Piotroski F-Score: Rebuild the Original Nine Signals

Reconstruct the Piotroski F-Score from point-in-time annual filings using the original nine binary definitions, exact asset denominators, strict thresholds, equity-offering evidence, and comparability controls.

For educational purposes only; not investment advice. Investing may result in loss.

Direct answer

The Piotroski F-Score sums nine binary financial-statement signals. Each favorable condition earns 1; otherwise it earns 0, so the total ranges from 0 to 9. Joseph Piotroski designed the score to separate financially stronger from weaker companies within a portfolio of high book-to-market firms. It is not a universal quality grade, bankruptcy probability, valuation, price target, or promise of excess return.

The original signals use annual accounting data and exact definitions. Net income before extraordinary items and cash flow from operations are each scaled by beginning-of-year total assets; asset turnover also uses beginning-of-year total assets. Long-term debt is scaled by average total assets. Current liquidity uses fiscal-year-end current assets and current liabilities. The equity signal asks whether common equity was issued, not whether diluted weighted-average shares or net shares outstanding increased.

Binary scoring discards magnitude and causality. A ratio that improves by a tiny amount gets the same point as a large improvement, while equality gets 0 under a strictly positive change rule. The score describes nine historical signals after the underlying filing became available. It does not explain whether improvement came from operations, acquisitions, inflation, accounting choices, asset sales, working-capital timing, or a weaker denominator.

How it works

Rebuild the original score in this order:

Fix the information date and company perimeter. Use the original annual filing and amendments available at the test date, not a later restatement or database refresh. Align issuer, consolidated group, fiscal-year length, currency, continuing operations, predecessor or successor periods, and acquisition or disposition perimeter. In the original return design, compounding began at the start of the fifth month after fiscal year-end to allow for financial-statement availability; a modern backtest should use actual filing timestamps and its stated execution rule.
Create consistent asset denominators. Let current fiscal year be t. For the original profitability and turnover signals, use Beginning assets_t = Ending assets_(t-1). Calculate ROA_t = Net income before extraordinary items_t / Beginning assets_t, CFO_t = Cash flow from operations_t / Beginning assets_t, and TURN_t = Net sales_t / Beginning assets_t. Build prior-year values on the same basis. Do not silently substitute average assets because a data vendor does.
Score four profitability signals. Set F_ROA = 1 if ROA_t > 0, else 0; F_CFO = 1 if CFO_t > 0, else 0; F_DROA = 1 if ROA_t - ROA_(t-1) > 0, else 0; and F_ACCRUAL = 1 if CFO_t > ROA_t, else 0. Because CFO and ROA share the same beginning-assets denominator, the last comparison is equivalent to current CFO exceeding current net income on the defined basis. Preserve discontinued operations, extraordinary-item legacy mapping, and cash-flow classification choices.
Score leverage, liquidity, and funding. Calculate long-term debt including its current portion over average total assets using a consistent convention, and set F_DLEVER = 1 only when the ratio declines. Calculate Current ratio = Current assets / Current liabilities at each fiscal year-end and set F_DLIQUID = 1 only when it increases. Set EQ_OFFER = 1 only if the company did not issue common equity during the year; otherwise set it to 0.
Score two operating-efficiency signals. Calculate MARGIN_t = (Net sales_t - Cost of goods sold_t) / Net sales_t and set F_DMARGIN = 1 only if current margin exceeds prior margin. Set F_DTURN = 1 only if current asset turnover exceeds prior turnover. Map revenue, cost of revenue, pass-through items, excise taxes, inventory methods, and service-company presentation consistently before comparing.
Sum without reweighting. Use F_SCORE = F_ROA + F_CFO + F_DROA + F_ACCRUAL + F_DLEVER + F_DLIQUID + EQ_OFFER + F_DMARGIN + F_DTURN. Missing data is not automatically 0, and a tie is not an improvement. If a bank, insurer, fund, REIT, or company without meaningful gross margin or current ratio cannot support the original inputs, report the score as unavailable or clearly label a modified implementation.
Audit every point and any backtest. Reconcile the income statement, balance sheet, cash-flow statement, equity statement, debt, EPS, acquisitions, discontinued operations, leases, supplier finance, factoring, and subsequent filings. Preserve delistings, corporate actions, survivorship, point-in-time universe membership, transaction costs, liquidity, portfolio formation, book-to-market selection, holding period, and benchmark. The original historical result was conditional on its sample and rules, not a timeless return guarantee.

The paper’s signals were selected for implementability and financially stressed high book-to-market firms; it expressly did not claim they were the optimal measures for every company. A provider may publish a useful modified score, but its label should state deviations in net-income definition, asset denominator, debt field, equity-offering proxy, period frequency, and missing-data treatment.

Example

Use a hypothetical industrial company with comparable annual filings:

Profitability and accrual signals: prior net income before extraordinary items is -$10.0000 million and prior beginning assets are $900.0000 million, so ROA_(t-1) = -$10.0000 / $900.0000 = -1.1111%. Current net income is $30.0000 million, current CFO is $40.0000 million, and current beginning assets are $1,000.0000 million; therefore ROA_t = $30.0000 / $1,000.0000 = 3.0000% and CFO_t = $40.0000 / $1,000.0000 = 4.0000%. Positive ROA, positive CFO, improving ROA, and 4.0000% > 3.0000% earn 4 points.
Leverage, liquidity, and equity issuance: prior long-term debt is $400.0000 million over $950.0000 million average assets, or $400.0000 / $950.0000 = 42.1053%; current long-term debt is $360.0000 million over $1,050.0000 million average assets, or $360.0000 / $1,050.0000 = 34.2857%, earning the leverage point. The current ratio rises from $300.0000 / $200.0000 = 1.5000 to $340.0000 / $200.0000 = 1.7000, earning the liquidity point. The equity statement and financing cash flows show no common-equity issue, so EQ_OFFER = 1; this group earns 3 points.
Operating efficiency and total: prior net sales are $700.0000 million and cost of goods sold is $455.0000 million, so prior margin is ($700.0000 - $455.0000) / $700.0000 = 35.0000%. Current sales are $800.0000 million and cost is $496.0000 million, so current margin is ($800.0000 - $496.0000) / $800.0000 = 38.0000%. Turnover rises from $700.0000 / $900.0000 = 0.7778 to $800.0000 / $1,000.0000 = 0.8000. Both signals earn 1, and total score is 4 + 3 + 2 = 9.
Why decomposition matters: suppose the company issued 1.0000 million shares to employees but repurchased 2.0000 million shares. Net shares fall by 1.0000 million, yet the original no-offering signal is 0 because common equity was issued. Conversely, a stock split can increase the share count without an equity financing inflow. If current ratio moves only from 1.7000 to 1.7001, it still earns 1 under the binary rule; that point does not establish meaningful liquidity improvement.

Risks

Use the filing version and data that were public at the analysis or portfolio-formation timestamp.
Align fiscal-year length, currency, entity perimeter, and predecessor or successor reporting periods.
Reconcile restatements, amendments, discontinued operations, acquisitions, divestitures, and foreign exchange.
Use net income before extraordinary items or document the modern mapping rather than mixing income definitions.
Scale original ROA, CFO, accrual, and turnover measures by beginning-of-year total assets.
Use average total assets for the original long-term-debt ratio and state the averaging convention.
Include the current portion of long-term debt when reconstructing the original leverage field.
Reconcile debt repayment, reclassification, leases, supplier finance, factoring, and acquisition debt.
Build current assets and current liabilities consistently and examine restricted cash and held-for-sale balances.
Treat strict positive tests consistently; equality, rounding noise, and missing values are not improvements.
Identify common-equity issuance from equity and financing records rather than diluted weighted-average shares.
Separate primary issuance, employee awards, options, conversions, acquisition shares, splits, and repurchases.
Map net sales, cost of goods sold, pass-through revenue, excise taxes, and cost classification consistently.
Check whether impairments, asset sales, write-downs, or acquisitions mechanically changed asset turnover.
Inspect receivables, inventory, payables, contract balances, supplier finance, and factoring behind CFO.
Do not infer earnings quality solely from CFO exceeding net income in one annual period.
Report unavailable or modified scores for business models that lack meaningful original inputs.
Preserve the original high book-to-market context when describing the paper’s historical evidence.
Include delistings, survivorship, corporate actions, filing lags, liquidity, costs, and benchmark in backtests.
Do not convert a binary historical score into a valuation, credit rating, causal diagnosis, or return promise.

Common misconceptions

“The original F-Score uses average assets for every asset-scaled ratio.” ROA, CFO, accrual, and turnover use beginning-of-year total assets; the long-term-debt ratio uses average total assets.
“No net increase in shares means the equity-offering point is one.” Issuance can occur alongside larger repurchases, while splits can increase shares without financing; inspect actual common-equity issuance.
“A score of nine proves quality, cheapness, or future outperformance.” It means nine historical binary conditions passed under the chosen definitions, not that valuation or future risk is favorable.
“CFO above net income proves clean earnings.” Working-capital timing, factoring, supplier finance, classification, taxes, and one-time cash flows can change the comparison.
“Any provider’s F-Score exactly replicates the paper.” Denominators, debt fields, equity proxies, missing-data rules, periods, restatements, and information timestamps can differ.

Authoritative sources

Journal of Accounting Research, Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers — original sample, nine definitions, binary composite, portfolio timing, historical results, and limitations.
Stanford Graduate School of Business, Joseph D. Piotroski — author and institutional research profile.
U.S. Securities and Exchange Commission, Beginners’ Guide to Financial Statements — income statement, balance sheet, cash-flow statement, equity, notes, ratios, and audit context.
Investor.gov, How to Read a 10-K/10-Q — filing structure, financial statements, footnotes, MD&A, risks, controls, exhibits, and comparison context.
U.S. Securities and Exchange Commission, Search Filings — point-in-time access to original and amended annual, quarterly, registration, and current reports through EDGAR.
Financial Accounting Standards Board, ASU 2022-04: Disclosure of Supplier Finance Program Obligations — program terms, balance-sheet presentation, outstanding obligations, and annual rollforward disclosures relevant to cash-flow and leverage review.