A new investment model automatically cuts risk during market crashes.

Investors often try to copy a target benchmark by buying a smaller basket of assets. Doing this with standard math tools usually creates unstable portfolios. These setups borrow too much and crash easily when markets drop. To fix this, researchers built a two-stage method based on information theory. First, they connect asset returns to target factors using an entropy minimizationentropy minimizationA method from information theory that finds the most probable solution given constraints. principle to keep parameters inside data-driven bounds. Next, they use that same entropy rule to pick the replication weights. Both steps rely on a Fermi-Dirac entropy function applied directly to the problem's constraints. Across five experiments, this model beat standard Ordinary Least Squares by lowering annualized turnover and boosting net-of-fees returns. It also works like a circuit breaker during market crashes and data errors by cutting capital to risky assets.

Why it matters

Investors and fund managers using standard methods can end up with fragile portfolios that fail during market shocks, leading to big losses. This model provides a way to build more stable portfolios that protect capital in crises like the COVID-19 crash.

Questions answered