Governance scores that survive real market tests

Most ratings tell you what feels right. We built ours by asking a harder question: which boardroom habits actually predict how firms are priced across markets?

Our methodology started with a simple frustration. A client showed us two governance reports on the same company. One scored it highly for having a separate chair and CEO. The other penalized it for the same structure. Both claimed to predict performance. Neither showed their work. We set out to build something different: a scoring system you can reproduce, criticize, and improve.

Every component is tested against cross-sectional market data before inclusion
Weights are derived from predictive power, not expert judgment alone
Full methodology, code, and sensitivity tables are shared with subscribers

Request a sample See our research

Analyst reviewing governance data on multiple screens

Score drift 12%

Average governance score migration for firms that later experienced major valuation events, measured twelve months prior

Firms scored 847

Data collection and cleaning

We extract governance disclosures from regulatory filings, proxy statements, and sustainability reports across eighteen jurisdictions. Every data point is manually verified against primary sources, flagged for restatements, and cross-checked for consistency across reporting periods. We discard approximately fifteen percent of raw observations due to ambiguity or non-comparability.

Factor identification

Candidate governance factors are drawn from academic literature, regulatory guidance, and historical case studies. We do not include a factor simply because it sounds important. Each candidate must demonstrate a plausible theoretical link to firm value and must be measurable with reasonable precision across our full coverage universe.

Predictive testing

Surviving factors enter a cross-sectional regression framework where we test their relationship to subsequent valuation multiples, return patterns, and volatility measures. We control for sector, size, profitability, and leverage. Factors that fail to show statistical significance across multiple time periods are removed or reformulated.

Weight calibration

Component weights are calibrated using out-of-sample validation periods. We deliberately withhold recent data during model construction and test whether the weighted score predicts outcomes in the held-out period. This prevents overfitting and gives us honest estimates of real-world performance.

Publication and monitoring

Final scores are published with confidence intervals, component breakdowns, and explicit discussion of model limitations. We monitor score stability, track prediction errors, and publish quarterly methodology reviews where we admit what broke, what we fixed, and what we still do not understand.

What sets us apart

Honest methodology in a world of black boxes

Traditional governance ratings mix together dozens of unrelated variables, apply opaque weighting schemes, and change their formulas without warning. We do the opposite. Our approach is designed to be criticized, replicated, and improved over time.

Predictive validity first

We only include factors that demonstrate measurable predictive power in out-of-sample tests, not factors that sound good in a committee meeting.

Full transparency

Subscribers receive the complete methodology, source code, and sensitivity analyses. You see exactly how each score is constructed and where the uncertainty lives.

Cross-sectional robustness

Our tests are designed to identify which governance patterns distinguish firms from their peers, not just which patterns coincide with broad market movements.

Continuous validation

Scores are back-tested against historical data and forward-tested against live outcomes. When a signal degrades, we tell you before you discover it the hard way.

Peer review culture

Internal analysts challenge each other's work before publication. External academics and client practitioners are invited to critique our methods and suggest improvements.

Technical foundations

Regression framework

Our primary analytical tool is a firm-level cross-sectional regression where governance scores and their components are regressed against subsequent valuation multiples and return metrics. We employ pooled OLS specifications with clustered standard errors at the sector level, supplemented by fixed-effects models that absorb time-invariant firm heterogeneity. Control variables include log market capitalization, return on assets, leverage ratio, and prior twelve-month return.

Econometric regression analysis on screen

Factor controls

To isolate governance effects from confounding influences, we systematically control for known drivers of cross-sectional variation. Size effects are captured through market capitalization deciles. Profitability effects use trailing return on equity and operating margin. Momentum and value exposures are measured using standard factor loadings. We also include sector fixed effects to prevent industry composition from driving apparent governance relationships.

Robustness checks

Every reported relationship undergoes multiple robustness tests. We re-estimate specifications using alternative variable definitions, different sample periods, and winsorized outliers at the one-percent level. Subsample analyses examine whether governance effects vary across small and large firms, high and low disclosure environments, and different legal traditions. Results that fail to replicate across these tests are flagged as fragile rather than reliable.

Limitations and caveats

We are explicit about what our methodology cannot do. Governance scores explain only a fraction of cross-sectional variation in firm outcomes. Causal identification remains challenging because governance structures are endogenous to firm performance. Our data relies on voluntary disclosures that may not reflect actual boardroom behavior. We discuss these limitations in every report and caution against overinterpreting any single score.

Score coverage

847 firms

4.9 Methodology clarity

Download the full methodology whitepaper

Our complete technical documentation includes factor definitions, regression specifications, weighting formulas, and a history of methodological changes. Available to qualified investors and academic researchers upon request.

Request whitepaper

Professional reading governance methodology whitepaper