How the Market Risk Score is built.

The Market Risk Score measures how vulnerable U.S. equities are to a meaningful market decline over the selected horizon. It combines valuation, interest rates, credit conditions, economic momentum, inflation, corporate profits and market internals.

The dashboard uses the 3-month composite as its headline reading and also lets members inspect separate 1-month, 6-month and 12-month model horizons. Interactive horizon controls update the category composition and qualitative downside-vulnerability views without converting the score into a probability.

The member Dashboard translates the selected 1-, 3-, 6- or 12-month model risk state into qualitative downside-vulnerability labels for a 10% correction, 15% decline and 20% bear market. The 6-month view is the default.

These labels are ordinal risk-state descriptions informed by the current MarketRiskScan model and retrospective research. They are designed to help users interpret relative vulnerability by decline severity and horizon.

A Treasury yield, inflation rate, CAPE ratio and volatility index cannot be averaged directly because they use different units and historical ranges. MarketRiskScan first maps each indicator to a comparable 0–100 risk scale.

For indicators configured with trend, the current change over the specified trend window is compared with that indicator's own historical changes. Some custom indicators use purpose-built nonlinear rules or composites instead of a simple percentile. For example, Yield Curve Stress and Oil & Gas Shock contain time-dependent stress components directly in the composite.

Each indicator receives a Data Quality score based on two operational factors: freshness and historical coverage.

Data Quality affects the score itself, not just the label beside it. An indicator with incomplete history has proportionally less influence.

Available effective weights are then renormalized. Missing data are never silently assigned a score of 50. The stored model field historically named Data Confidence summarizes weighted data freshness, coverage and completeness across the model. On public-facing pages this is described as Data Quality to avoid implying forecast certainty.

MarketRiskScan stores daily model snapshots going forward. These snapshots can be graphed directly because they represent what the operational model calculated on that date.

Reconstructing earlier periods such as the dot-com peak, the 2007 market peak, or 1929 requires a separate point-in-time research process. A valid historical reconstruction should only use observations that were actually published and available by the simulated date. Revised economic data must not be treated as though investors knew the revised values in real time.

Model definitions are versioned. Changes to indicators, weights, lookbacks or transformations should create a new model version so historical snapshots remain interpretable. The current public-data framework is v1.2.

Market Risk Scan is an informational risk-monitoring tool. It is not investment advice, is not individualized to any investor, and does not predict future returns. Scores should be considered alongside other information and judgment.