What is measured?
Price, accounting data, market behavior, and corporate events describe different parts of a business. A method cannot answer a question its inputs do not represent.
FREE STOCK RESEARCH EDUCATION
Understand the method before trusting the score.
Stock Scanner is a free educational research tool. This guide explains what twelve widely used approaches measure, why they can appear to work, and the conditions in which they can fail. The public scanner uses synthetic companies and does not provide live recommendations, individualized advice, or a promise of returns.
FOUR QUESTIONS FOR EVERY SIGNAL
A score is only the start of the investigation.
When was it known?
A fair historical test must use information available at that date. Revised filings and future index membership create misleading hindsight.
What could break it?
Trading costs, crowding, structural change, weak accounting, and a different market regime can erase an apparent advantage.
Is evidence independent?
Three similar price signals are not three independent opinions. Related methods should be grouped before forming a combined view.
TWELVE RESEARCH LENSES
What each method sees—and misses
01
Relative value composite
Plain explanation: Compare similar companies and ask which price looks low relative to earnings, sales, assets, and cash generation. Combining several ratios reduces dependence on a single accounting number.
Why it may work: Investors can overreact to unpopular businesses. The central risk is the value trap: a company may be cheap because its economics are deteriorating. Compare within industries, normalize unusual items, and pair price with balance-sheet and business-quality evidence.
Associated thinkers: Benjamin Graham and Warren Buffett. Primary case source: Berkshire Hathaway 1985 letter.
02
Piotroski F-score
Plain explanation: Use nine accounting checks covering profitability, funding, liquidity, and operating efficiency to separate financially improving firms from weak ones.
Why it may work: A low valuation alone says little about financial health. The F-score adds a structured quality check, but it depends on reliable statements and may not transfer cleanly to banks, insurers, or businesses with unusual accounting.
Researcher: Joseph Piotroski. Original Journal of Accounting Research paper.
03
Profitability and ROIC quality
Plain explanation: Look for businesses that repeatedly earn attractive returns on the capital required to operate, while converting reported profit into cash.
Why it may work: Durable economics can support compounding, but a superb company can still be a poor investment at an extreme price. Estimate maintenance investment carefully and test whether competitive advantages are weakening.
Associated capital allocators: Charlie Munger and Warren Buffett. Primary discussion: Berkshire Hathaway 2007 letter.
04
Growth and earnings surprise
Plain explanation: Measure whether revenue, earnings, and guidance are improving faster than the market expected—not merely whether growth is positive.
Why it may work: Expectations may adjust gradually after important news. It can fail when one-time items create a surprise, estimates are stale, or an expensive valuation already assumes exceptional growth.
Early evidence is associated with Ray Ball, Philip Brown, Victor Bernard, and Jacob Thomas. Review of post-earnings-announcement drift research.
05
Factor exposure and residual return
Plain explanation: Separate broad, repeatable market exposures—such as size, value, profitability, and investment—from company-specific return.
Why it may work: It prevents a portfolio from mistaking a common market exposure for unique skill. Factor definitions are models rather than laws; results depend on the market, period, implementation, and trading costs.
Researchers: Eugene Fama and Kenneth French. Fama–French data library.
06
12–1 relative momentum
Plain explanation: Compare recent performance across securities, usually skipping the latest month, and test whether stronger trends persist for a while.
Why it may work: Information can spread gradually and investors may underreact. Momentum can reverse abruptly, concentrate in crowded trades, and incur high turnover. It needs diversification, cost controls, and crash-risk limits.
Researchers: Narasimhan Jegadeesh and Sheridan Titman. Original Journal of Finance paper.
07
Moving-average trend
Plain explanation: Compare shorter and longer price trends to describe direction and reduce exposure when persistent weakness appears.
Why it may work: Large economic adjustments often unfold over time. Trend rules suffer repeated small losses in sideways markets and can react late after sudden reversals. Position sizing matters more than a perfect moving-average length.
Associated pioneer: Richard Donchian. Long-run trend-following evidence review.
08
Short-term oversold reversal
Plain explanation: Identify unusually sharp recent declines and test whether temporary selling pressure may normalize.
Why it may work: Liquidity shocks and emotional trading can push price beyond a reasonable short-run response. A falling price may also reflect new permanent information, so reversal signals require news, liquidity, and balance-sheet checks.
Researchers: Werner De Bondt and Richard Thaler. Original overreaction paper.
09
Low volatility and downside risk
Plain explanation: Prefer stocks with more stable returns, smaller drawdowns, or lower sensitivity to broad market moves.
Why it may work: Investors may overpay for lottery-like upside or face constraints that push them toward high-beta securities. Low-risk portfolios can become crowded, lag speculative rallies, and hide sector concentration.
Researchers and practitioners include David Blitz and Pim van Vliet. Low-volatility research overview.
10
Corporate execution events
Plain explanation: Evaluate observable actions such as repurchases, guidance changes, capital allocation, and whether management follows through.
Why it may work: Actions can reveal information that promotional language does not. Repurchases destroy value when completed above intrinsic value, while guidance can be managed or withdrawn. Evidence needs dates, authorization size, actual execution, and funding source.
Capital-allocation cases appear in Berkshire Hathaway shareholder letters. Berkshire Hathaway 2011 letter.
11
Accrual and earnings quality
Plain explanation: Compare accounting profit with cash generation and working-capital movements to identify earnings that may be less persistent.
Why it may work: Investors may focus on headline profit without separating cash and accounting estimates. High accruals are not automatically manipulation; rapid growth, acquisitions, or industry structure can produce legitimate differences.
Researcher: Richard Sloan. Original accruals research.
12
Explainable multi-family consensus
Plain explanation: Combine genuinely different method families, cap the influence of correlated signals, and show both agreement and disagreement.
Why it may work: No single lens works in every environment. A combined view can reduce dependence on one assumption, but complexity can hide overfitting. Every component, weight, conflict, missing input, and decision to withhold a score should remain visible.
Multi-style research is associated with Cliff Asness and Antti Ilmanen. Investing with Style.
RESPONSIBLE COMBINATION
A practical five-step research workflow
- Define the question. Specify market, date, and horizon before examining the answer.
- Choose independent families. Combine business value, quality, price behavior, and events instead of repeating similar signals.
- Record contrary evidence. Every favorable claim needs a clear condition that could invalidate it.
- Withhold when evidence is weak. Missing, stale, conflicting, or unauthorized data should produce “not assessed,” not a fabricated probability.
- Preserve the research state. Save inputs, dates, assumptions, and limitations so another person can reproduce the reasoning.
Historical associations and famous practitioners do not prove that a method will generate future profits. Fees, taxes, market impact, data revisions, survivorship, and model selection can materially change results.
FREQUENTLY ASKED QUESTIONS
Important boundaries
Does Stock Scanner predict which real stock will rise?
No. The current public edition uses synthetic companies and deterministic example scores for education. It does not publish live stock probabilities or trading recommendations.
Is a high score evidence of future return?
No. A score is a structured summary of selected inputs. It must be considered with uncertainty, opposing evidence, data timing, implementation costs, and method failure modes.
Will ads influence rankings?
No. Advertising is restricted to educational reading surfaces. Advertisers cannot purchase a score, ranking, favorable method description, or inclusion in scanner results.
What data is stored?
The public scanner stores watchlists, saved scans, and research notes in the current browser. Review the privacy notice and use the built-in backup or deletion controls.
Use the ideas on transparent synthetic examples.
Compare multiple lenses, inspect contrary evidence, and save the reasoning locally—without creating an account.
Open Stock Scanner