What is the Piotroski F-score?
By Ghost Man Trades · Last reviewed · Education only, not advice
The F-score in plain English
Cheap stocks are cheap for a reason. Some are good businesses having a bad stretch; others are getting worse and will stay cheap, the classic value trap. In 2000, accounting professor Joseph Piotroski asked whether simple facts from the annual report could tell the two apart. His answer was a checklist of nine yes/no questions. Each "yes" is worth one point, so the total runs from 0 (weak on every test) to 9 (strong on every test).
Every test compares the latest fiscal year with the one before. So the F-score is about direction: is the business making money, is that money real cash, and did profitability, debt, liquidity and efficiency get better or worse?
The 9 checks
As defined in Piotroski (2000), section 2. "Starting assets" means total assets at the beginning of the year, which is why three years of balance sheets are needed.
| # | Check | One point if |
|---|---|---|
| 1 | Return on assets (profit) | Net income ÷ starting assets is above zero |
| 2 | Cash from operations | Cash flow from operating activities is above zero |
| 3 | Change in return on assets | Return on assets is higher than last year |
| 4 | Accruals (quality of earnings) | Cash from operations is bigger than net income |
| 5 | Change in leverage | Long-term debt ÷ average total assets went down |
| 6 | Change in liquidity | The current ratio (current assets ÷ current liabilities) went up |
| 7 | Equity offering | The company did not issue common stock in the year |
| 8 | Change in gross margin | Gross profit ÷ revenue went up |
| 9 | Change in asset turnover | Revenue ÷ starting assets went up |
Checks 1 to 4 are about profitability, 5 to 7 about debt, liquidity and outside money, and 8 and 9 about operating efficiency. Piotroski used net income before extraordinary items; U.S. accounting rules removed "extraordinary items" in 2015 (FASB ASU 2015-01), so today plain net income is used.
Example (hypothetical numbers). Example Co., fiscal years 2024 and 2025, all figures in millions.
- Total assets: $1,000 at the start of 2024, $1,100 at the start of 2025, $1,150 at the end of 2025.
- Net income: $60 in 2024, $77 in 2025. Return on assets: $60 ÷ $1,000 = 6.0%, then $77 ÷ $1,100 = 7.0%. Check 1: yes. Check 3: yes.
- Cash from operations in 2025: $95. Above zero (check 2: yes) and above net income of $77 (check 4: yes).
- Long-term debt ÷ average assets: 2024 $300 ÷ $1,050 = 28.6%; 2025 $330 ÷ $1,125 = 29.3%. It went up. Check 5: no.
- Current ratio: 1.40 last year, 1.52 this year. Check 6: yes.
- Shares outstanding: 50 million both years. Check 7: yes.
- Gross margin: 41.0% last year, 40.2% this year. Check 8: no.
- Revenue ÷ starting assets: $900 ÷ $1,000 = 0.90 last year, $1,045 ÷ $1,100 = 0.95 this year. Check 9: yes.
- F-score: 7 of 9.
What the research found, and what it didn't
Piotroski tested U.S. stocks in the cheapest fifth by book-to-market from 1976 to 1996. Within that cheap group, buying high scorers and avoiding low scorers improved the average one-year return in his sample. That is a finding about averages across many cheap stocks, in one historical period, before trading costs. It does not say a 9 will beat a 4 next year, and later research disagrees on how much of the effect remains now that the score is widely known.
Common mistakes
- Treating it as a buy signal. It describes last year's statements. It knows nothing about price, valuation or what happens next.
- Scoring banks, insurers and REITs. Their balance sheets don't fit the tests (no gross margin or current ratio in the usual sense).
- Guessing missing numbers. If a company doesn't report gross profit or a current ratio, a check can't be scored. Treating "not reported" as "no" or as zero changes the score.
- Mixing data definitions. Comparing one year's long-term debt figure with a differently defined figure the next year can flip a check.
- Forgetting it's a year old. An annual report can be up to 15 months old by the time the next one arrives.
How Ghost Man uses this
Every stock page and every stock row in the Ghost Report shows a Quality score, which is the Piotroski F-score computed from the company's own annual filings, with all nine checks, the actual numbers and the two fiscal years compared. Our choices, stated plainly:
- Data: SEC EDGAR XBRL financial data from the company's 10-K filings. When SEC data can't be reached or a company doesn't file a U.S.-GAAP 10-K, we fall back to Yahoo Finance annual financials and label the source on the page.
- Check 7: we use "shares outstanding did not go up," the usual stand-in for "did not sell new stock," because it works with both data sources.
- Check 5: a company that reports no long-term debt in either year gets the point (debt did not rise).
- Missing numbers: a check that needs a number the filings don't report shows n/a and no total out of 9 is given. Fewer than three annual reports: "n/a (not enough filings)."
- Banks, insurers and REITs are not scored.
- It is a health check, not part of the Ghost Man buy rules, and its track record with Ghost Man hasn't been tested yet. We'll publish that test, good or bad, when it's done.
Related terms
- Free cash flow · What is free cash flow?
- Return on equity (ROE) · What is return on equity (ROE)?
- Debt-to-equity ratio · What is the debt-to-equity ratio?
- Value trap · What is a value trap?
- Backtesting and survivorship bias · What is backtesting, and how does survivorship bias distort it?
- Trailing P/E and EPS · What are EPS and the trailing P/E ratio?
Learn it step by step in Swing School
Frequently asked questions
What is a good Piotroski F-score?
In Piotroski's study, 8 or 9 counted as high and 0 or 1 as low. Most companies land in the middle, between 4 and 7. The score only compares this year's statements with last year's, so it says whether the business got stronger or weaker, not whether the stock is cheap.
Does the F-score work for banks?
Not well. Gross margin, the current ratio and long-term debt mean something different for banks and insurers, and REITs are built around property and debt. Ghost Man Trades does not score banks, insurers or REITs and says so on their pages.
Where does the data come from?
From the company's annual report (Form 10-K for U.S. companies): the income statement, balance sheet and cash flow statement for the latest three fiscal years. Three years are needed because return on assets and asset turnover use the assets at the start of each year.
Does a high F-score mean the stock will go up?
No. Piotroski found that, among cheap value stocks in 1976 to 1996, high scorers did better on average than low scorers. Later studies disagree on how much of that edge is left, and an average over hundreds of stocks says little about any one of them.
Why can two websites show different F-scores?
Each step has choices: which net income line, how long-term debt is defined, whether "no new shares" uses shares outstanding or stock actually sold, and whether a missing number counts as zero. Ghost Man Trades shows every number behind each check and leaves a check blank (n/a) instead of guessing.
Sources
We check every definition and formula against these sources. Our own rules come from the Ghost Man scan code that runs this site.
- Piotroski, J. (2000). Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers. Journal of Accounting Research 38 (Supplement), 1-41
- SEC: Beginners' Guide to Financial Statements
- SEC EDGAR: XBRL financial data APIs (companyfacts)