Building a stock screener that fits your thesis

The first time you use a stock screener, one of two things happens. You set a couple of filters and get back 400 companies, which is useless — that’s not a shortlist, that’s a phone book. Or you crank every knob to “great” and get back three names, two of which are companies you’ve never heard of in countries you can’t pronounce.

Both failures come from the same mistake: treating the screener as a machine that finds good stocks, when what it actually does is find stocks matching a description. The quality of the output is entirely the quality of your description. So the real skill isn’t the tool. It’s writing a good description of what you’re hunting for — and that starts with a thesis, not a filter.

How do you build a stock screener around a thesis?

Write the thesis as one sentence, then translate each clause into a filter. Before you touch a single filter, finish this sentence: “I’m looking for companies that ___.”

• “…grow fast but aren’t wildly overpriced.”
• “…gush cash and pay me while I wait.”
• “…got beaten up recently but have healthy balance sheets.”

Each of those maps to a completely different set of filters. If you can’t finish the sentence, no screener on earth will help you, because you’re asking it to read your mind. Write the thesis first. The filters are just its translation.

Why is a single-filter screen a trap?

Because one factor alone, like a low P/E, surfaces value traps as readily as bargains. The classic beginner screen is a single filter: “low P/E.” The problem is that a low P/E, by itself, is at least as likely to surface a dying business as a bargain. Cheap is often cheap for a reason. This is the “value trap,” and single-factor screens walk straight into it.

Good screens layer factors so they check each other:

• Valuation (P/E, P/S, EV/EBITDA, free-cash-flow yield) — is it reasonably priced?
• Growth (revenue growth, earnings growth) — is the business actually going somewhere?
• Quality (margins, return on equity, debt levels) — is it a sturdy company or a house of cards?
• Technicals (price vs. moving averages, distance from highs) — is the timing sane, or am I catching a falling knife?

“Low P/E” alone finds value traps. “Low P/E and growing revenue and manageable debt” finds something closer to an actual opportunity. Each filter you add isn’t just narrowing the list — it’s disqualifying a specific way the first filter could have fooled you.

How many stocks should a screen return?

A list you can actually research — roughly 10 to 30 names. A screener’s job is to hand you a list a human can work through — call it 10 to 30 names — not to pick the winner. If you’re getting hundreds, you haven’t described anything specific. If you’re getting two, you’ve over-specified and you’re now curve-fitting to a fantasy.

When the list is too long, add a quality gate, not a stricter valuation one — it thins the herd without pushing you toward the cheapest, most fragile names. When it’s too short, loosen your single strictest filter by a notch and watch what floods in; that tells you which constraint was really doing the work.

Is a screener result a buy list?

No, it is a reading list. Here’s the uncomfortable truth. The screen doesn’t find you a stock to buy. It finds you a stock to research. Everything a screener knows is a number in a database — it has no idea that the “growing” company just lost its biggest customer, or that the “cheap” one is cheap because a lawsuit is coming.

So the output of a good screen isn’t a buy list. It’s a reading list. Ten names, and now you go do the actual work: read the filings, understand how the company makes money, decide whether today’s price makes sense for you. The screener saved you from reading about the thousands of others. That’s the entire value, and it’s a lot — but it’s not the same as an answer.

What does a thesis-driven screen look like?

Here is one worked example. Say your thesis is “quality compounders that got cheap in a selloff.” That translates to roughly:

• Return on equity above 15% (quality)
• Revenue growth positive over three years (it’s still growing)
• Debt-to-equity under 1 (won’t get wiped out by rates)
• Price more than 20% below its 52-week high (the “got cheap” part)

Run that and you’ll likely get a workable handful — good businesses the market has temporarily soured on. Some will deserve the pessimism. A few might not. That’s the list worth a Saturday.

---

Tradune’s screener runs across price, valuation, growth, quality, and technicals so you can build multi-factor screens from a thesis instead of a single slider. What it hands you is a starting point for research — never a recommendation.

Related feature: Tradune’s stock screener — compose the multi-factor screen this post describes, then save and re-run it.