Above the historical screening cutoff. Accounting quality, sector and reporting date can change its usefulness; solvency is not guaranteed. Below the historical screening cutoff. This signals a need for further investigation, not an individual default forecast.
Choose the model before interpreting the number
The public manufacturing model has five ratios and uses market equity. The four-ratio non-manufacturing model uses book equity and excludes sales. It is not the separate private-manufacturing Z′ model. The third option keeps the four-ratio result and adds 3.25, with equally shifted screening bands; it does not supply an independently calibrated emerging-market rating or default probability. Choose the variant before entering the statement, and keep its source visible when comparing results.
For example, the worked public case produces 1.9792. Its financial amounts can be entered in USD units, thousands or millions, provided the same scale is applied throughout. The ratio of working capital to assets remains 0.2 when both amounts are divided by one million. Changing only one amount would instead change the economic assumption. The unit selector therefore changes display rather than the underlying amounts. Supply the reporting period and market-equity valuation date separately, or leave them explicitly unspecified. No missing date is inferred from today or from a preset.
Read a signed bridge rather than a probability
Each factor contributes its raw ratio multiplied by the coefficient of the selected model. Negative working capital, retained earnings, EBIT and book equity remain negative where supported. Assets and liabilities must be positive; the calculator does not replace a zero denominator with a convenient monetary value. Public market equity and sales cannot be negative. The bridge includes the intercept and total separately, so the standardized score cannot hide its extra 3.25 inside an operating factor.
Select a factor button to inspect the numerator, denominator, ratio and weighted contribution. A comparison baseline is a snapshot of entered assumptions, not a company valuation. After changing EBIT, for example, its contribution change and the total score change should reconcile within ordinary floating-point precision. A different model disables that score-delta interpretation. Independently rounded display columns can differ in the final digit, so CSV retains raw values. The historical screening band remains a prompt for further investigation; it neither establishes solvency nor forecasts an individual bankruptcy percentage.
Keep provenance with scenarios and exports
The scenario view varies one selected input while retaining the same model and statement-period metadata. A zero starting value cannot meaningfully change by a percentage, so that case uses disclosed absolute steps tied to total assets instead. Unsupported negative public equity or other invalid scenarios remain unavailable rather than receiving a fabricated score. Use these changes to understand the model’s arithmetic, not to assume that a particular balance-sheet repair is feasible.
Trend inputs identify their model separately. Changing from public market equity to book equity changes the model basis, so a line is not drawn as though those scores formed a continuous comparable series. Presets are hypothetical input illustrations and carry no verified company or filing claim. Copy, CSV and paper receipts preserve coefficients, raw ratios, intercept, thresholds and source references with the entered period. Default URLs contain no financial inputs. Reviewed sharing discloses the encoded assumptions, while device-local saving is opt-in. Neither feature independently verifies the statements or supplies accounting, lending or investment advice.