Stock screener tools all agree on the easy part. Set a return on equity floor and a debt ceiling, and any of them will hand back the companies that clear both. What the result list does not carry is the fact that one of those two rules did nearly all of the eliminating while the other barely changed the outcome. Without it you cannot tell which threshold you are actually betting on, and you end up adjusting the rule that was never the constraint.
This article publishes that breakdown for eight value screens. Every figure below is recomputed from the GuruScreens run effective 2026-08-01, which evaluated 3,787 US companies against SEC EDGAR filings, with Fiscal.ai as a data partner. Data as of that run. The screens rebuild monthly, so the counts move.
The eight screens, and how much each one cuts
| Screen | Rules | Companies passing | Share of 3,787 |
|---|---|---|---|
| Enterprising Investor (Graham) | 5 | 102 | 2.69% |
| Acquirers Multiple (Carlisle) | 2 | 75 | 1.98% |
| Graham Net-Net (NCAV) | 1 | 43 | 1.14% |
| Buffett Quality | 5 | 16 | 0.42% |
| Neff Value | 4 | 10 | 0.26% |
| Magic Formula (Greenblatt) | 2 | 8 | 0.21% |
| Quality at Reasonable Price | 6 | 5 | 0.13% |
| Defensive Investor (Graham) | 5 | 3 | 0.08% |
Two rows deserve a second look. Graham Net-Net applies a single rule and still cuts the universe to 43 names. Quality at Reasonable Price applies six rules to reach five. Rule count and selectivity are almost unrelated, which is the first sign that counting your filters tells you very little about what your screen is doing.
What a stock screener throws away
Most screening tools answer one question: which companies satisfy all of my conditions at once. That is a single true or false per company, and everything used to produce it can be discarded once the list exists.
The discarded detail is where the useful information lives:
- Which rule failed. A company that fails only your debt test is a different research prospect from one that fails four tests.
- By how much it failed. A current ratio of 1.98 against a 2.0 threshold is noise. A current ratio of 0.4 is a finding.
- What the actual value was. Without the underlying number you cannot tell whether your threshold is doing real work or just reflecting how you rounded it.
- What the screen would look like without a given rule. This is the question that identifies the threshold you are really betting on.
None of that is exotic. It is computed while the screen runs, and a plain result list has nowhere to put it.
The audit trail: threshold against actual value
On GuruScreens, each of the 3,787 evaluated companies keeps its full evaluation. For every rule you get the metric, the comparison, the threshold, the company's actual value wherever the filing supplies one, and the resulting pass or fail. A company is not merely absent from the Buffett Quality list. It is absent for a stated reason that carries a number you can check against the filing. Where the filing supplies no value at all, the rule cannot be evaluated and the company drops out anyway, and how often that happens on each screen is worth reading before you treat any pass count as a measure of strictness.
Every screen page prints its rules twice, once in words and once as the underlying metric and threshold. The Buffett Quality screen shows roe >= 0.15 directly beneath "ROE >= 15%", so you can read the computation rather than the description. That is the point of printing both, and it also means a disagreement between the two becomes visible instead of staying hidden. One such disagreement is covered below. The criteria, the metric definitions and the current pass counts are open to everyone. The complete passing list and the per-stock audit detail sit on the Pro plan.
Stock screener rules ranked by how much they cut
For each rule you can ask three separate questions:
- Clears it alone: how many of the 3,787 companies satisfy this rule, ignoring every other rule.
- Sole blocker: how many companies fail this rule and pass every other rule in the screen.
- If removed: how many companies the screen would return if this rule were deleted.
The third number is the honest measure of how much a rule matters. Run it and the intuitive answer is wrong more often than not.
One reconciliation note before the tables, because this article asks you to check its arithmetic. The published screen pages report a holdings count. The audit data reports every company's per-rule result. On some screens the two differ by a few names, because the published list applies a final construction step after the rules are evaluated. The columns below are computed entirely from the audit population, so every row reconciles internally. Both figures are given wherever they differ.
Buffett Quality: the ROE floor, not the margin rule
Buffett Quality publishes 16 holdings, and 17 companies satisfy all five rules in the audit data. Read those rules and the 20% net profit margin requirement looks like the brutal one. Very few businesses keep a fifth of revenue as profit.
The data says otherwise:
| Rule | Clears it alone | Sole blocker for | Screen without it |
|---|---|---|---|
| ROE at least 15% | 248 | 151 | 17 to 168 |
| EPS growth 5Y at least 5% | 1,212 | 18 | 17 to 35 |
| Debt/Equity 0.5 or lower | 2,087 | 23 | 17 to 40 |
| Net profit margin at least 20% | 695 | 6 | 17 to 23 |
| Positive equity | 3,313 | 1 | 17 to 18 |
The margin rule that reads as severe is cleared by 695 companies on its own, and it is the sole reason for exclusion in just 6 cases. Drop it and the screen goes from 17 to 23. The return on equity floor is cleared by only 248 companies, blocks 151 that would otherwise pass everything, and removing it takes the screen from 17 to 168. Each row reconciles: 168 companies clear the other four rules, 151 of them are stopped by return on equity alone, and 17 remain.
If you run this screen, the return on equity threshold is essentially the screen. Everything else is trim. The closest single miss on that rule sits at 14.08% against the 15% requirement, which is the kind of margin that should make you examine the threshold rather than the company.
Defensive Investor: the growth rule, not the ten year streaks
The Graham Defensive Investor screen returns 3 companies, and 3 companies satisfy all five rules in the audit data. Its two most quotable rules are the ten year dividend record and the ten year run of positive earnings. Those sound like the constraints that leave only three names standing.
They are not. The ten year dividend record is cleared by 1,133 companies on its own and is the sole blocker for 2. The ten year positive earnings record is cleared by 1,446 and is the sole blocker for 1. The binding rule is the ten year earnings growth rule: 333 companies clear it alone, it is the sole reason for exclusion in 114 cases, and removing it takes the screen from 3 to 117.
One caveat belongs with that finding, and it is exactly the kind of thing printing the computation is supposed to surface. On this screen the human readable label and the computed threshold behind it do not currently agree. The label describes a percentage growth test. The computation applies an absolute change threshold, which is a materially stricter test than the label implies. We have flagged it for correction. Until it is fixed, read the growth rule as this screen's own construction rather than as Graham's published one third test, and read the result above as a statement about our implementation.
A screen that goes from 3 results to 117 when you remove one rule is not really a five rule screen. It is a one rule screen with four sanity checks attached. Knowing which one you are on matters before you conclude that the market has no conservative balance sheets left.
Magic Formula and Acquirers Multiple: two rules, one live rule
The two rule screens make the pattern unmistakable. Magic Formula publishes 8 holdings, and 9 companies satisfy both rules in the audit data. The earnings yield test is cleared by 299 companies alone and is the sole blocker for 58. The 20% return on invested capital test is cleared by only 67 alone and is the sole blocker for 290. Remove the return on capital rule and the audit population goes from 9 to 299. Remove the yield rule and it goes from 9 to 67. The quality half of the formula, not the cheapness half, is doing the filtering in this market.
Acquirers Multiple is more lopsided still. It publishes 75 holdings, and 81 companies satisfy both rules in the audit data. Its market capitalization floor of 100 million dollars is cleared by 2,997 companies and is the sole blocker for 25. Its 15% earnings yield requirement is cleared by 106 and is the sole blocker for 2,916. Both rows reconcile to the same 81. That screen is a yield screen with a size filter attached, and one line of the audit says so.
The general lesson is worth stating plainly: you cannot tell which threshold is binding by reading how strict it sounds. The intuitive answer was wrong on both Buffett Quality and Defensive Investor. A screener that keeps the per-rule detail lets you check instead of guess.
Near misses are where the judgment lives
The pass or fail boundary is the least interesting part of a screen, because it is the part you chose arbitrarily. The shape of the distribution around it matters more.
Buffett Quality has 17 companies clearing all five rules, and another 199 clear four of five. Defensive Investor has 3, and another 128 clear four of five. Those near miss groups run from roughly twelve to roughly forty three times the size of the pass list, and every company in them has a documented single point of failure.
That is a research queue, not a rejection pile. A company failing only the return on equity test in the fourteens is a reasonable candidate for a closer look. A company failing four rules is a lower priority for the same amount of your time. Without a per-rule breakdown, both look identical: absent.
Running your first screen in four steps
If you want the practical version rather than the analysis, this is the whole workflow:
- Pick a screen by what you want it to prove, not by the name attached to it. Quality bias points to Buffett Quality or Quality at Reasonable Price. Deep value points to Graham Net-Net or Acquirers Multiple. Balance sheet conservatism points to the two Graham screens.
- Read the criteria panel before the results. Each rule appears with its metric and threshold. If a threshold looks arbitrary to you, that is worth knowing before the list influences your thinking.
- Check the pass count against the evaluated count. A screen returning 3 of 3,787 is telling you something about the rules as much as about the market.
- Work the near misses, not just the passes. The companies failing one rule, with the actual value shown, are where your own judgment can add something a fixed threshold cannot.
Reading the backtest without fooling yourself
Nothing in this section is a recommendation, a forecast, or a claim that any screen will make money. Every screen links to a monthly rebalanced backtest with an equity curve and the usual risk statistics. The full charts and period returns are part of the Pro plan. Two honest caveats belong with those figures, and we would rather print them than have you discover them later.
The performance panel on the screen cards covers a trailing twelve month window. The CAGR shown there matches the one year return figure to within a fraction of a point on every screen, because both measure the same twelve months. It is a recent result, not a durable property of the strategy.
The stored histories are reconstructed, not a live track record. The eight screens hold between 28 and 134 monthly versions, and they begin on eight different dates ranging from May 2015 to May 2024. That spread is the signature of a backfill built from point in time filings rather than a record published in real time and left untouched. The reconstruction uses the data on file at each date, which is the right way to build it, but it is not a decade of live results and we do not present it as one. Past screen performance does not indicate future results.
If a screener shows you a smooth equity curve without telling you which of those two situations you are in, that is the question to ask it.
How the underlying data gets there
The screens rebuild on the first of each month from SEC EDGAR filings, with Fiscal.ai as a data partner. The run effective 2026-08-01 evaluated 3,787 companies. In that run the median company filing carries a date of 2026-05-11 and the most recent is 2026-07-30, which is the ordinary gap between a reporting period and the screen that reads it.
The tail is worth stating too, because a screen is only as fresh as the filing under each company. In the same run, 270 companies carry a filing more than 180 days old, 62 carry one more than a year old, and the oldest is from March 2017. Those are typically companies that have stopped filing regularly. If a name looks surprising on a screen, the filing date behind it is the first thing to check.
One precision point on the word audited. Annual reports on Form 10-K are audited. Quarterly reports on Form 10-Q are reviewed rather than audited. The accurate description of these screens is that they run on figures companies filed with the SEC, some audited and some reviewed. That distinction is worth carrying into any screener you use, including this one.
Where this sits against the other options
GuruFocus is the deepest fundamental database of the group, with extensive 13F tracking and hundreds of metrics per company. AAII has decades of published screen research behind it. Validea is built around following specific guru model portfolios and publishes per criterion detail for its models. All three are serious products, and if you need what they specialize in, use them.
What GuruScreens does is narrow and specific. Every threshold is printed as a metric and a number beside its human readable label. Every evaluated company keeps a per-rule result with the actual value behind it wherever the filing supplies one. The screen library, its criteria and the current pass counts open on the free tier. Weigh that against what each of the others gives you, because the right tool depends on which of those things you actually need. There is a fuller side by side in the guru stock screener comparison and a migration focused piece for people moving off GuruFocus.
A note on what these screens are. The thresholds are the GuruScreens interpretation of a publicly described method, not rules the named investors published. Warren Buffett never published a 15% return on equity cutoff. Greenblatt's Magic Formula ranks earnings yield against return on capital and excludes financials and utilities rather than applying fixed floors. Carlisle's Acquirers Multiple is enterprise value over operating earnings rather than an earnings yield minimum. Graham's defensive growth test was a one third increase in per share earnings over ten years. We think the interpretations are reasonable, and we print the computation so that you can disagree with them precisely.
Choosing a starting screen
Nothing here is a recommendation to buy or sell anything, and a passing stock is not a suggested purchase. With that said, if you want a starting point rather than a survey:
- Learning how screens behave: start with a two rule screen such as Magic Formula or Acquirers Multiple. With two rules the sensitivity analysis is trivial to follow and you build intuition quickly. The Magic Formula explainer covers the reasoning behind the ranking approach.
- Quality bias: Buffett Quality or Quality at Reasonable Price. Expect small result counts and expect the return on equity threshold to be the lever.
- Balance sheet conservatism: the two Graham screens. The Defensive Investor walkthrough explains why it returns so few names, and the Piotroski F-Score guide is a useful companion for judging whether a cheap company is improving or deteriorating.
Whichever you pick, read the criteria before you read the list. The list tells you what survived. The criteria, and the counts behind them, tell you what your screen believes.
If you want a different kind of screener
Two adjacent tools are worth knowing about, because a guru value screen is not the right instrument for every question. Both are MarketXLS products, and GuruScreens is itself operated by MarketXLS, so treat this as a pointer to sister tools rather than an independent survey of the market.
The MarketXLS AI stock screener takes a plain language description of what you are looking for and turns it into filters, which suits exploratory work where you do not yet know the thresholds you want. It has a different design goal from the screens described here, which are deterministic and run fixed published criteria, so the same inputs always produce the same output and every result is reproducible.
Prefer to build the logic yourself with your own thresholds? Here is how to build a custom stock screener in Excel, which walks through the same screening ideas in a workbook you control.
Frequently asked questions
What is a stock screener?
A stock screener applies numeric filters to a universe of companies and returns the ones that satisfy every filter. The screens described here apply value investing criteria to 3,787 US companies and rebuild monthly from SEC filings.
Why does my stock screener return no results?
Usually because one threshold in the set is far more restrictive than the others, not because all of them are tight. Defensive Investor returns 3 companies, and removing a single rule takes it to 117. A per-rule breakdown identifies which threshold is responsible instead of leaving you to guess.
Can I see why a specific company failed a screen?
Every evaluated company keeps its result for each rule, including the threshold, the actual value from the filing where one is available, and the pass or fail outcome. The rules and thresholds are open to everyone. The per-stock audit detail is part of the Pro plan.
How often does the screening data update?
The screens rebuild on the first of each month. Each run records the filings behind it, so the gap between a reporting period and the screen reading it stays visible.
Is the backtest a real track record?
No. It is a reconstruction built from point in time filings, and the eight screens begin on eight different dates between 2015 and 2024. The performance panel on the screen cards covers a trailing twelve month window. Neither is a live published record, and past screen performance does not indicate future results.
Do I need to pay to see the screening rules?
No. The free tier opens the screen library, the criteria with their metrics and thresholds, and the current pass counts. The Pro plan at 99 dollars per year adds the complete passing list, the per-stock audit detail, and the full backtest charts and period returns. Both appear on the pricing page.
The bottom line
A stock screener that returns only a list asks you to trust it. A stock screener that returns a list plus the per-rule evidence behind every inclusion and exclusion lets you audit it, and auditing changes conclusions. Buffett Quality is a return on equity screen wearing four other rules. Defensive Investor is a growth rule wearing two famous streak requirements. Neither fact is visible from the result count, and both are obvious from the per-rule counts.
Open the full screen library and read the criteria and pass counts on the free tier, then compare the plan details if you want the complete passing lists, the per-stock audit detail and the full backtest range.
GuruScreens is a MarketXLS product. GuruScreens is not affiliated with, endorsed by, or sponsored by Warren Buffett, Berkshire Hathaway, Joel Greenblatt, Tobias Carlisle, John Neff, Joseph Piotroski, or the estate of Benjamin Graham. All names, trademarks and product names referenced here, including GuruFocus, AAII, Validea, Fiscal.ai and QuoteMedia, belong to their respective owners and are used only for identification. Nothing here is investment advice, a recommendation to buy or sell any security, or a prediction of future results. Company names shown in screen results are examples of screen output and are not suggested purchases. Screening criteria are our interpretation of publicly described methods and may differ from the originals. Past screen performance does not indicate future results. Always do your own research.