Altman Z-Score: The 1968 Bankruptcy Predictor That Still Works

A formula written in 1968 still flags roughly 80-90% of corporate bankruptcies a full year before the filing. Most retail investors have never run it on a stock they own.

What is the Altman Z-Score?

The Altman Z-Score is a multi-factor formula that estimates the probability a public manufacturing company will file for bankruptcy within roughly two years. NYU finance professor Edward Altman developed it in 1968 by analyzing 66 firms — half of which had filed Chapter 11 — and identifying the combination of ratios that best discriminated between survivors and casualties.

The result is a single composite score. Above ~2.99 sits a "safe zone." Below ~1.81 is the "distress zone." In between is a gray area Altman himself called ambiguous.

The model's enduring usefulness is that it forces an investor to look at five different angles of solvency at once — liquidity, retained earnings, operating profitability, market confidence, and asset turnover — instead of fixating on a single ratio.

A high P/E is fashionable; a low Z-Score will end your position. The order of operations matters more than most investors admit.

How is the Altman Z-Score calculated?

The original formula for public manufacturing companies is:

Z = 1.2 × (Working Capital / Total Assets) + 1.4 × (Retained Earnings / Total Assets) + 3.3 × (EBIT / Total Assets) + 0.6 × (Market Cap / Total Liabilities) + 1.0 × (Sales / Total Assets)

Each ratio captures a distinct dimension:

The weights are not arbitrary. Altman derived them through discriminant analysis — the same statistical technique used to classify spam emails. Higher coefficients reflect ratios with more bankruptcy-predictive power.

Note that Market Cap / Total Liabilities is the only market-based input. Strip it out and Altman essentially becomes an accounting model. Including it lets the market "vote" on solvency, which is part of why the formula performs as well as it does.

How do you interpret the score?

Three zones, simple to memorize:

The original 1968 study reported the model correctly classified about 95% of bankruptcies one year before filing. Subsequent out-of-sample tests over the next ~30 years showed accuracy degrading to roughly 80-90% — still strong, but worth knowing the limit.

There is also a meaningful Type II error rate of around 15-20% — firms flagged as distress that ultimately survive. That false-positive rate is acceptable for a screening tool but disqualifying as a sole-criterion stock-picking model. Use the score to flag risk, not to bet against companies.

How does Altman Z-Score look on real stocks?

Three illustrative examples using approximate FY2025 numbers (rounded for clarity, since denominators move quarter to quarter):

Two takeaways from this list.

First, mature, asset-light, cash-generative businesses like Apple (AAPL) and Costco (COST) routinely score in the safe zone — but their stocks can still underperform if growth disappoints. Z-Score does not measure stock attractiveness; it measures balance-sheet survivability.

Second, capital-heavy businesses with leverage like AT&T (T) often print sub-1.81 readings without being in obvious trouble. The model penalizes leverage uniformly, which means investment-grade-rated firms still flunk. Pre-revenue EVs like Lucid (LCID) and Rivian (RIVN) flunk for the obvious reason: no retained earnings, negative EBIT, large liabilities relative to market cap.

What are the model's biggest weaknesses?

Three caveats that matter in 2026.

Financials are excluded by design. Altman's discriminant analysis was trained on industrials. Banks, insurers, and asset managers have liability structures that violate the model's assumptions. Never apply Z-Score to JPMorgan (JPM), Bank of America (BAC), or BlackRock (BLK). Use bank-specific solvency ratios (Tier 1 capital, common equity tier 1) instead.

Asset-light tech distorts the score. Software companies like Palantir (PLTR), Salesforce (CRM), and Adobe (ADBE) have low total assets and high market caps, which can produce inflated Z-Scores even when operating performance is mediocre. The "Sales / Total Assets" denominator was calibrated for inventory-heavy manufacturing, not 80% gross-margin SaaS.

Goodwill and intangibles complicate interpretation. Mergers stuff the balance sheet with goodwill, which inflates total assets and pushes ratios down. After major M&A, a firm may "fail" the Z-Score until amortization runs through earnings — even when underlying operations are healthy.

For these reasons, Altman developed two variants — the Z' (private manufacturing) and Z" (non-manufacturer / emerging markets) — with different coefficients and different cutoffs.

When should you actually use the Altman Z-Score?

Three scenarios where it is genuinely useful.

Pre-screening watchlists. Run Z-Score across your watchlist quarterly. Anything that drops from Safe to Gray is worth a deeper look — the model works as a "fire alarm" even when it is not a stock-picking system.

Stress-testing dividend-paying stocks. A 5% yield from a Z = 1.2 firm is often an artifact of a falling stock price, not management generosity. Pair Z-Score with free cash flow yield to separate sustainable distributions from those at risk of cuts.

Evaluating cyclical bottoms. During recessions, plenty of solid businesses temporarily flunk Z-Score because their EBIT compresses. Buying Z = 1.5 firms at the bottom of a cycle has historically produced strong returns — provided they are not flunking for structural reasons.

The model is a screening tool, not an oracle. Pair it with fundamental analysis and the Piotroski F-Score for a more complete view of financial health.

What is the modern alternative to Altman Z-Score?

Several. None has fully displaced Altman because none combines the same simplicity, transparency, and accuracy at zero cost.

The Beneish M-Score detects earnings manipulation. The Zmijewski X-Score uses logit regression. Moody's KMV model uses option-pricing theory on equity volatility. CDS spreads from credit derivatives markets effectively price-discover bankruptcy probability in real time. Each has strengths Altman lacks; none is as easy to compute on the back of an envelope.

The smartest use of Altman Z-Score in 2026 is as a triangulation point — not as a final answer. It tells you whether a balance sheet would survive a normal recession. It does not tell you whether the stock is cheap, the business is great, or management is honest. Those are separate questions, and they require separate tools.

For more frameworks built on similar logic, the investor profiles on MainRatios cover Howard Marks and Seth Klarman — both of whom use distress-signal frameworks like this one as gating filters before deeper work.

Ready to analyze these stocks yourself? Search any ticker on MainRatios to see valuations from 6 legendary investors - free.