Altman Z-Score in Excel: Predict Bankruptcy Risk (2026)

July 7, 2026 · VeloraAI Team
Data Analysis Financial Modeling Formulas

Edward Altman's 1968 model correctly flagged roughly 80–90% of bankrupt U.S. manufacturers a full year before they filed — with a false-positive rate under 20%. That's not a legacy statistic; the Altman Z-Score in Excel still shows up in bank credit memos, distressed-debt screens, and turnaround diligence packs in 2026, because five ratios and one weighted sum still beat most machine-learning classifiers on interpretability. This guide walks through building the original Z-Score, plus the Z' and Z'' variants for private firms and non-manufacturers, in a single Excel workbook you can point at any 10-K.

What Is the Altman Z-Score?

The Altman Z-Score is a linear discriminant model that combines five financial ratios into a single number designed to separate healthy firms from those two years away from bankruptcy. Higher scores mean lower distress risk. The score is not a probability — it's a distance measure between a company and Altman's 1968 sample of failed vs. surviving manufacturers.

The original model was estimated on 66 firms (33 bankrupt, 33 solvent) and validated on out-of-sample data. Its enduring value is that every input comes straight from published financial statements, so you can score any public company in five minutes with only a balance sheet, income statement, and closing share price.

ℹ️ Note: The Z-Score predicts financial distress, not equity returns. A company can have a Z-Score of 4.5 and still be a bad stock — the model has nothing to say about valuation, only about balance-sheet resilience.

The Five Ratios (X1 to X5)

Each variable measures a different dimension of financial health:

Ratio Formula What It Captures
X1 Working Capital / Total Assets Short-term liquidity cushion
X2 Retained Earnings / Total Assets Cumulative profitability over time
X3 EBIT / Total Assets Operating productivity of assets
X4 Market Value of Equity / Total Liabilities Market's confidence in solvency
X5 Sales / Total Assets Asset turnover efficiency

X3 is the workhorse. It carries the highest coefficient (3.3) in the original model because operating earnings relative to the asset base is the single strongest historical predictor of who survives.

How Do You Calculate the Altman Z-Score in Excel?

Enter the five inputs into named cells, compute each ratio in its own cell, then multiply by the coefficients and sum. The original formula is Z = 1.2×X1 + 1.4×X2 + 3.3×X3 + 0.6×X4 + 0.999×X5. In Excel, that becomes a one-line SUMPRODUCT across a ratios column and a coefficients column — reusable across the three model variants.

Step 1: Set Up the Input Block

Pull these seven line items from the target company's most recent 10-K or 10-Q. Put them in a labeled input block on a tab called Inputs:

Current Assets           (BS - Current Assets)
Current Liabilities      (BS - Current Liabilities)
Total Assets             (BS - Total Assets)
Total Liabilities        (BS - Total Liabilities)
Retained Earnings        (BS - Retained Earnings)
EBIT                     (IS - Operating Income)
Sales                    (IS - Net Revenue)
Shares Outstanding       (Cover page or notes)
Share Price              (Latest close)

Name these cells (Formulas → Define Name) so downstream formulas read as EBIT / TotalAssets instead of E12/E8. This is not cosmetic — named ranges cut audit time on a 15-cell model roughly in half.

Step 2: Compute the Five Ratios

Build a small Ratios block with each X value in its own cell:

X1 = (CurrentAssets - CurrentLiabilities) / TotalAssets
X2 = RetainedEarnings / TotalAssets
X3 = EBIT / TotalAssets
X4 = (SharesOutstanding * SharePrice) / TotalLiabilities
X5 = Sales / TotalAssets

For X4, market cap goes on top because Altman found market pricing captures information that book equity misses — investors bake in expected future losses long before write-downs hit the balance sheet.

💡 Pro Tip: Wrap each ratio in IFERROR(..., NA()) rather than returning zero on a divide-by-zero. A zero silently inflates the Z-Score toward "safe"; #N/A propagates up and forces you to look.

Step 3: Multiply and Sum

Two ways to write the final formula. The literal version:

=1.2*X1 + 1.4*X2 + 3.3*X3 + 0.6*X4 + 0.999*X5

The reusable version, with the five ratios in B2:B6 and the coefficients in C2:C6:

=SUMPRODUCT(B2:B6, C2:C6)

The SUMPRODUCT form is what you want when you're computing all three variants side by side, because swapping the coefficient column gives you Z, Z', or Z'' without rewriting the formula.

Example: Apple's fiscal 2025 filings hypothetically give X1 = 0.05, X2 = 0.20, X3 = 0.32, X4 = 12.5, X5 = 0.85. Original Z = 1.2(0.05) + 1.4(0.20) + 3.3(0.32) + 0.6(12.5) + 0.999(0.85) = 9.75 — well into the safe zone, driven overwhelmingly by X4 (equity value vastly exceeds liabilities).

Step 4: Add a Zone Classifier

Turn the raw number into a verdict with a nested IFS:

=IFS(Z >= 2.99, "Safe",
     Z >= 1.81, "Grey",
     TRUE,      "Distress")

Pair it with conditional formatting: green fill for Safe, amber for Grey, red for Distress. That single visual is the difference between a Z-Score cell that gets read and one that gets ignored on page 47 of the credit memo.

Interpreting the Score: The Three Zones

Once you have a Z, drop it into Altman's three-zone framework. The zones come from the original 1968 study and have been re-validated across dozens of subsequent papers.

graph TD
    A[Altman Z-Score Result] --> B{Z Value}
    B -->|Z > 2.99| C[Safe Zone]
    B -->|1.81 <= Z <= 2.99| D[Grey Zone]
    B -->|Z < 1.81| E[Distress Zone]
    C --> F[Bankruptcy unlikely in 24 months]
    D --> G[Elevated risk - monitor closely]
    E --> H[High distress probability - deeper diligence]
    G --> I[Recompute quarterly]
    H --> J[Stress test debt covenants]

Safe Zone (Z > 2.99). Historical bankruptcy rate within 24 months is under 5% for firms in this range. This does not mean the company is a good investment — just that the balance sheet is unlikely to collapse.

Grey Zone (1.81 ≤ Z ≤ 2.99). This is the "type II error" band where Altman's model makes most of its mistakes. Roughly 15–20% of grey-zone firms have historically ended up bankrupt; the other 80% stabilize or recover. Treat a Grey score as a trigger for deeper diligence — pull the debt schedule, check covenant headroom, model a revenue shock.

Distress Zone (Z < 1.81). In the original sample, 95% of firms with Z below 1.81 filed for bankruptcy within two years. That's the headline stat everyone quotes. In modern samples the rate is lower (closer to 50–70%) because the coefficients were fit to 1960s manufacturers, but a sub-1.81 Z is still a serious red flag that warrants stress-testing every debt covenant and refinancing wall.

⚠️ Warning: Do not use the original Z-Score cutoffs on private firms or non-manufacturers. The zones shift — a "Safe" Z of 3.0 on the original model is actually Grey on the Z' model. Match the variant to the company type before you look at the number.

Which Altman Variant Should You Use?

Altman published three official versions of the model, each fit to a different corporate population. Using the wrong variant is the single most common analyst error with this framework.

Variant Population Formula Distress / Safe Cutoffs
Z-Score Public manufacturers 1.2 X1 + 1.4 X2 + 3.3 X3 + 0.6 X4 + 0.999 X5 < 1.81 / > 2.99
Z'-Score Private manufacturers 0.717 X1 + 0.847 X2 + 3.107 X3 + 0.420 X4 + 0.998 X5 < 1.23 / > 2.90
Z''-Score Non-manufacturers, emerging markets 6.56 X1 + 3.26 X2 + 6.72 X3 + 1.05 X4 < 1.10 / > 2.60

Z' for Private Companies

Private firms have no market price, so X4 has to switch to book value: X4 = Book Value of Equity / Total Liabilities. Altman re-estimated the coefficients on private-firm data and published Z' in 1983. In Excel, this is the cleanest change to make — swap one input, swap the coefficient column, and read a different cutoff. Everything else stays.

Z'' for Non-Manufacturers and Emerging Markets

Non-manufacturers have wildly different asset turnover — a bank's Sales/Assets ratio is not comparable to a retailer's — so Altman dropped X5 entirely and reweighted the remaining four variables. Z'' is also the version used for emerging-market sovereigns and corporates, sometimes with a +3.25 constant to align to Altman's calibration:

Z'' (EM) = 3.25 + 6.56*X1 + 3.26*X2 + 6.72*X3 + 1.05*X4

X4 for Z'' uses book value of equity / total liabilities, same as Z'.

💡 Pro Tip: If you're screening a portfolio that spans manufacturers and non-manufacturers, compute all three variants for every name and label which one is the "correct" reading. It takes 30 extra seconds and eliminates the entire class of misapplied-model errors.

Building a Screening Model for a Portfolio

The Z-Score's real leverage shows up when you run it across dozens of companies at once — a distressed-debt screen, a credit-portfolio review, or an equity long-short book's short list.

Structure the Workbook

Three tabs is enough for most workflows:

  1. Inputs — one row per company, columns for the seven raw line items plus ticker, sector, and reporting date
  2. Ratios & Score — computed X1–X5, then Z, Z', and Z'' side by side, plus the "correct variant" column
  3. Dashboard — sorted table of all companies with zone flags, conditional formatting, and a distress-count summary

Pull Data Efficiently

For public companies you have three practical options:

  • Manual entry from 10-Ks — highest quality, best for 5–20 names
  • Power Query from XBRL — the SEC's EDGAR API returns structured facts; pulls all seven inputs in one query
  • API from a data provider — FactSet, Bloomberg, or Refinitiv terminals push the ratios straight into Excel

The XBRL / Power Query route is the sweet spot for most analysts. A 300-line M script populates the entire Inputs tab from EDGAR in under a minute, and the refresh runs on the same cadence as quarterly filings.

Flag Watchlist Candidates

The most useful screen isn't the Distress zone itself — the market already knows about those names — it's firms that crossed from Safe into Grey during the last quarter. Two additional columns catch it:

PriorZ  = [prior-quarter Z-Score]
Trend   = IF(Z - PriorZ < -0.5, "Deteriorating",
          IF(Z - PriorZ > 0.5,  "Improving", "Stable"))

A firm dropping 0.5+ points quarter-over-quarter within Safe territory is where the model still finds signal that isn't in consensus estimates.

Common Mistakes That Break the Model

Five errors show up repeatedly in analyst spreadsheets:

1. Using EBITDA Instead of EBIT

X3 is defined as EBIT / Total Assets. Substituting EBITDA inflates X3, and because the coefficient on X3 is 3.3, that error compounds fastest. Depreciation is real; the model wants it deducted.

2. Book Value of Equity in the Original Z

X4 in the original Z uses market cap, not book equity. Analysts trained on private-company modeling default to book value and get a Z that's systematically low for stable large-caps and systematically high for stressed small-caps. If you're computing the original Z, pull the share price.

3. Trailing-Twelve-Month vs. Annual Mixing

Sales in X5 should match the period of Total Assets. Either use fiscal-year Sales and year-end Total Assets, or use TTM Sales and average Total Assets across the same period. Mixing TTM revenue with a point-in-time balance sheet distorts asset turnover.

4. Applying Original Z to Banks and Insurers

Financials have leverage ratios that make X4 look catastrophically bad and asset-turnover ratios that make X5 look catastrophically low — even for a healthy bank. Original Z is meaningless for financial institutions. Use Z'' if you must use an Altman model, but bank-specific frameworks like the Texas Ratio or CAMELS give better signal.

5. Ignoring Off-Balance-Sheet Debt

Post-ASC 842, operating leases are on the balance sheet. Pre-2019 comparisons and some private-company filings still exclude them. If lease liabilities are missing from Total Liabilities, X4 is overstated. For consistency, capitalize leases before computing the ratios.

⚠️ Warning: If a company reported goodwill impairments or restructuring charges in the current year, EBIT can go negative even for a going concern. A single-year negative EBIT will drag the Z-Score into the Distress zone even when a normalized three-year average would land in Grey or Safe. Compute both and note the difference.

When Should You Use Altman Z-Score in Financial Analysis?

Use the Z-Score as a first-pass distress screen on public non-financial companies, particularly when you need a defensible single number that any credit committee or investment memo reader will recognize. It's ideal for initial portfolio triage, distressed-debt watchlists, covenant early-warning systems, and academic or teaching contexts.

Skip it — or supplement it heavily — when you're scoring banks, insurers, or asset-light software companies whose economics diverge from Altman's manufacturer sample. And never use Z as the only input to a real credit decision; pair it with debt schedule stress tests, liquidity runway analysis, and management commentary from the last two earnings calls.

Where Z-Score Adds Real Value

  • Distressed-debt screening. Rank the S&P 500 by Z and pull the bottom decile for deeper diligence
  • Credit committee packages. A single number that references a well-known academic framework is easier to defend than a bespoke internal model
  • Covenant early warning. Recompute quarterly; a two-quarter downward trend within Grey often precedes a covenant violation
  • Teaching cases. The model is transparent enough to teach undergraduates and rigorous enough to publish

Where It Falls Short

  • Financial institutions (see above)
  • Asset-light firms where Total Assets is a poor denominator — software, consulting, staffing
  • High-growth startups where retained earnings are structurally negative regardless of quality
  • Post-restructuring firms where cost cuts haven't hit the reported numbers yet

Alternatives and Complementary Models

The Z-Score is not the only game in town. A serious credit workflow triangulates:

Model Best Use Key Trade-off
Altman Z-Score Public manufacturers Legacy calibration; simple, defensible
Ohlson O-Score Broader coverage, probability output Nine variables; less intuitive
Merton Model Market-based distance to default Requires equity volatility estimate
Piotroski F-Score Value investing, quality overlay Not a distress model per se
Zmijewski Probit Bankruptcy probability estimate Sample-selection bias in original fit

If you're building a serious internal distress scorecard, run Z, Ohlson O, and Merton in parallel and flag disagreements. When all three agree the firm is in distress, the signal is loud. When they disagree, the interesting story is why — usually a market-versus-accounting divergence that deserves its own memo.

Frequently Asked Questions

What is a good Altman Z-Score?

For public manufacturers, a Z above 2.99 is considered safe and bankruptcy is unlikely within 24 months. Above 3.5 is comfortably strong; above 5.0 is exceptional and typically limited to firms with large cash balances and low leverage. Anything between 1.81 and 2.99 is the Grey Zone — not immediate danger, but warrants a look at debt maturity walls and covenant headroom.

Is Altman Z-Score still reliable in 2026?

The five ratios and the ordinal ranking remain useful, but the exact cutoffs have drifted. Recent studies show original Z now flags about 60–70% of bankruptcies one year out, versus 95% in the 1968 sample. Treat the score as a ranking tool across peers rather than an absolute probability. For calibrated bankruptcy probabilities, layer Ohlson O-Score or a Merton distance-to-default on top.

Can Altman Z-Score be used for private companies?

Yes — use the Z'-Score with coefficients 0.717, 0.847, 3.107, 0.420, and 0.998, and replace market cap in X4 with book value of equity. The cutoffs shift: Safe is above 2.9, Distress is below 1.23. Do not apply the original Z coefficients to private firms; the model was refit specifically because private-company data behaves differently.

How often should I recalculate the Altman Z-Score?

Recompute at every quarterly filing — 10-Q for U.S. public companies, half-year for most international filers. For distressed names or credit-monitored positions, recompute monthly using preliminary flash results plus current market cap. The most predictive signal isn't the level of the Z-Score but its trajectory: two consecutive quarters of a 0.5+ drop is a stronger warning than a single-quarter Distress reading.

Does the Altman Z-Score work for tech companies?

Original Z was fit to 1960s manufacturers and struggles with modern tech firms: asset turnover is structurally low, retained earnings are often negative for years, and much of the value sits in intangibles that never hit the balance sheet. Use Z'' for tech companies — it drops the Sales/Assets ratio and reweights the remaining four variables — and cross-check with a Merton-style distance-to-default that captures equity market signals.

Putting It Into Production

Once the workbook is built, the real value is running it every quarter without re-typing anything. A refreshable Power Query connection to EDGAR, a coefficient table you can swap in seconds, and a conditional-formatted dashboard turn a one-off calculation into a live credit-monitoring tool. Analysts using VeloraAI often layer natural-language prompts on top — asking "which portfolio names moved from Safe to Grey this quarter, and what drove the change in X3?" — to surface the two or three names worth a full re-read of the 10-Q.

Build the Z-Score once, wire it to your data source, and it'll pay back the hour of setup every earnings season for as long as the portfolio exists.