Round Lot Jump to the manual 100 shares = 1 lot

Screening small-cap stocks: float, silent filters and liquidity

A screen behaves differently at the small end of the market, and the reasons are mechanical rather than philosophical. Fields go missing, float diverges from capitalization, and the price that looks like a bargain may be one nobody has transacted at recently.

Written under the standing byline R. Devlin screening and filings Published

Float vs cap
often far apart
Missing fields
common
Liquidity filter
not optional
A brass jeweller's sieve stack half-filled with tiny blank metal discs, the finest mesh on top
The finer the mesh, the more the sieve itself decides the result. At the small-cap end the tooling is part of the finding.

Free float is what is actually tradable

Market capitalization multiplies the share price by all shares outstanding. Free float counts only the shares genuinely available to trade, excluding founder holdings, strategic stakes and locked-up shares. On a large company the two are usually close. On a small one, a founder holding 70 percent means the float is a fraction of the headline figure.

This matters because most screens rank by capitalization, which is not what constrains you. A position that is trivial against capitalization can be substantial against float, and it is float that determines whether you can enter and leave at a comparable price.

Missing fields are the strongest filter in the screen

A loupe resting on its side on a shallow tray of tiny blank metal blanks under raking light
At this end of the market the tooling is part of the finding. What has no data does not appear at all.

Coverage thins as companies get smaller. Fewer analysts, fewer standardized metrics, more empty cells. When a screener filters on a field, rows with no value are almost always dropped rather than flagged, so a five-criterion screen has quietly added a sixth: "has data for all five".

At the small-cap end that unrequested criterion can remove a large share of the universe, and it removes it non-randomly. The companies with thinner coverage are exactly the ones a small-cap screen was supposed to surface. The fix is not clever criteria; it is checking the row count before and after each filter and asking where the difference went.

A test worth running onceApply your criteria one at a time and note the surviving count after each. If one criterion removes far more than expected, it is probably a data-coverage effect rather than a business finding. This takes two minutes and changes how much weight the output deserves.

Liquidity is a criterion, not a footnote

Average daily volume and the typical bid-ask spread belong in a small-cap screen as first-class criteria, because they determine whether any conclusion is actionable. A spread of several percent is a cost paid on entry and again on exit, and it can exceed the advantage the screen was built to find.

The practical rule is to size against volume rather than conviction. A position that would take several days of typical volume to exit is illiquid for you regardless of what the chart shows, and that arithmetic should happen before entry rather than during an attempt to leave.

Survivorship inside the universe itself

Backtests and long-run statistics at the small end are distorted by companies that no longer exist. Delisted, acquired and failed businesses drop out of most historical databases, so a screen tested on today's universe was tested only on the survivors. The effect is largest exactly where failure is most common.

This does not invalidate small-cap screening. It means a strategy that looked excellent in a backtest deserves skepticism proportional to how much of its universe would have gone missing, and that the more reliable use of a screen is generating a shortlist to read rather than a rule to follow.

FAQ

Questions this page raises

What counts as a small-cap company?

Conventionally a market capitalization roughly between 300 million and 2 billion dollars, but the boundaries are conventions rather than definitions and index providers disagree. For screening purposes the useful boundary is where data coverage and liquidity begin to bite, which is not a fixed number.

Why does my screen return so few results?

Most often missing data rather than strict criteria. Each filter drops rows with no value for that field, so several criteria compound into a coverage requirement. Run the criteria one at a time and watch where the count falls.

Should I screen by market cap or by float?

Float, if the field is available, because it reflects what is tradable. Where it is not available, treat capitalization as an upper bound and check the largest holders before assuming a position is easy to build or exit.

Do the Piotroski and Altman scores work on small caps?

They compute, and both were designed around conventional operating companies with two years of filings. Where a small company has incomplete history, negative earnings or an unusual balance sheet, the arithmetic still returns a number while the interpretation stops applying.

Next

Read on in screening