How to Use a Stock Screener to Build a Short List

Open a screener, leave every filter at its default, and the results count at the top covers the whole database. Thousands of tickers. That count is the first thing to shrink, and the order in which you shrink it decides more about the final list than most people realize, because every filter you add works only on whatever the filters before it left behind.

Start with a question

A screen needs a question behind it. “Profitable small caps that have been paying down debt” is a question. “Good stocks” is a mood.

Write the question as one sentence before you open the filter panel. It tells you which fields you need. It also gives you a test for the results later, since a name either fits the question or it doesn’t. Skip this step and the usual pattern takes over: you add filters one at a time, watch the count fall, and stop when the number feels comfortable, which leaves you with a list shaped by the order of the dropdown menus more than by any idea about the businesses on it.

Keep it narrow. Broad questions produce broad lists.

Set the universe, then liquidity

The first filters decide where you’re allowed to look: which exchanges, which countries, and how big a company has to be. They’re blunt tools, and that’s fine.

Liquidity comes next. An average daily volume floor and a minimum share price clear out the names where a quote can jump on a small order, where the bid and ask sit far apart, and where your own buying could move the price against you before the order is even filled. Setting those floors early stops the thesis filters from packing your list with stocks you could never trade at a fair price. The argument for doing it in this order is made in set liquidity minimums first.

What counts as liquid depends on your order size. Someone buying a few hundred shares can live with thinner trading than a fund can. Pick floors that fit your own orders.

Add the thesis filters

Now the filters that carry the idea. Keep them few.

Two or three well-chosen conditions usually say more than ten loose ones. Each extra filter is one more place where a data quirk can drop a good company or wave a bad one through. For every field you use, read its definition. Most screeners keep a glossary that says whether the P/E ratio uses trailing or forward earnings, whether growth is annual or quarterly, and what happens when a value is missing.

Missing values deserve their own check. Some tools drop a company when a field is blank. Others let it through. Either way the results change and nothing on the screen tells you.

Trim to a list you can research

Aim for 10 to 30 names. Fewer than ten usually means one filter is doing all the work. More than thirty and you won’t read the filings, which hands the decision to the screen.

If the count runs high, tighten the filter closest to your question. If it runs low, loosen the one furthest from it. Change one thing at a time.

Sorting helps. Sort by the field that matters most to the question and read down the list, paying extra attention to the names at the extremes, because an extreme value is about as likely to come from a data error, a tiny denominator or a one-off accounting event as from anything real about the business.

Save, export and check the data date

Save the screen under a name that states the question. Export the results with the date you ran them.

Then find the data date. Screeners refresh fundamentals on their own schedules, so a company that reported last week may still show the prior quarter’s numbers, and a list built on stale figures can include companies that no longer pass while missing ones that now do. Look for an “as of” note on the results table or in the field glossary. A tool that never shows one has a gap, and the stock screener buyer’s guide treats that as a reason to look elsewhere.

Rerun saved screens on a schedule. Results drift as prices move and new filings land. A screen that fit one market can also go quiet in the next, which is the subject of screens can stop working.

Read the filings

The screener’s job ends here. Yours starts.

Open each company’s latest annual or quarterly report and check that the numbers the screen relied on are real, recurring and consistent with the business the company describes. A one-time gain can make a company look profitable for a year. An acquisition can make growth look homegrown when most of it was bought. Neither shows up in a screener column. Reading the 10-K in a set order keeps this to about an hour per company.

Cut hard. Most of a list can fall away once you read the filings. The names that survive go on a watchlist with one line each on why they passed.

What goes wrong

• Filtering before the question exists, then explaining whatever comes out.
• Stacking so many conditions that the list reflects gaps in the data more than anything about the companies.
• Trusting a field without reading how the tool calculates it.
• Forgetting the data date.
• Treating the final list as a buy list, when it is only a reading list that happens to be sorted by the numbers you chose.