lab gates data and candidates

PatternX's Momentum Lab acts automatically, so its scanning stage is strict. Its data-quality gate rejects:

• a symbol with no quote;
• a feed that reports itself as delayed;
• data without an exchange timestamp;
• data older than 10 seconds;
• a crossed quote.

A bad-print guard also checks the last trade: if it sits more than 3% outside the bid and ask, Momentum Lab uses the midpoint instead. Stale data never arms, confirms or triggers a strategy.

Alongside the gate sit the liquidity filters. Defaults: minimum price $1; minimum premarket volume 50,000 shares (100,000 for the Reversal strategy); minimum premarket dollar volume $250,000; maximum spread 1.0% of the midpoint, which a stock without a two-sided quote fails; all exchanges; halt detection on; volatility-pause detection on in the regular session. An RVOL filter exists but is off by default; it divides today's volume by average daily volume, isn't adjusted for time of day, and is only available with IBKR data. A minimum market cap is mandatory: $100M, $500M (the default) or $2B. Any enabled minimum fails when its input is unknown, so missing data never counts as a pass.

For example, a stock with 60,000 premarket shares at $3.50 passes the share-volume minimum but has only $210,000 of dollar volume, so it fails the $250,000 default. The $24.80 by $25.10 quote above, at about 1.2%, fails the 1.0% spread maximum.

What your data source supplies changes which checks can run:

With IBKR data, a stock IBKR has no figure for falls back to one derived from SEC EDGAR filings, then to Polygon where PatternX has it configured; with Alpaca data, Polygon is the only source. With no figure at all, the stock is skipped. The guide lists every parameter, and the market-cap filter article explains the presets.

What to be careful of

• Loosening filters until something shows up. An empty scanner on a quiet morning is information, not a fault.
• Trusting early-morning RVOL. A version that isn't time-adjusted is structurally low before the open.
• Assuming full coverage. Know your feed's start time and whether the scanner covers the whole market or only the top of a movers list.
• Ignoring timestamps. A move on a 20-minute-old print isn't a current move.
• Treating "passed the scanner" as "worth trading". Filters remove bad candidates; they don't make the remaining ones good.