Lesson 4 of 7
The forensic pass
Ten years of cash against earnings, three forensic scores run together, the auditor and 8-K history, and the share count, before trusting any figure.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- Ten years of operating cash flow set against ten years of net income is the sturdiest single check, because either number alone is noisy for a single year.
- Piotroski, Altman and Beneish each ask a different question of a different sample, and run together cover more ground than any one alone.
- The auditor's own opinion, two years of Form 8-K, and the share count add three checks no ratio computes.
One fixed pass, run the same way on every idea that survives the first hour, before any figure from the filing is trusted for the steps that follow it.
The forensic pass, in order
- Cash against earnings, ten years summed, not one year at a time.
- Piotroski, Altman and Beneish, read together rather than any one alone.
- The auditor's opinion, and two years of Form 8-K for items 4.01 and 4.02.
- The diluted share count, five years, since a reported buyback can mask dilution elsewhere.
Why ten years, not one
A single year's accrual, net income minus operating cash flow, moves for reasons that have nothing to do with earnings quality: a large customer paying late at year end, one receivables build ahead of a strong quarter. Summed across ten filed years, the same kind of gap stops being a swing and starts being a pattern.
Did ten years of profit arrive as cash?
- Net income, ten years$1,000
- Operating cash flow, ten years$600
Three scores, three different questions
Piotroski's nine pass-or-fail tests ask whether a cheap company's accounts are strong and improving, chosen in advance and tested only inside the cheapest fifth of the market by book to market. Altman's five ratios ask about failure, fitted on public manufacturers that went bankrupt against ones that did not. Beneish's eight indices ask about manipulation, fitted on companies later found to have manipulated earnings. A company passing only one of the three has answered only one of the three questions, not the other two.
What no ratio computes
A disclosed material weakness in internal control says more about whether the numbers can be trusted than any ratio built from those numbers. So does a change of auditor (Form 8-K item 4.01), or a non-reliance determination (item 4.02). A diluted share count that keeps rising, read against buybacks the company reports separately, says whether the business is shrinking the share count or quietly growing it.
Check yourself
Why sum ten years of cash against earnings instead of reading a single year?
- Because one year's cash flow is never audited
- A single year's gap is noisy; summed across ten years it becomes a pattern
- Because Sloan's own test measured ten-year sums
Show the answer
Answer: B.
One year of accruals can move for reasons unrelated to earnings quality. The multi-year sum is the sturdier check, not a rule from a study: Sloan's own test used single-year accruals.What sample was the Altman Z-Score fitted on?
- The cheapest fifth of the market by book to market
- Public manufacturers that failed, matched against ones that did not
- Companies already found to have manipulated earnings
Show the answer
Answer: B.
Altman's 1968 model was built and tested on publicly traded manufacturers, which is also why it reads oddly on banks and asset-light services.A wide gap between ten-year earnings and ten-year operating cash flow proves what, on its own?
- That the company has manipulated its results
- That the company is heading toward failure
- Nothing on its own. It says where to read next
Show the answer
Answer: C.
A working-capital build, gains on asset sales and manipulation can all produce a similar gap. The pass tells you where to look, not the answer.Sources
- Sloan, The Accounting Review, 1996: the accrual part of earnings persists less than the cash part, the reason step one checks cash against earnings at all.
- Beneish, Financial Analysts Journal, 1999: the M-Score, eight indices, a published cutoff of minus 1.78.
- Altman, Journal of Finance, 1968: the Z-Score, fitted on public manufacturers, zones at 1.81 and 2.99.
- Piotroski, Journal of Accounting Research, 2000: the F-Score, tested inside a high book-to-market cohort.
- Dechow, Ge, Larson and Sloan, 'Predicting Material Accounting Misstatements', Contemporary Accounting Research, 2011: misstating firms showed unusually high accruals and were raising new financing, which also motivates the share-count step.
What this leaves out
- The ten-year figures in the worked example are illustrative. No filed company appears here.
- Each score answers the question it was built for, and no other. Passing all three is not a finding about value or quality on its own.
- Operating cash flow can be managed too: stretched payables, sold receivables, costs moved into investing.
- The forensic pass flags where to read next. It does not replace reading the filing, and it does not detect every kind of misstatement.
- Nothing here is advice about any security.