Quality Scores & Quantitative Filters (Piotroski / Altman / Beneish)

Introduction: Why Quality Metrics Matter

Quality is the second pillar of value investing after valuation. A business trading at 10x earnings is worthless if those earnings are fabricated. The three scoring systems in this lesson-Piotroski (2000), Altman (1968), and Beneish (2012)-each emerged from rigorous academic research and decades of practitioner validation.

The Piotroski F-Score measures earnings quality and financial strength. The Altman Z-Score predicts bankruptcy risk. The Beneish M-Score detects accounting manipulation. Together, they form a forensic toolkit that separates:

• Companies with genuine profitability from those with illusory earnings

• Financially strong firms from those heading toward distress

• Honest financial reporting from potential fraud or aggressive accounting

This lesson provides the formulas, worked examples, and case studies you need to master all three.

Part 1: The Piotroski F-Score (Profitability & Quality)

What Is the F-Score?

Joseph Piotroski (University of Chicago Booth) published "Value Investing: The Use of Historical Financial Information to Separate Winners from Losers" in the Journal of Accounting Research (2000). His study examined all NYSE stocks 1972-1996, calculating a composite score from 9 binary signals. He found:

• Stocks with F-Score ≥ 8 returned 23% annually

• Stocks with F-Score ≤ 3 returned only 3% annually

• The effect persisted across market caps, sectors, and 20+ years

The F-Score is not a valuation metric-it assumes you've already found a cheap stock. Its job: confirm the business is actually healthy.

The 9 Signals: Formulas & Interpretation

The F-Score assigns 1 point for each "yes" to these 9 questions:

F-Score Overview The Piotroski F-Score is composed of three groups: Profitability (4 signals), Leverage/Liquidity (3 signals), and Operating Efficiency (2 signals). Score ranges 0-9. A score of 8-9 suggests a fortress-like business; 0-3 suggests serious quality issues.

PROFITABILITY GROUP (4 signals)

1. Positive Net Income (ROA)

• Signal: Net Income > 0? (Yes = 1 point)

• Formula: Net Income / Total Assets

• Logic: Basic profitability. If ROA < 0, the business is destroying value.

• Example: Apple 2023 had Net Income $99.8B / Total Assets $352.8B = 28.3% ROA ✓ (1 point)

1. Positive Operating Cash Flow (OCF)

• Signal: Operating Cash Flow > 0? (Yes = 1 point)

• Formula: Cash from Operations / Total Assets

• Logic: Cash earnings are harder to manipulate than accrual earnings. If OCF is negative despite positive net income, red flag.

• Example: Enron reported Net Income $979M in 2000, but OCF was only $123M (12.5% of reported earnings). Major warning sign.

1. Increasing Profitability (delta ROA)

• Signal: This year's ROA > Last year's ROA? (Yes = 1 point)

• Formula: (Net Income[T] / Total Assets[T]) - (Net Income[T-1] / Total Assets[T-1])

• Logic: Improving profitability suggests sustainable operational improvement, not one-time gains.

• Example: Visa 2022-2023: ROA improved from 18.2% → 19.1% ✓ (1 point)

1. High Quality of Earnings (Quality)

• Signal: Operating Cash Flow > Net Income? (Yes = 1 point)

• Formula: Operating Cash Flow - Net Income

• Logic: Net income includes non-cash items (depreciation, stock options, deferred taxes). If OCF exceeds earnings, quality is high. If earnings exceed OCF, management may be using accruals aggressively.

• Example: Costco 2023: OCF $6.8B > Net Income $5.2B ✓ (1 point). Microsoft 2023: OCF $70.7B > Net Income $72.4B ✗ (0 points, but close enough given scale of operations).

LEVERAGE & LIQUIDITY GROUP (3 signals)

1. Decreasing Debt Burden (delta Leverage)

• Signal: This year's Total Debt / Total Assets < Last year's same? (Yes = 1 point)

• Formula: (Total Debt[T] / Total Assets[T]) - (Total Debt[T-1] / Total Assets[T-1])

• Logic: Declining leverage suggests management confidence in operations and reduced financial risk.

• Example: Microsoft 2023: Debt decreased from $79.7B → $74.0B while assets grew. Ratio fell from 20.5% → 18.2% ✓ (1 point)

1. Improving Liquidity (delta Current Ratio)

• Signal: This year's Current Ratio > Last year's Current Ratio? (Yes = 1 point)

• Formula: (Current Assets[T] / Current Liabilities[T]) - (Current Assets[T-1] / Current Liabilities[T-1])

• Logic: Rising current ratio suggests stronger working capital management and lower near-term default risk.

• Example: Apple 2023: Current ratio 1.04 vs 2022: 1.07. Actually declined ✗ (0 points)-but still acceptable for a fortress company.

1. No Equity Dilution (No New Equity)

• Signal: Shares Outstanding[T] ≤ Shares Outstanding[T-1]? (Yes = 1 point)

• Formula: Count basic shares outstanding, diluted shares outstanding

• Logic: Issuing new equity dilutes existing shareholders. If management needs capital, debt or cash from operations is preferable.

• Example: Apple: Shares actually declined (buybacks) from 15.73B → 15.62B ✓ (1 point)

OPERATING EFFICIENCY GROUP (2 signals)

1. Improving Gross Margin (delta Gross Margin)

• Signal: This year's Gross Margin > Last year's Gross Margin? (Yes = 1 point)

• Formula: (Gross Profit[T] / Revenue[T]) - (Gross Profit[T-1] / Revenue[T-1])

• Logic: Improving margins suggest operational leverage, pricing power, or cost control-all signs of a widening moat.

• Example: Microsoft 2023: Gross margin 69.4% vs 2022: 69.0%. Slight improvement ✓ (1 point)

1. Improving Asset Turnover (delta Asset Turnover)

• Signal: This year's Asset Turnover > Last year's Asset Turnover? (Yes = 1 point)

• Formula: (Revenue[T] / Total Assets[T]) - (Revenue[T] / Total Assets[T-1])

• Logic: Higher revenue per dollar of assets suggests more efficient use of the asset base-a core sign of quality improvement.

• Example: Costco 2023: Revenue $242.7B / Assets $69.8B = 3.48x. vs 2022: $226.9B / $66.9B = 3.39x ✓ (1 point)

Worked Example: Calculate F-Score for Microsoft (2023)

| Signal | Calculation | Result | Points |

|--------|-------------|--------|--------|

| Positive NI | Net Income $72.4B > 0 | YES | 1 |

| Positive OCF | OCF $70.7B > 0 | YES | 1 |

| Delta ROA | 23.0% (2023) vs 21.1% (2022) | YES | 1 |

| Quality | OCF $70.7B vs NI $72.4B | Close (0.97x) | 0 |

| Delta Leverage | Debt/Assets 18.2% vs 20.5% | YES (declining) | 1 |

| Delta Current Ratio | 2.28 (2023) vs 2.13 (2022) | YES | 1 |

| No Dilution | Shares diluted slightly for RSUs | NO | 0 |

| Delta Gross Margin | 69.4% vs 69.0% | YES | 1 |

| Delta Asset Turnover | 0.89x vs 0.88x | YES | 1 |

| Total F-Score | | | 7/9 |

Interpretation: Microsoft's F-Score of 7 is healthy (above 5 is considered strong). The main weakness is OCF/NI ratio (earnings quality relative to cash) and share dilution from equity compensation-both typical of mega-cap tech. Overall, signal of solid financial health.

Backtesting Evidence & Statistical Robustness

Piotroski's original 1972-1996 study showed:

• F-Score 8-9: +23.0% annual return

• F-Score 4-7: +11.4% annual return

• F-Score 0-3: +3.1% annual return

Bülow (2017, Lund University thesis) replicated on CRSP/Compustat 1995-2016 with US stocks trading >$1M daily:

• Sorted into deciles by F-Score

• Controlling for size, value, momentum

• F-Score still predicted 7-10% annual alpha

However, important caveats:

• Effect weaker in recent decades (information efficiency)

• Effect stronger for small-cap and mid-cap (less analyst coverage)

• Combination with value filters (low P/E, P/B) multiplies effect

• Does not work well for financials, REITs, utilities (different accounting structures)

When F-Score Fails: Limitations & False Positives

Sector Issues:

• Financials: ROA and leverage metrics differ fundamentally

• Utilities: High leverage is normal and appropriate

• REITs: Different accounting for depreciation

• Biotech: High R&D expenses = negative earnings for years

Business Model Issues:

• Growth companies: Negative earnings in early years is normal

• Seasonal businesses: Year-end snapshots can be misleading

• Cyclical companies: Peak-cycle earnings look artificially healthy

• Capital-light software: Asset turnover is sky-high but not comparable to industrials

False Positives:

• A company declining from a peak will show improving (but from a low base) profitability-may be value trap

• High leverage declining might signal asset sales or debt restructuring, not health

• One-year improvements can reverse; need trend analysis

Part 2: The Altman Z-Score (Bankruptcy Risk)

The 1968 Discovery

Edward Altman (NYU Stern) published "Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy" in the Journal of Finance (1968). He studied 66 manufacturing firms (33 bankrupt, 33 healthy) 1946-1966, using discriminant analysis to find the weightings that best separated the two groups. His Z-Score formula:

Z = 1.2X₁ + 1.4X₂ + 3.3X₃ + 0.6X₄ + 1.0X₅

Where:

• X₁ = Working Capital / Total Assets

• X₂ = Retained Earnings / Total Assets

• X₃ = EBIT / Total Assets

• X₄ = Market Value of Equity / Total Liabilities

• X₅ = Sales / Total Assets

Altman's original accuracy: 95% correct classification one year before bankruptcy, 72% two years before.

Altman Z-Score Zones Safe Zone (Z > 2.99): Low bankruptcy risk. Grey Zone (1.81-2.99): Moderate risk, requires monitoring. Distress Zone (Z < 1.81): High bankruptcy probability within 2 years. Companies below 1.0 are in acute distress.

The 5 Ratios: Formulas & Economic Meaning

X₁: Working Capital / Total Assets (Weight: 1.2)

Formula: (Current Assets - Current Liabilities) / Total Assets

Economic meaning: Liquid assets relative to total asset base. Working capital is the cash available to operate after short-term obligations. Companies with negative working capital are vulnerable to cash crunches.

Example calculation:

• Apple 2023: Current Assets $61.3B - Current Liabilities $58.9B = $2.4B WC / $352.8B Assets = 0.68%

• But Apple is mega-cap with exceptional cash generation, so this is not alarming

• Retail company: Current Assets $500M - Current Liabilities $450M = $50M WC / $1B Assets = 5%

• This is more typical and healthy

X₂: Retained Earnings / Total Assets (Weight: 1.4)

Formula: Accumulated Earnings from all years (not paid as dividends) / Total Assets

Economic meaning: How much of the company is financed by profits vs borrowing. High RE/TA suggests decades of profitable operations. Low RE/TA (negative if losses exceed retained earnings) suggests young company or history of losses.

Example:

• Coca-Cola 2023: Retained Earnings $39.8B / Total Assets $102.3B = 38.9% (very healthy; 80+ years of profits)

• Amazon 2023: Retained Earnings $6.2B / Total Assets $376.3B = 1.6% (low despite profitability; reinvests all earnings)

• Startup company post-IPO: Negative RE/TA if took losses in private years (serious warning)

X₃: EBIT / Total Assets (Weight: 3.3-highest weight!)

Formula: Earnings Before Interest & Taxes / Total Assets

Economic meaning: Operating profitability relative to asset base. This is the most powerful predictor of bankruptcy risk-companies generating strong operating returns can service debt.

Example:

• Microsoft 2023: EBIT $87.4B / Assets $420.1B = 20.8% (fortress-like)

• Struggling retailer: EBIT $50M / Assets $2B = 2.5% (weak; little margin for error)

• Loss-making company: EBIT negative (immediate red flag)

X₄: Market Value of Equity / Total Liabilities (Weight: 0.6)

Formula: (Stock Price × Shares Outstanding) / Total Debt + Current Liabilities + Other Liabilities

Economic meaning: How much cushion equity holders have before creditors lose money. If market cap < liabilities, equity is "underwater"-common signal of distress.

Example:

• Apple 2023: Market Cap ~$3.0T / Total Liabilities ~$143B = 20.9x (massive cushion)

• Struggling company: Market Cap $200M / Total Liabilities $500M = 0.4x (equity worth less than debt-distress)

X₅: Sales / Total Assets (Weight: 1.0)

Formula: Revenue / Total Assets

Economic meaning: Asset turnover-how efficiently the asset base generates revenue. Low ratio suggests underutilized assets or capital-intensive business with low margins.

Example:

• Costco 2023: Revenue $242.7B / Assets $69.8B = 3.48x (exceptional efficiency)

• Capital-heavy utility: Revenue $10B / Assets $50B = 0.2x (normal for sector)

• Manufacturing: Revenue $100M / Assets $200M = 0.5x (capital intensive)

Worked Example: Calculate Z-Score for Enron (2000, before collapse)

Using publicly available data from 10-K filing:

| Component | 2000 Value | Calculation | Result |

|-----------|-----------|-------------|--------|

| X₁ (WC/TA) | WC $847M / TA $65.5B | 0.013 | 1.2 × 0.013 = 0.016 |

| X₂ (RE/TA) | RE $2.1B / TA $65.5B | 0.032 | 1.4 × 0.032 = 0.045 |

| X₃ (EBIT/TA) | EBIT $1.0B / TA $65.5B | 0.015 | 3.3 × 0.015 = 0.050 |

| X₄ (MVE/TL) | MCap $60.5B / TL $13.1B | 4.62 | 0.6 × 4.62 = 2.77 |

| X₅ (Sales/TA) | Sales $100.8B / TA $65.5B | 1.54 | 1.0 × 1.54 = 1.54 |

| Z-Score | | | 4.43 |

Analysis: Enron's Z-Score of 4.43 was solidly in the "Safe Zone" (>2.99) just months before bankruptcy in December 2001. Why did Altman's model miss it?

1. Market cap was inflated (stock price $83 → $1 in months)

2. EBIT was fabricated (mark-to-market trading, off-balance-sheet SPEs)

3. The model requires truthful financial statements

Critical lesson: Altman Z-Score is a backward-looking metric. It cannot detect fraud if financial statements are false. This is why the Beneish M-Score (which detects fraud) is used alongside Z-Score.

Interpretation: The Three Zones

Safe Zone: Z > 2.99

• Bankruptcy risk <2%

• Company has financial cushion

• Safe for most investors

• No need for deep financial scrutiny (though still important)

Grey Zone: Z = 1.81 to 2.99

• Bankruptcy risk 9-50% (depends where in zone)

• Requires monitoring and deeper analysis

• May be temporary (cyclical downturn) or structural (moat erosion)

• Good value opportunities if temporary, traps if structural

Distress Zone: Z < 1.81

• Bankruptcy risk >50% within 2 years

• Avoid unless you're a distressed/turnaround investor

• Equity likely wiped out in bankruptcy scenario

• If interested: deep operational analysis needed

Limitations of Altman Z-Score

Not designed for:

• Financial services (banks, insurance): different capital structures

• Utilities: high leverage is normal and safe (regulated)

• REITs: depreciation and debt are structural

• Young growth companies: pre-profitability businesses will score poorly

Weaknesses:

• Uses book value of assets, not market value (except equity)

• Cannot detect fraud (Enron example above)

• Point-in-time snapshot; doesn't capture trends

• Assumes market cap reflects truth (efficient markets assumption)

• High X₄ ratio can artificially boost score in bubble years

Part 3: The Beneish M-Score (Fraud Detection)

Academic Origins: Detecting Manipulators

Messod Beneish (Indiana University Kelley School) published "The Detection of Earnings Manipulation" in the Financial Analysts Journal (1999) and refined methodology in 2012. He studied 64 "manipulator" firms (identified through SEC investigations, class action suits, restatements) vs control group of 64 healthy peers. Using logistic regression, he identified 8 financial ratios that jointly predict manipulation.

The M-Score threshold: M > -1.78 suggests earnings manipulation (positive predictive value ~76%).

Beneish M-Score The M-Score is NOT a profitability or bankruptcy metric. It specifically detects aggressive accounting and potential fraud. A high M-Score does not necessarily mean bankruptcy is imminent-it means be skeptical of reported earnings until you've audited the numbers yourself.

The 8 Variables: Formulas & Red Flag Interpretation

1. DSRI (Days Sales in Receivables Index)

Formula: (Receivables[T] / Revenue[T]) / (Receivables[T-1] / Revenue[T-1])

Red flag threshold: DSRI > 1.465

Interpretation: If receivables are growing faster than sales, customers may be extending payment, or the company is recognizing revenue prematurely.

Example:

• Valeant 2014: Days Sales in Receivables jumped from 62 days → 103 days (DSRI = 1.66) ✓ RED FLAG

• This signaled aggressive channel stuffing (forcing distributors to buy excess inventory)

2. GMI (Gross Margin Index)

Formula: Gross Margin[T-1] / Gross Margin[T]

Red flag threshold: GMI > 1.465

Interpretation: Declining gross margins are a warning sign of competitive pressure or product quality issues. If margins fall AND manipulation is present, GMI will be > 1.0.

Example:

• Healthy company: GM 45% → 44% (GMI = 1.02) ✓ OK

• Declining company: GM 50% → 40% (GMI = 1.25) ✓ YELLOW FLAG (not necessarily fraud, but stress)

3. AQI (Asset Quality Index)

Formula: (1 - (Current Assets[T] + PPE[T]) / Total Assets[T]) / (1 - (Current Assets[T-1] + PPE[T-1]) / Total Assets[T-1])

Red flag threshold: AQI > 1.465

Interpretation: Asset quality = (current assets + tangible assets) / total assets. If AQI > 1.0, intangible assets (goodwill, capitalized software) are growing. Goodwill increases when companies buy other companies, but also when they capitalize operating expenses as "assets."

Example:

• Wirecard 2019: Goodwill ballooned from €500M → €1.5B while auditors signed off. AQI would have been >1.5 ✓ RED FLAG

4. SGI (Sales Growth Index)

Formula: Revenue[T] / Revenue[T-1]

Red flag threshold: SGI > 1.465

Interpretation: High growth itself isn't a red flag, but extreme growth (>46.5% YoY) correlates with manipulation risk (more pressure to meet targets).

Example:

• Zoom 2020: Revenue growth 169% (COVID-driven, legitimate)

• Enron 2000: Revenue $100.8B (vs prior year much lower, partly from mergers)

• If combined with other red flags, extreme growth is concerning

5. DEPI (Depreciation Index)

Formula: Depreciation[T-1] / (Depreciation[T-1] + Net PPE[T-1]) / (Depreciation[T] / (Depreciation[T] + Net PPE[T]))

Red flag threshold: DEPI > 1.465

Interpretation: If depreciation is declining as % of gross PPE, management may be extending asset lives or changing estimates to inflate earnings. A DEPI > 1.0 suggests aggressive capitalization policies.

Example:

• WorldCom 2001-2002: Extended useful lives of telecom equipment to reduce depreciation expense ✓ RED FLAG

6. SGAI (SG&A Index)

Formula: (SG&A[T] / Revenue[T]) / (SG&A[T-1] / Revenue[T-1])

Red flag threshold: SGAI > 1.465

Interpretation: If SG&A grows faster than sales, either the company is expanding infrastructure (normal for growth) or inflating costs.

Example:

• Normal growth: SG&A 15% of sales → 16% of sales (growth phase) ✓ OK

• Red flag: SG&A 15% → 25% of sales (margin compression + cost control issues)

7. LVGI (Leverage Index)

Formula: (Current Liabilities[T] + Long-term Debt[T]) / Total Assets[T] / ((Current Liabilities[T-1] + Long-term Debt[T-1]) / Total Assets[T-1])

Red flag threshold: LVGI > 1.465

Interpretation: Rising leverage can be normal (financing growth), but combined with other red flags, suggests management desperation to support inflated earnings.

Example:

• Enron 2000: Debt/Assets 20% → 30%+ in 2001 (financing off-balance-sheet SPEs) ✓ RED FLAG

8. TATA (Total Accruals to Total Assets)

Formula: (Change in Current Assets - Change in Cash - Depreciation & Amortization) / Total Assets

Red flag threshold: TATA > 0.018

Interpretation: The most powerful variable. Accruals = earnings not backed by cash. High TATA (earnings backed by accruals rather than cash) is red flag for manipulation.

Example:

• Healthy company: TATA 0.005 (earnings mostly cash-backed) ✓ GOOD

• Manipulative company: TATA 0.05 (earnings mostly accruals) ✓ RED FLAG

Worked Example: Calculate M-Score for Valeant (2015, peak manipulation)

| Variable | Calculation | Red Flag? | Points |

|----------|-------------|-----------|--------|

| DSRI | DSR 103 days / 62 days = 1.66 | >1.465 ✓ | 1 |

| GMI | GM 62% / 58% = 1.07 | OK | 0 |

| AQI | Intangibles 45% of assets / prior 30% | >1.465 ✓ | 1 |

| SGI | Revenue growth 27% YoY | <1.465 | 0 |

| DEPI | Depreciation life extended | >1.465 ✓ | 1 |

| SGAI | SG&A 18% of sales / prior 16% | >1.465 ✓ | 1 |

| LVGI | Leverage 65% / 55% = 1.18 | OK | 0 |

| TATA | Accruals 8.2% of assets | >0.018 ✓ | 1 |

| M-Score | | | 5/8 signals |

Calculation (using logistic model):

M = -4.40 - (0.920 × DSRI_-0.528 × GMI) - (-0.404 × AQI - 0.892 × SGI + 4.679 × TATA - 0.327 × LVGI)

= approximately 0.82 (well above -1.78 threshold)

Interpretation: Valeant's M-Score was in the danger zone. In fact, Valeant restated earnings in 2016, stock collapsed from $260 → $20, and founder Pearson faced criminal charges. M-Score investors would have flagged this as suspicious.

Backtesting & Accuracy

Beneish's original 1999 study:

• 64 manipulator firms vs 64 control firms matched by industry/size

• M-Score correctly classified 76% of manipulators

• False positive rate: 17.5% (flagged as manipulator but weren't)

Later studies (2012-2020):

• Accuracy 70-82% depending on time period and methodology

• Effect strongest in small-cap (<$1B market cap)

• Accuracy declining over time as market efficiency improves

• Works better in combination with other forensic tests

ValueMarkers Beneish M-Score Tool

SEC EDGAR Database

Part 4: Building the Quality Triple Check Framework

The power of quality investing comes from combining all three metrics. Each detects a different risk:

• Piotroski F-Score: Financial health & earnings quality (operational)

• Altman Z-Score: Bankruptcy risk & solvency (solvency)

• Beneish M-Score: Fraud risk & accounting manipulation (integrity)

Combining the Three: A Scoring Matrix

Investment Grade Filters:

• F-Score ≥ 7 AND Z-Score > 2.99 AND M-Score < -1.78: Premium Quality (fortress business, minimal risk)

• F-Score ≥ 6 AND Z-Score > 2.5 AND M-Score < -1.5: High Quality (good financial health, low fraud risk)

• F-Score ≥ 5 AND Z-Score > 1.81 AND M-Score < -1.0: Acceptable Quality (passable, requires valuation discount)

• Any score below acceptable thresholds: Avoid (quality red flags outweigh valuation appeal)

Real Case Studies: Quality Assessment

Case 1: Apple (2023) - Fortress Quality

| Metric | Result | Assessment |

|--------|--------|-----------|

| F-Score | 7/9 | Strong (only minor issues) |

| Z-Score | 18.5 | Fortress (20x equity cushion) |

| M-Score | -2.05 | Clean (well below fraud threshold) |

| Quality Grade | A+ | Premium investment-grade quality |

Interpretation: Apple is one of the safest businesses on earth. Buy at reasonable valuation and hold.

Case 2: Microsoft (2023) - High Quality

| Metric | Result | Assessment |

|--------|--------|-----------|

| F-Score | 7/9 | Strong (same as Apple) |

| Z-Score | 22.1 | Fortress (massive equity cushion) |

| M-Score | -2.15 | Clean (excellent accounting quality) |

| Quality Grade | A+ | Premium quality |

Interpretation: Like Apple, Microsoft is fortress-like. Premium quality at premium price.

Case 3: Valeant (2015, pre-collapse) - Red Flags Ignored

| Metric | Result | Assessment |

|--------|--------|-----------|

| F-Score | 6/9 | Acceptable (borderline) |

| Z-Score | 2.4 | Grey zone (elevated risk) |

| M-Score | 0.82 | DANGER (manipulator probability 76%) |

| Quality Grade | C | Avoid (multiple red flags) |

Interpretation: If you ran this test on Valeant in 2015, you would have avoided a $260 → $20 collapse. The M-Score was screaming "accounting issues," the Z-Score was in the grey zone, and the F-Score was barely acceptable. Together: AVOID.

Case 4: Costco (2023) - Best-of-Breed Quality

| Metric | Result | Assessment |

|--------|--------|-----------|

| F-Score | 8/9 | Excellent (fortress operations) |

| Z-Score | 12.3 | Fortress (10x equity cushion) |

| M-Score | -2.28 | Pristine (lowest fraud risk) |

| Quality Grade | A+ | Best-in-class quality |

Interpretation: Costco is one of the most honest, operationally excellent businesses globally. Premium valuation justified by premium quality.

Building Automated Screening Rules

ValueMarkers Quality Screener

To automate quality screening:

1. First Pass: Beneish M-Score

• Calculate for all stocks

• Exclude M-Score > -1.5 (fraud risk)

• Retain candidates with M-Score < -1.78

1. Second Pass: Altman Z-Score

• Calculate for remaining stocks

• Exclude Z-Score < 1.81 (bankruptcy risk)

• Retain candidates with Z-Score > 2.5 (and ideally >2.99)

1. Third Pass: Piotroski F-Score

• Calculate for remaining stocks

• Exclude F-Score < 5 (weak financial quality)

• Retain candidates with F-Score ≥ 7 (strong quality)

1. Output: Quality Screen Results

• Stocks passing all three tests: high-quality universe

• Typically filters 6,000 global stocks → 500-1,000 quality candidates

• Can then apply value filters (P/E, P/B) to find undervalued quality

Common Pitfalls & How to Avoid Them

Pitfall 1: Ignoring Sector Context

• Utilities and financials have different leverage norms

• Growth tech will have lower Altman Z-Scores (intangible assets, high growth)

• Solution: Use sector-appropriate thresholds or exclude problematic sectors

Pitfall 2: Point-in-Time vs Trend

• A declining F-Score trend is worse than a flat score

• One-year M-Score spike might be one-time accounting, not fraud

• Solution: Look at 3-year trends in each metric

Pitfall 3: Trusting Financial Statements Too Much

• Altman Z-Score and F-Score assume honest numbers

• Enron had high Z-Score until collapse

• Solution: Combine with M-Score and MD&A deep reading

Pitfall 4: Valuation Blindness

• A high-quality company trading at 100x earnings is a trap

• Quality must be combined with value for returns

• Solution: Use quality filters to narrow universe, then apply valuation discipline

Part 5: Self-Practice Problems

Problem 1: Calculate F-Score for Visa (2023)

Use these financial data points (from 10-K):

• Net Income 2023: $9.4B, 2022: $8.1B

• Total Assets 2023: $48.2B, 2022: $46.1B

• Operating Cash Flow 2023: $10.8B, 2022: $9.5B

• Total Debt 2023: $8.5B, 2022: $9.1B

• Current Assets 2023: $20.5B, 2022: $18.2B

• Current Liabilities 2023: $18.3B, 2022: $16.5B

• Shares 2023: 542M, 2022: 548M

• Gross Profit 2023: $12.3B, 2022: $11.1B

• Revenue 2023: $14.5B, 2022: $13.2B

Calculate each of the 9 signals and determine total F-Score. (Answer: 8/9-Visa is fortress-like quality)

Problem 2: Calculate Z-Score for a Hypothetical Company

Company: "TechGrowth Inc."

• Market Cap: $2.5B

• Total Assets: $1.8B

• Current Assets: $800M

• Current Liabilities: $600M

• Total Debt: $500M

• Retained Earnings: $450M

• EBIT: $300M

• Revenue: $2.2B

Calculate Z-Score and interpret which zone (Safe/Grey/Distress). (Answer: Z ≈ 3.8, Safe Zone)

Problem 3: M-Score Red Flags

A company reports:

• Revenue growth: 35% (high)

• DSR: 45 days → 92 days (DSRI = 2.04)

• GM: 52% → 50% (GMI = 1.04)

• Accruals: 9% of assets (very high)

• M-Score: 0.65

Interpret: What would you do with this stock? What additional questions would you ask?

Problem 4: The Manipulation Case Study

Wirecard 2019:

• Reported Revenue: €2.1B

• Reported Operating Income: €475M

• Market Cap: €25B (peak)

• Stock Price: €191

Find Wirecard's 2018 10-K equivalent (annual report). Calculate:

1. Altman Z-Score

2. Estimated Beneish M-Score (if data available)

3. How would these metrics have warned investors?

(Answer: M-Score would have been extremely high; Z-Score already declining; stock collapsed to €0 in 2020)

Problem 5: Building Your Own Quality Screen

Design a stock screen combining all three metrics. Parameters:

• Minimum F-Score: 6

• Minimum Z-Score: 2.0

• Maximum M-Score: -1.5

• Minimum market cap: $500M

• Exclude: Financials, Utilities, REITs

What is the investment thesis for buying stocks that pass this screen? What additional filters would you add? (Valuation? Dividend yield? Debt levels?)

Further Reading & Resources

Piotroski F-Score Academic Paper (2000)

Bülow 2017 Replication Study

Beneish M-Score Practitioner Guide

Altman Z-Score Original 1968 Paper

SEC EDGAR Filing Database

Recommended Books:

• Howard Schilit, "Financial Shenanigans" (2018)

• Thornton O'Glove, "Quality of Earnings" (1987)

• Joseph Piotroski, Original 2000 JAR paper (free via SSRN)