Scoring Factors

Net income is polluted by capital structure decisions (debt vs. equity), tax strategies, and one-time charges. Gross profit strips all of that away to reveal the core economics of the business. A company with high gross profitability relative to its asset base is extracting real value from its operations, regardless of how it is financed or what accounting choices it makes. Novy-Marx showed that gross profitability has roughly the same power predicting returns as book-to-market — the classic value factor — but captures a completely different dimension of quality.

WACC (weighted average cost of capital) blends the cost of equity (estimated via CAPM using beta and equity risk premium) with the after-tax cost of debt, weighted by the company's capital structure. The spread matters more than ROIC in isolation because a company earning 12% ROIC with a 10% WACC (2% spread) is creating less value than one earning 8% ROIC with a 5% WACC (3% spread). The spread is what drives compounding.

This formula produces a negative number when cash flow exceeds reported earnings, which is the healthy case — the company is generating more cash than its income statement suggests. A positive number means reported earnings exceed cash flow, indicating heavy reliance on accruals. The engine ranks stocks so that more negative values (higher cash-flow quality) receive higher percentiles. Sloan's research showed that the market systematically misprices the accrual component of earnings, creating a predictable return pattern.

Profitability (4 signals): Positive ROA, positive operating cash flow, ROA improvement year-over-year, and cash flow exceeding net income (quality of earnings). Leverage & Liquidity (3 signals): Decrease in long-term debt ratio, increase in current ratio, and no new equity issuance. Operating Efficiency (2 signals): Increase in gross margin and increase in asset turnover. Each signal is binary — the company either improved or it did not. This simplicity makes the score robust and hard to game.

While Earnings Quality (Sloan's accrual anomaly) captures the single-period accrual level, the Accrual Ratio looks at the cumulative pattern. A company might have a one-off accrual spike that does not repeat, but if the accrual ratio remains elevated across multiple periods, it signals a structural disconnect between reported earnings and cash generation. The two factors together provide a more complete picture of earnings reliability.

Scale Economics: Unit costs decline as volume grows, visible through improving margins at higher revenue. Pricing Power: The company can raise prices without proportional customer loss, measured through stable or expanding margins over time. Switching Costs: Customer retention rates and revenue concentration patterns suggest meaningful friction to leaving. Capital Efficiency: Consistently high ROIC relative to the sector median, indicating an economic moat protecting margins. Each signature is scored independently; a company with all four has a wide moat, while zero indicates no detectable competitive advantage.

Compounding requires reinvestment at attractive rates. If ROIC is volatile, the company cannot reliably predict what its next dollar of reinvestment will earn. Management faces harder capital allocation decisions, and investors face wider valuation uncertainty. Stable ROIC also signals that the competitive moat is durable — the company is not subject to boom-bust cycles driven by commodity prices, regulatory changes, or customer concentration.

The engine projects free cash flows using a two-stage model: a near-term growth phase based on analyst consensus and historical trends, followed by a terminal growth phase converging to GDP growth. The discount rate is the company-specific WACC. Because DCF is highly sensitive to terminal assumptions, the engine uses conservative terminal growth rates (2-3%) and cross-checks the result against three other valuation methods.

Two companies can have the same market cap but very different total costs to acquire if one carries significant debt. Enterprise value captures the true cost to an acquirer — you buy the equity and assume the debt, offset by cash on hand. Normalizing by free cash flow (operating cash flow minus capital expenditures) rather than earnings avoids accounting distortions from depreciation methods, tax strategies, and non-cash charges.

Greenblatt's "magic formula" combines a high earnings yield (EBIT/EV, the inverse of the Acquirer's Multiple) with high return on capital. Carlisle tested the two components separately and found that cheapness alone — the Acquirer's Multiple — drove most of the returns. Adding a quality filter actually reduced performance in his dataset, because the cheapest stocks included deep-value turnarounds that a quality screen would eliminate. The engine uses both approaches but keeps them as separate factors.

Standard free cash flow subtracts total capital expenditures, which includes both maintenance capex (replacing worn-out equipment) and growth capex (expanding capacity). Owner earnings attempts to subtract only maintenance capex, giving a truer picture of the cash available to owners. The distinction matters most for capital-intensive businesses where a large share of capex goes toward growth. A company spending heavily on expansion may have low reported FCF but high owner earnings.

The engine uses net buybacks (shares repurchased minus shares issued) to avoid crediting companies that buy back stock with one hand and dilute with stock-based compensation with the other. A company repurchasing $500M in stock while issuing $400M in SBC is only returning $100M net. This adjustment is particularly important in the technology sector where SBC is often a significant expense.

Instead of estimating intrinsic value from projected growth (forward DCF), Reverse DCF takes the current price as given and solves for the implied growth rate that would justify it. This implied rate is compared against the company's trailing 3-year revenue CAGR and reinvestment-driven sustainable growth rate. When the implied rate is lower than the demonstrated rate, there is a positive growth gap — the market is pricing in deceleration that has not yet materialized.

The engine uses tangible book value, excluding goodwill and other intangible assets that may not survive a liquidation. This is deliberately conservative — intangibles like brand value or patents have real economic worth in a going concern, but in a downside scenario they may be worth far less. The asset floor is meant to answer: "If everything goes wrong, what is the hard floor on this stock's value?"

Jegadeesh and Titman tested multiple formation periods and found that 12 months captures the intermediate-term momentum effect — the tendency of stocks with positive returns over the past year to continue outperforming. The 1-month skip avoids the short-term reversal (also called the "microstructure" effect), where stocks that have risen sharply in the last few weeks tend to pull back. The 12-1 construction has been replicated across decades and international markets.

When a company reports an earnings surprise, the stock price adjusts — but research shows it does not adjust fully. Positive surprises are followed by continued drift upward for 60-90 days, and negative surprises by continued drift downward. SUE standardizes the surprise magnitude, making it comparable across companies with different earnings volatility. The engine uses SUE as part of the catalyst signal for Track B scoring, where it indicates fundamental momentum that the market has not yet fully priced.

Lakonishok and Lee found that individual insider purchases have modest predictive power — insiders buy for many reasons including diversification plans and option exercises. But when multiple insiders buy in a tight window, the signal strengthens dramatically. These clusters indicate that several people with the best possible information about the company independently decided to put their own money at risk. The role weighting reflects that C-suite executives typically have better information than outside directors.

Not all institutional buying is informative. Index funds buy mechanically, and many active managers are closet indexers. Cohen, Polk, and Silli showed that the "best ideas" of skilled managers — their highest-conviction new positions — significantly outperform. The engine tracks a curated list of managers with demonstrated long-term alpha, focusing on their new position initiations rather than incremental additions. This filters out noise from mechanical rebalancing and focuses on genuine conviction signals.

Analyst revisions: Net upgrades minus downgrades over 90 days, normalized by coverage count. Short interest: Percentage change in shares sold short over 30 days — declining short interest means bears are covering. Options skew: The relative pricing of put vs. call options, which reflects the aggregate positioning of options traders. Each component is standardized and combined into a single composite, so no single signal source dominates.

The Runway Score combines the current reinvestment rate, the marginal return on new capital deployed, and the total addressable market relative to current revenue. A company reinvesting 40% of earnings at 20% incremental ROIC in a large market has a long runway. The same company in a saturated niche has a short one. This factor adds a forward-looking dimension to the momentum pillar — it distinguishes sustainable momentum from a growth rate that is about to hit a wall.