Altman Z / Piotroski F / Beneish M — Three Scores for Fundamentals & Integrity
4-section structure: Concept / How We Compute / How to Read / Caveats.
1. Concept
A company's "quality" decomposes into three questions:
| Question | Tool | Author / Year |
|---|---|---|
| Will it go bankrupt? | Altman Z-Score | Edward Altman, 1968 |
| Are fundamentals strong? | Piotroski F-Score | Joseph Piotroski, 2000 |
| Are the financials honest? | Beneish M-Score | Messod Beneish, 1999 |
These three are complementary, not substitutes. Even financially strong companies that manipulate financials eventually blow up (Enron, Wirecard). Honest but deteriorating companies still lose money long-term.
Only companies passing all three are truly worth holding.
2. How We Compute
2.1 Altman Z-Score
Manufacturing 5-factor version:
Z = 1.2·A + 1.4·B + 3.3·C + 0.6·D + 1.0·E
| Variable | Meaning |
|---|---|
| A | (Current Assets − Current Liabilities) / Total Assets |
| B | Retained Earnings / Total Assets |
| C | EBIT / Total Assets (operating_income proxy) |
| D | Market Equity Value / Total Liabilities (shares × latest close) |
| E | Annual Revenue / Total Assets |
Thresholds: Z > 2.99 🟢 Safe · 1.81–2.99 🟡 Grey · < 1.81 🔴 Distressed
2.2 Piotroski F-Score
9 binary checks, max 9:
Profitability (4): ROA>0 · CFO>0 · ROA improving · CFO>NI Structure (3): LT-debt ratio decreasing · Current ratio improving · No new shares Efficiency (2): Gross margin improving · Asset turnover improving
Thresholds: 8-9 🟢 Excellent · 6-7 Good · 4-5 🟡 Average · 0-3 🔴 Poor
2.3 Beneish M-Score
M = −4.84 + 0.92·DSRI + 0.528·GMI + 0.404·AQI + 0.892·SGI + 0.115·DEPI − 0.172·SGAI + 4.679·TATA − 0.327·LVGI
Threshold: M > −1.78 🔴 possible manipulation · M ≤ −1.78 🟢 unlikely
2.4 Data Handling
- All scores use annual data (Q4 = full year)
- 5-year trend shown; each year computed independently (Altman uses latest MV for simplicity)
- TW IS has no standalone depreciation column → Beneish DEPI uses neutral 1.0
- TW long-term debt approximated as
total_liabilities − current_liabilities
3. How to Read
Combined Reading Matrix
| Altman | Piotroski | Beneish | Verdict |
|---|---|---|---|
| 🟢 | 🟢 | 🟢 | Truly quality — core holding candidate |
| 🟢 | 🟡/🔴 | 🟢 | Financially OK, fundamentals deteriorating — mature company at cycle bottom? |
| 🟡/🔴 | 🟢 | 🟢 | Growing but financially stretched — high-leverage startup, watch CFO |
| 🟢 | 🟢 | 🔴 | Looks perfect but has accounting red flags — beware |
| 🔴 | 🔴 | — | Stay away |
Ring Visuals
- Altman ring: Z mapped 0–4 → 0–100% (fuller = better)
- Piotroski ring: score / 9
- Beneish ring: M in [-4, 0] → lower is better → fuller = better
5-Year Trend
- Downward trend is worse than a single low score
- Piotroski < 5 for 3+ consecutive years often indicates structural decline
- Beneish spikes above −1.78 — investigate that year's revenue/AR spikes
4. Caveats
⚠️ Altman Doesn't Apply to Financials
Original 1968 sample was US manufacturing. Banks, insurance, leasing, REITs have fundamentally different capital structures — applying Altman directly will be wrong. For financial stocks, ignore the Z and rely on the other two.
⚠️ Altman's Market-Value Behavior
Variable D uses market equity value (shares × current price). Benefit: reflects current market view. Drawback:
- Price crash → D drops → Z drops → flagged "distressed"
- But this may be market panic, not fundamental deterioration
Cross-check with Piotroski / Beneish.
⚠️ Piotroski F7 Share-Count Tolerance
F7 "no new shares" uses current ≤ prior × 1.001 (0.1% tolerance) because:
- Stock dividends, options exercises, CB conversions cause small increases
- Strict
≤would penalize healthy companies constantly
Significant equity issuance (rights offering, placement) still fails F7.
⚠️ Beneish's High-Growth Trap
SGI (revenue growth) has a positive coefficient. Meaning: faster revenue growth → higher M → more "manipulation-looking".
This is a statistical regularity, but misclassifies genuine high-growth companies (early Amazon/Tesla/semiconductor startups). TSMC during 2020–2022 CapEx boom may show elevated M.
When M is grey, cross-check with:
- Piotroski still high?
- Revenue growth matched by CFO growth?
- Is TATA (accruals) elevated?
Only if multiple signals align is it more likely true manipulation.
⚠️ Honest Accuracy Disclosure
- Altman Z: original paper ~80–90% accuracy for 1-year bankruptcy (manufacturing), ~72% for 2-year. Fails on startups, transformations, services.
- Piotroski F: not a predictive model but a descriptive score. Historically, F ≥ 8 outperforms F ≤ 2 significantly in subsequent returns.
- Beneish M: 1999 paper ~76% accuracy. Enron 1998 had M = −2.74 (missed it); honest high-growth firms get false positives.
No single score can judge a company with 100% certainty. Three together reduce error, not eliminate it.
⚠️ Data Lag
- Annual filing delays: TW Q4 due by Mar 31; US 10-K within 60–90 days of fiscal year-end
- "Latest year" on the dashboard has a 2–5 month delay
- Not suitable for short-term trading — use for long-term screening
Further Reading
- Income Statement, Balance Sheet & Cash Flow: All 3 Statements Explained
- Financial Rankings: Find the Best Companies by ROE & EPS
- What Is DCF?
Try It
- Open Stock Analysis → Financial Ratios, scroll to the bottom
- Click any card's "▼ Expand" to see variable breakdowns
- Switch the trend chart between the three scores
- Click 📐 to view full formulas and platform parameters