Beneish M-Score
EliteEight ratios that flag possible earnings manipulationThe Beneish (1999) M-Score combines eight indices from the latest two fiscal years. It was built to detect the statistical fingerprint of earnings manipulation, and it deliberately flags many fast-growing, healthy companies too. A high score is a prompt to dig in, never a verdict. The model was calibrated on industrial firms, so REITs and financials (whose balance sheets it reads poorly) score noisily; the indices are winsorised to keep a single degenerate ratio from dominating. Not investment advice.
How it works
Messod Beneish built this to answer a forensic question: does this company's accounting resemble the accounting of companies later caught overstating earnings? Eight ratios, each comparing this year to last, weighted into one score.
They look for fingerprints rather than crimes. Receivables growing faster than sales. Margins quietly deteriorating while the story stays intact. Assets shifting into softer categories. Accruals running ahead of cash.
When it misleads
A high score is not evidence of fraud, and should never be read as an accusation. It says the accounting matches a pattern. Honest companies match that pattern all the time — during an acquisition, a turnaround, or a change in how revenue is recognised.
It was fitted on manufacturers in the 1990s. Subscription businesses and anything carrying heavy deferred revenue trip it for structural reasons that have nothing to do with manipulation.
This screen ranks the most fragile names first, so we don’t publish its list — naming companies as distressed on a public page isn’t something a screening model should do on its own. What today’s run found, in aggregate: