If finding undervalued stocks were as simple as filtering on P/E below 15, the value-investing premium would have been arbitraged away in 1934 when Benjamin Graham first wrote about it. It wasn't, and it still exists, because finding genuinely undervalued stocks is structurally harder than a single-ratio screen. This post walks through the 5-screen method that actually works.
The screens, in order:
- Cheapness (the obvious one)
- Quality (so the cheapness isn't a trap)
- Sector positioning (so you understand the contextual valuation)
- Contrarian indicators (so you're buying when others are selling)
- Behavioural patience (so you actually buy)
Screen 1 - Cheapness
The starting filter. The right cheapness metric depends on the business type:
- Owner-earnings yield (FCF / Enterprise Value): the most-robust general metric. Above 7 percent is interesting for a stable business; above 10 percent is statistically cheap.
- EV/EBIT: useful for cross-sector comparison. Below 10x is on the cheap side; below 6x is deeply cheap (often a value trap warning, see Screen 2).
- P/B (price-to-book): only meaningful for businesses with substantial tangible assets. Banks, real estate, industrial-asset-heavy stocks. P/B below 1.0 is the Graham defensive threshold.
- P/E (price-to-earnings): most-cited but most-misleading. P/E ignores capital structure, ignores quality of earnings, and is meaningless for cyclicals at trough or peak earnings. Use only as a quick first-cut filter.
Run the first screen with FCF/EV > 7 percent or EV/EBIT < 12x. Get the candidate universe down to maybe 200-400 names from the 21,000+ initial universe.
Screen 2 - Quality
Cheap-only screens catch all the value traps. A stock at EV/EBIT 4 with collapsing margins and rising debt is "cheap" right up to the moment it goes to zero. Layer in quality:
- Sustained ROIC > 12 percent over the trailing 5 years (genuinely profitable use of capital)
- Gross margin stable over the trailing 5 years (no sudden margin compression signalling moat erosion)
- Positive operating cash flow in at least 4 of the last 5 years (no story stocks)
- Net debt to EBITDA < 3.0x (no leverage-driven blow-up risk)
Apply these in addition to Screen 1. Your universe drops to maybe 50-100 names.
This is the screen that separates "cheap-and-trapped" from "cheap-and-mispriced." Both look identical on Screen 1; only Screen 2 distinguishes them.
Screen 3 - Sector positioning
A stock that looks cheap on absolute metrics may be expensive within its sector. A stock that looks expensive may be cheap within sector. Context matters.
- Compare each candidate's EV/EBIT to the median EV/EBIT of its GICS sector. Trading below the sector median is interesting; trading at the sector low quartile is more interesting.
- Compare each candidate's FCF yield to the sector median. Above-sector FCF yield with a passing quality screen is the strongest "cheap relative to peers" signal.
- Check the sector's own valuation history. Tech sector at sector-wide EV/EBIT of 25x is high relative to a 10-year median of 18x; energy sector at sector-wide 9x is high relative to a 10-year median of 6x. The sector itself can be cheap or expensive.
After Screen 3, your candidates have all of: absolute cheapness, passing quality, AND being cheap within their sector context. Universe down to 20-40 names.
invest-like.com surfaces the per-sector medians live at /sector/[slug]/ for each of the 11 GICS sectors.
Screen 4 - Contrarian indicators
The strongest value opportunities exist when consensus has turned against a stock. Indicators:
- Recent 12-month price performance below market: stocks down 20+ percent over 12 months while the broader market is up are forced to defend their thesis publicly. If the thesis holds, the price catches up.
- Negative analyst revisions: when 70+ percent of covering analysts have recently downgraded forecasts. The stock is unloved.
- Short interest above 10 percent of float: someone is actively betting against the company. If you disagree with the short thesis after research, the asymmetric setup is strong.
- Recent dividend cut or initiation: cuts trigger selling by yield-focused funds; initiations rarely. Either is informational.
- Insider buying after price decline: insiders buying their own stock after a 20+ percent drop is one of the more reliable signals (insider selling is noisy; insider buying is signal).
Apply these as positive signals on the 20-40 candidates from Screen 3. Stocks that pass at least one Screen 4 indicator are statistically more likely to be genuinely mispriced.
Screen 5 - Behavioural patience
The final screen happens between your ears. Of the 10-20 stocks that survive Screens 1-4:
- Can you write the 2-sentence buy thesis without help? If you can't, you don't understand the company well enough.
- Can you specify what would make you sell? "If the moat erodes" beats "if it goes up 50 percent."
- Can you commit to the 24-hour rule? Sleep on it. The Munger pre-mortem trick: if you slept on the buy decision and still want to execute, your conviction is real. If the urgency faded, it was probably noise.
This is the part that no screener can do for you. The Buffett-Fit Score on invest-like.com handles Screens 1-3 mechanically and surfaces signals for Screen 4. Screen 5 is your job.
A concrete worked example
Let's run the 5-screen process at the universe level to see how it narrows.
Initial universe: 21,000+ US-listed and major-international stocks.
Screen 1 (cheapness): filter for FCF/EV > 7 percent OR EV/EBIT < 12x. Reduces to roughly 800-1,200 stocks (depending on market conditions).
Screen 2 (quality): ROIC > 12%, stable gross margin, positive OCF, ND/EBITDA < 3x. Drops to roughly 80-150 stocks. The contrast with Screen 1 is stark - most "cheap" stocks fail at least one quality test.
Screen 3 (sector position): trading below sector median EV/EBIT. Drops further to roughly 30-50 stocks.
Screen 4 (contrarian): at least one positive contrarian indicator. Drops to roughly 10-20 stocks. This is your active research universe.
Screen 5 (behavioural): your specific conviction filter applied to your portfolio. Typically you'll buy 1-3 names from Step 4 in a given quarter.
The process is intentionally restrictive. The biggest mistake retail investors make is skipping Screens 2-5 and going straight from Screen 1 to a purchase decision. That's how you collect a portfolio of value traps.
Common false signals
A few stocks routinely look "undervalued" but aren't:
- Cyclicals at trough earnings: look cheap on TTM P/E because TTM is the cycle bottom. Use 10-year normalised earnings instead.
- Single-customer concentration: a stock with 60 percent of revenue from one customer is cheap for a reason; the customer loss is a real risk.
- Litigation overhang: a stock cheap on EV/EBIT but with a pending class-action of 20+ percent of market cap is not really cheap. Subtract the expected litigation cost from market cap and re-check.
- Foreign currency mismatch: a company earning mostly in USD but reporting in EUR can show artificial EBIT volatility from forex; the underlying business may be fine.
- Accounting one-offs: a stock that just took a goodwill impairment looks cheap on TTM P/E because the impairment depressed it. Normalised earnings ex-impairment is the real number.
The 5-screen method catches most of these because Screen 2 (quality) and Screen 3 (sector positioning) require sustained metrics across cycles, not point-in-time TTM ratios.
How invest-like.com does this
invest-like.com runs Screens 1-3 mechanically on every stock in the indexed universe at every refresh:
- The /best/owner-earnings-yield/ ranking is Screen 1 cheapness sorted
- The /fit/buffett/ ranking layers Screen 2 quality on top
- The /sector/[slug]/ pages show Screen 3 sector context
- The /buffett/[ticker]/ pages show recent price action, dividend changes, insider activity for Screen 4 inputs
Screen 5 is by definition your own. The 27-question checklist at /blog/value-investing-checklist-printable-2026/ is the closest formalisation we offer.
Further reading
Educational only. Not investment advice.