VMCI Scoring Methodology

VMCI is a weighted sum of 5 pillar scores: VALUE (35%) + QUALITY (30%) + INTEGRITY (15%) + GROWTH (12%) + RISK (8%) = 100%. Each pillar is itself an average of all its indicators (percentile-ranked 0-100). So VMCI = 0.35*VALUE + 0.30*QUALITY + 0.15*INTEGRITY + 0.12*GROWTH + 0.08*RISK. Final score is 0-100. Scores 70+ are excellent; 50-70 are good; 25-50 are fair; <25 are poor.

Each indicator is ranked against all stocks in the universe. If a stock's P/E is lower than 85% of other stocks, it gets a P/E percentile score of 85. This is more robust than absolute values because P/E of 10 is cheap for a tech stock but expensive for a utility. Percentile ranking normalizes comparisons across industries and geographies.

VALUE gets 35% because valuation is the foundation of value investing-buying low is the entry point. QUALITY gets 30% because only quality companies at low valuations compound wealth; cheap junk stays cheap. INTEGRITY gets 15% to filter out balance sheet disasters. GROWTH gets 12% because we want stocks that compound. RISK gets 8% as a tiebreaker. These weights reflect the philosophy of Ben Graham + Warren Buffett: price (35%) + quality (30%) + safety (15%) + growth (12%) + risk (8%).

Three complementary integrity tests: (1) Piotroski F-Score (9-point, checks 9 financial health signals), (2) Altman Z-Score (bankruptcy risk, 0-10 scale), (3) Beneish M-Score (earnings manipulation risk, -2 to +2). A stock passing all three (F-Score 8-9, Z-Score > 2.99, M-Score < -1.78) has passed "the greatest financial quality screen in value investing." Failing any one is a red flag.

A simpler, game-like scoring for beginners: Price Test (40%) asks "Is it cheap?" using P/E, P/B, and valuations. Quality Test (40%) asks "Is the business good?" using ROE, margins, and cash flow. Safety Test (20%) asks "Is it safe?" using debt ratios and balance sheet strength. Score 0-100. But if Beneish M-Score > -1.78 (likely manipulator), Safety is capped at 30 regardless. The gamification version helps beginners intuitively understand value, quality, and safety.

If a company has negative earnings, P/E is not meaningful and gets no score for that indicator (the company is scored on other indicators in the Value pillar). If a metric is unavailable (e.g., FCF for early-stage companies), we skip it and score on available data. Free tier gets 30 core indicators (always available); paid tiers unlock more specialized indicators (some may be sparse in emerging markets).

Yes. VMCI is 100% transparent. Every stock page shows: the individual score for each of the 120 indicators, the percentile for each indicator, the category average (Value, Quality, etc.), and the final weighted score. You can see exactly how a stock scored on P/E (41st percentile), ROE (78th percentile), D/E (52nd percentile), etc. Most competitors hide their formulas. We publish ours because glass-box > black-box.

VMCI recalculates daily when new market prices are published and quarterly when companies report earnings/financials. Indicator scores are continuously updated so comparisons always reflect the latest fundamental data. Annual reports (10-K) are incorporated as soon as they're filed. The leaderboard snapshots the top/bottom 10 stocks daily.

Piotroski F-Score (1-9) measures balance sheet improvement and is a binary signal across 9 dimensions. Altman Z-Score predicts bankruptcy risk from 5 ratios. Beneish M-Score detects earnings manipulation from 8 variables. Each catches one dimension. VMCI integrates all three into the INTEGRITY pillar but adds VALUE (35%), QUALITY (30%), GROWTH (12%), and RISK (8%) — giving a single 0-100 score that captures the full investment thesis: am I paying a low price for a high-quality business with trustworthy numbers, room to grow, and bounded downside?

Magic Formula ranks stocks on two metrics: Earnings Yield (EBIT/EV) and Return on Capital. Greenblatt averages the two ranks. VMCI overlaps with Magic Formula on the Value/Quality axis but extends the framework: VMCI uses 28 value indicators (not just earnings yield), 41 quality indicators (not just ROC), and adds INTEGRITY/GROWTH/RISK pillars. In practice, Magic Formula top-decile stocks typically score 65-80 on VMCI; high VMCI stocks usually rank well on Magic Formula too. They agree on the direction; VMCI gives more granular separation.

Yes, indirectly through three channels: (1) the QUALITY pillar uses FCF/Net Income — heavy SBC inflates net income relative to cash, so cash conversion drops; (2) the GROWTH pillar uses per-share metrics (EPS growth, FCF/share) — share count growth from SBC suppresses both; (3) the INTEGRITY pillar Piotroski component awards a point for not issuing new shares. A company with 5%+ annual share-count growth from SBC will see meaningful percentile-rank deductions across all three pillars.

Empirically, valuation explains more cross-sectional return variance than quality alone (Fama-French 1992; Asness, Frazzini, Pedersen 2019). Buying a wonderful business at any price has destroyed returns multiple times in history (Nifty Fifty, 1999 tech, 2020 SaaS). The 35% Value weight ensures price discipline. The 30% Quality weight ensures we are not buying value traps. The 65% combined V+Q weight reflects the empirically-validated combination most associated with long-run abnormal returns.

VMCI uses GAAP earnings as the base. Non-GAAP adjustments are common ways to flatter results (excluding "one-time" charges that recur every year, capitalizing operating expenses). When non-GAAP earnings significantly exceed GAAP earnings, the GAAP/non-GAAP gap is a red flag captured in the INTEGRITY pillar via FCF/Net Income and the Beneish M-Score. Companies relying on aggressive non-GAAP reporting see their VMCI scores deduct on cash quality and earnings persistence.

Yes. VMCI is a fundamental ranking, not a forecast. It tells you the stock looks attractive on 120 metrics today. It does not predict: macro shifts, regulatory changes, technology disruption, management changes, or one-time events. A high VMCI is a strong starting point for further qualitative analysis, not a buy signal. Always combine VMCI with: business model understanding, competitive position, management track record, and your own portfolio fit. The methodology is glass-box exactly so you can audit each input and decide if it fits your investment thesis.

The illustrative chart shows the cumulative return of equal-weight VMCI top-decile stocks rebalanced quarterly vs the S&P 500 total return from 2010 to 2025. Limitations: (1) survivorship bias is not yet fully adjusted — historical universe excludes delisted names; (2) backtest is point-in-time-as-of-publication, not full as-of-decision-date with the data lags that existed historically; (3) does not include trading costs, taxes, or capacity constraints; (4) factor performance is regime-dependent and value+quality underperformed in 2017-2020. Treat the chart as directional evidence the framework has historical merit, not as a forecast.

Banks and insurance companies require sector-adjusted percentile ranking because their balance sheets differ structurally from industrials. VMCI groups financials into a separate peer group for percentile ranking — a bank with 12% ROE is compared against other banks, not against software companies (which routinely hit 30%+ ROE). For REITs, FFO and AFFO replace traditional FCF in the QUALITY pillar. Capital structure metrics (D/E) use sector-normalized thresholds. The 5-pillar weights remain identical; only the underlying ranking universe changes.

Yes, on the Professional plan. The /api/internal/methodology/score-history endpoint returns the full VMCI breakdown (each pillar, each indicator) for any ticker on any date back to January 2018. Free tier shows current snapshot only. Analyst plan shows last 12 months of monthly snapshots. Professional plan includes daily resolution and CSV export for custom backtesting.

Earnings restatements trigger a full VMCI recomputation for the affected periods. The historical score record reflects the restated values, with a "restated" flag on the audit trail. The Beneish M-Score and Piotroski F-Score are particularly sensitive — restatements often trigger a transition from pass to fail. The methodology assumes restated data is more accurate than the original; this matches the academic convention but means historical performance figures change slightly when restatements occur.