What the Altman Z-Score is

The Altman Z-Score is a number that estimates how likely a public company is to go bankrupt within the next two years. It does this by combining five accounting ratios into a single weighted sum. The higher the score, the safer the company looks on paper. The lower the score, the louder the alarm.

Edward Altman, then a finance professor at NYU, published the model in 1968 in the Journal of Finance ("Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy," 23(4):589–609). It was the first widely used multivariate statistical model for predicting corporate failure, and in Altman's original calibration sample it correctly classified roughly 95% of firms one year before they went bankrupt.

Sixty-plus years later, the Z-Score is still one of the first sanity checks credit analysts run on a balance sheet. It's not magic. It's a fast filter that tells you whether a company has the basic financial shape of a survivor or the basic financial shape of a distressed name. Anyone with a 10-K and a calculator can compute it in about a minute.

The formula

The original Z-Score, designed for publicly traded manufacturing firms, looks like this:

code-highlight
Z = 1.2·X1 + 1.4·X2 + 3.3·X3 + 0.6·X4 + 1.0·X5

The five variables:

X1 = Working Capital / Total Assets — liquidity. Working capital is current assets minus current liabilities. A company that can't cover near-term bills has a problem.
X2 = Retained Earnings / Total Assets — cumulative profitability. This implicitly captures age too. A young company that has burned cash for years will have a tiny or negative X2.
X3 = EBIT / Total Assets — operating productivity. How much pre-tax, pre-interest profit the asset base is generating. This is the heaviest-weighted ratio in the formula, and for good reason: it strips out capital structure and tax noise to measure the core engine.
X4 = Market Value of Equity / Total Liabilities — solvency cushion. How much equity buffer sits above the debt stack. Notice this uses market cap, not book equity.
X5 = Sales / Total Assets — asset turnover. How efficiently assets are being converted into revenue.

Every number you need is in the 10-K and the current quote. The whole calculation is six division operations and one weighted sum.

A worked example

Imagine a hypothetical public manufacturer we'll call Acme Widgets. We pull its latest 10-K and quote and compute the five ratios. (Every number below is invented for illustration.)

Working capital: $300M
Retained earnings: $800M
EBIT (trailing twelve months): $400M
Market cap: $2,500M
Total liabilities: $1,800M
Sales (trailing twelve months): $4,000M
Total assets: $5,000M

Now we plug in:

X1 = 300 / 5,000 = 0.06
X2 = 800 / 5,000 = 0.16
X3 = 400 / 5,000 = 0.08
X4 = 2,500 / 1,800 = 1.39
X5 = 4,000 / 5,000 = 0.80

And weight:

code-highlight
Z = 1.2(0.06) + 1.4(0.16) + 3.3(0.08) + 0.6(1.39) + 1.0(0.80)
Z = 0.072 + 0.224 + 0.264 + 0.834 + 0.80
Z ≈ 2.19

Acme Widgets lands at about 2.19, which (as we'll see in the next section) is the grey zone. Not screaming distress, not in the clear either. A real analyst would dig into trend: is X3 rising or falling quarter over quarter? Did X4 just collapse because the stock dropped 40%?

What the score tells you

Altman calibrated three zones on his 1968 sample:

Z > 2.99 — Safe zone. Low statistical probability of bankruptcy over the next two years based on these five ratios.
1.81 ≤ Z ≤ 2.99 — Grey zone. Mixed signal. The model can't confidently classify the company either way. Treat the name as "needs more work."
Z < 1.81 — Distress zone. Historically, firms with Z below 1.81 had a high probability of bankruptcy within two years.

The thresholds are not laws of physics. They're empirical cutoffs from a specific dataset (manufacturing firms, mostly mid-20th century). Treat them as ranges, not bright lines. A 2.95 and a 3.05 are essentially the same number even though one is technically "safe" and the other is technically "grey."

Variants for non-public and non-manufacturing firms

The original formula has two well-known siblings.

Z' for private firms. If a company isn't traded, you can't compute X4 with a market cap. Altman re-calibrated the model using book value of equity in place of market value, with different coefficients. The Z' thresholds are different too. Use this when valuing private targets or running credit on a non-listed name.

Z'' for non-manufacturing and emerging-market firms. Asset turnover (X5) varies wildly across industries. A consulting firm has tiny sales-to-assets compared to a retailer, even if both are healthy. Altman's Z'' drops X5 entirely, reweights the remaining four ratios, and re-calibrates the cutoffs. The rough thresholds are Z'' below 1.1 for distress and above 2.6 for safe. This is the version to reach for when evaluating service businesses, tech companies, or names outside developed markets.

Don't use any Z-Score variant on banks, insurers, or other financials. Their balance sheets are structurally different — leverage of 10–15x is normal, and "total liabilities" includes customer deposits or policy reserves. The model was never built for them.

When to use it

The Z-Score earns its keep in a few specific situations:

Quick screen on a watchlist of unfamiliar names. Run Z across 50 candidates and you'll immediately spot the 3–4 that look financially fragile.
Cross-check on a long thesis. Before sizing into a deep-value name, confirm it isn't a value trap with a Z below 1.81.
Trigger for further diligence. If Z drops a full point quarter over quarter, that's a signal — go read the cash flow statement and the notes.
Credit-style monitoring of leveraged holdings. For names with meaningful debt, recomputing Z each quarter is cheap insurance against catching a downgrade cycle late.

It is not a buy or sell signal on its own. A high Z doesn't mean the stock will outperform. A low Z doesn't mean the stock is going to zero next month — plenty of distressed names recover.

Limitations

The Z-Score has real weaknesses, and pretending it doesn't is how people get blown up:

It's backward-looking. All five inputs come from already-reported financial statements. By the time the 10-K shows a problem, the market often already knows.
The calibration is old. Altman's original sample was U.S. manufacturers from 1946–1965. Several decades of academic work have argued the model's accuracy has decayed over time, especially outside manufacturing.
It misses qualitative risk. Pending litigation, a fraud investigation, a key-person dependency, a regulatory shift, a customer concentration risk — none of these show up in the five ratios.
It's blind to off-balance-sheet liabilities. Operating lease accounting has tightened, but unfunded pensions, contingent obligations, and complex financial instruments can still distort the picture.
A "safe zone" score is not a guarantee. It means the model didn't find a statistical bankruptcy signal in these five ratios. That's a much weaker claim than "this company is healthy."

Use Z as one input. Never as the only input.

How to use the Z-Score with Stock Alarm Pro

You don't have to compute the score by hand for every name. The Stock Alarm Pro screener exposes the underlying inputs directly — total assets, total liabilities, retained earnings, working capital proxies, EBIT margins, asset turnover, market cap, and debt-to-assets are all filterable columns. A practical workflow:

Start with a universe of mid- and large-cap names.
Screen out the obvious distress shape: filter for positive retained earnings, debt-to-assets below 0.6, positive operating margin, and a healthy current ratio.
Sort the survivors by the metric you actually care about — valuation, growth, momentum — and dig into the top 10–20.
For any name you're seriously considering, spend the 60 seconds to compute Z by hand from the 10-K. Re-run it next quarter. Watch the trend.

The point isn't to chase a perfect bankruptcy predictor. The point is to make sure you never accidentally own a name whose financials were quietly screaming for two quarters before the market caught on.