How to Use the Altman Z-Score to Avoid Distressed Companies
Edward Altman published his bankruptcy prediction model in 1968, and it has held up better than almost anything else in finance from that era. In his original study, the Z-Score correctly sorted companies that would go bankrupt from those that wouldn't a couple of years ahead of the event, and later work across different markets and downturns has mostly backed that up. For a formula built on five ratios you can pull straight off a filing, that's a genuinely good track record.
Most people use it wrong, though. They look the number up once, nod, and move on. A single snapshot doesn't tell you much. The real value comes from watching the score move over time, knowing what each piece is actually telling you about the business, and pairing it with a couple of other tools so you're not leaning on one model to make a call.
The formula and what each piece means
For a public manufacturing company, the Z-Score is Z = 1.2A + 1.4B + 3.3C + 0.6D + 1.0E. The letters are: A is working capital divided by total assets, B is retained earnings divided by total assets, C is EBIT divided by total assets, D is market value of equity divided by total liabilities, and E is sales divided by total assets.
Each ratio is looking at a different corner of the balance sheet. Working capital to total assets is short-term liquidity. If it's negative, the company is consuming more than it's generating in the near term, which is worth a second look. Retained earnings to total assets is cumulative profitability relative to size, so young companies and serial money-losers score low here almost by definition. EBIT to total assets is operating profitability with taxes and leverage stripped out, which lets you compare the actual business across companies with very different capital structures. Market value of equity to total liabilities is the market's read on solvency. When equity value slips below total liabilities, investors are telling you they doubt the company can cover what it owes. Sales to total assets is asset efficiency, how much revenue the company squeezes out of its asset base.
The reading is simple. Above 2.99 is the safe zone. Between 1.81 and 2.99 is the gray zone, where nothing is on fire but you want to keep an eye on it. Below 1.81 is the distress zone, where the odds of trouble over the next couple of years climb sharply.
Where the inputs live and a quick worked example
Every number you need is in the filings. Total assets, total liabilities, retained earnings, and the pieces of working capital (current assets minus current liabilities) come off the balance sheet in the 10-K or 10-Q. EBIT and sales come off the income statement. Market value of equity is just the current share price times shares outstanding, and you'll find the share count on the cover of the filing or in the equity footnotes. All of it is free on EDGAR if you'd rather go to the source than trust a data vendor's calculated field.
Say a mid-size manufacturer reports total assets of 500, total liabilities of 300, current assets of 180 and current liabilities of 120 (so working capital of 60), retained earnings of 90, EBIT of 40, sales of 450, and a market cap of 250. All numbers are made up to show the mechanics. Plugging in: A = 60/500 = 0.12, B = 90/500 = 0.18, C = 40/500 = 0.08, D = 250/300 = 0.83, E = 450/500 = 0.90. Then Z = 1.2(0.12) + 1.4(0.18) + 3.3(0.08) + 0.6(0.83) + 1.0(0.90) = 0.144 + 0.252 + 0.264 + 0.498 + 0.900, which comes to about 2.06. That's squarely in the gray zone, so nothing screams emergency, but I'd want to see whether last year's score was 2.4 and falling or 1.7 and recovering before I did anything with it.
Why a model this old still works
You'd think a formula calibrated on mid-century data would have aged out by now. It hasn't, because it's measuring things that don't really change. Companies fail when they run out of liquidity, can't earn an operating profit, carry too much debt against their equity value, or use their assets badly. None of that is tied to a particular decade or technology.
The coefficients (1.2, 1.4, 3.3, 0.6, and 1.0) came out of Altman's original sample, and they hold up because the link between these ratios and distress is structural, not a quirk of the data he happened to have. Researchers have re-run the model through later periods, including the 2008 crisis and the COVID shock, and it has kept flagging trouble ahead of time reasonably well.
The variants for non-manufacturers
Altman built several modified versions for companies that don't fit the original mold. The Z'-Score swaps market value of equity for book value in the fourth ratio, which lets you run it on private companies that don't have a market cap. The Z''-Score drops the sales-to-assets term entirely so the model travels across industries with wildly different asset intensity. That's the one to reach for with service businesses, financials, and software, where asset turnover just isn't comparable to a factory.
There are further tweaks for emerging-market companies that fold in country risk and accounting differences. The skeleton stays the same, but the coefficients and the zone boundaries shift to reflect a higher baseline level of risk.
Using it as a screen
The most useful job the Z-Score does is a negative screen. It's there to help you skip disasters rather than to hand you winners. Before you buy anything, run the number. If it lands below 1.81, the company is in the distress zone and you'd better have a specific, well-argued reason to go in anyway.
For things you already own, check the trend every quarter. A score sliding from the safe zone down into the gray zone is an early warning. It doesn't mean sell tomorrow, it means go figure out what's dragging it down. Is working capital shrinking because the company is burning cash? Is equity value falling because the market is pricing in more risk? Are operating profits eroding relative to the asset base? The score tells you where to point the flashlight.
And the direction usually matters more than the level. A company sitting at 2.5 that has dropped four quarters running worries me more than one at 2.0 that's been climbing back. One is healing, the other is coming apart, and the absolute number hides that.
Pair it with a second model
The Z-Score gets a lot stronger next to a couple of complementary tools, because each one covers the others' blind spots.
Pair it with the Piotroski F-Score. Joseph Piotroski's 2000 paper scored companies on nine fundamental signals, and it answers a different question than Altman does. The Z-Score tells you how far from distress a company sits; the F-Score tells you whether the underlying financials are getting better or worse. Together you get position and momentum.
Pair it with the Beneish M-Score for earnings manipulation. A company with a falling Z-Score and an M-Score that's flagging possible manipulation is throwing off two warnings at once, and when a distress signal and a manipulation signal show up together, I take it seriously.
Pair it with free cash flow. The Z-Score leans on accrual figures from the statements, so a cash reality check catches the cases the ratios miss. A company can post a fine Z-Score while quietly burning cash if its accounting profits aren't turning into real money. Watching both keeps you honest.
Where it works and where it breaks
The model is at its best on manufacturing and industrial companies, which is what Altman calibrated it on. For banks, insurers, and REITs, the standard version is unreliable, because these businesses run enormous leverage by design. The plain formula would tag a perfectly healthy bank as distressed simply because of how its balance sheet is built.
For software and other asset-light companies, the asset-based ratios can mislead you the other direction. A healthy SaaS business with strong recurring revenue and fat margins can score poorly just because its asset base is tiny next to its market value. Use the Z''-Score, which drops the asset turnover term, for those.
Retail and consumer names tend to fit the model well. They carry the mix of tangible assets, working-capital swings, and debt that Altman designed it to read, and a string of retail bankruptcies over the past decade showed up as sliding Z-Scores well before the filings hit.
What it can't do
No single model calls bankruptcy with certainty, and this one is no exception. It throws false positives, flagging some companies that never actually fail, and it misses failures too, especially the ones driven by fraud, a sudden market shock, or anything that doesn't show up in trailing financials.
It's also backward-looking by construction, since it runs on reported numbers. A company can carry a strong Z-Score built on last year's statements while the current business quietly falls apart. That's exactly why the quarterly trend matters more than any one reading. What you want to track is the direction of travel over several quarters.
Management can also nudge the inputs, at least for a while. Aggressive revenue recognition props up sales to assets. Drawing down a credit line lifts working capital. Buybacks shrink total assets and can flatter a ratio or two. None of that lasts, but it can delay the warning by a quarter or two, which is another reason to watch the direction rather than trust a single clean number.
Wiring it into your process
If you invest systematically, calculating Z-Scores across your portfolio and watchlist every quarter is worth building in. The inputs come straight from standard financial data providers, and the math is simple enough to sit in a spreadsheet or a short Python script.
Set thresholds so you're not eyeballing it. Anything that drops below 2.5 goes on a watchlist for a closer look. Anything below 1.81 triggers a full review of the thesis. Any new candidate scoring below 2.0 needs a written reason for why you think the distress signal is wrong before it goes in.
The Z-Score won't make you a better stock picker on its own. What it does is keep you out of the ugliest holes, and over a long enough run, dodging the names that go to zero does as much for your returns as catching the ones that double. Protecting the downside is unglamorous, and it's most of the job.