Skip to content

Quantitative Signals & Scoring

Edge Half-Life Calculation

A quantitative measure of the time period required for an insider trading signal or informational edge to decay to half its original predictive power or alpha generation capacity.

Edge half-life calculation is central to insider-trading quant platforms, as insider signals exhibit predictable information decay. Unlike public market signals that may persist indefinitely, insider information asymmetries erode as the market digests Form 4 filings, Schedule 13D updates, and PDMR transaction reports. The half-life metric quantifies this degradation by measuring the time elapsed until a signal's risk-adjusted return (alpha or information coefficient) drops to 50% of its initial value. This decay function is non-linear, typically following an exponential or power-law distribution.

Practitioners calculate edge half-life by regressing cumulative abnormal returns (CAR) or signal-conditioned excess returns against time elapsed since transaction disclosure. The half-life value directly informs position sizing, signal reweighting schedules, and portfolio turnover optimization. A signal with a 5-day half-life requires rapid capital deployment and tactical exit windows, while a 60-day half-life permits medium-term conviction holding. In insider-trading surveillance contexts, half-life analysis also serves as a proxy for information leakage speed, flagging whether trades exhibit leaked information characteristics versus genuine fundamental insight.

Formula

IC(t)
Information coefficient at time t post-disclosure
IC(0)
Information coefficient at time of initial signal generation
t
Time elapsed in trading days
Sources: Easley, D., López de Prado, M. M., and O'Hara, M. (2012). The microstructure of the flash crash: Flow toxicity, liquidity crashes, and the probability of informed trading. Journal of Portfolio Management, 38(2), 102-113. · Campbell, R. H., Huisman, R., and Koedijk, K. (2001). Optimal portfolio selection with derivatives. Journal of Banking and Finance, 25(9), 1595-1617.