Alpha decay describes the erosion of a signal's predictive accuracy — measured by its Information Coefficient (IC) — as the return measurement horizon extends. A signal with a short half-life decays quickly: its IC at a one-week horizon may be meaningful, but by one month it has fallen toward zero.
Decay is measured empirically by computing IC at successive forward-return horizons (1-day, 5-day, 10-day, 21-day, and beyond) and plotting the resulting curve. The horizon at which IC has fallen to half its peak defines the signal's half-life.
Implications for strategy design
- Fast-decaying signals require high turnover and very low transaction costs to remain profitable.
- Slow-decaying signals tolerate higher trading costs and lower frequency, but compete with more market participants over time.
Common drivers of decay include arbitrage by competitors, news incorporation, liquidity cycles, and regime shifts. Monitoring rolling IC over time detects whether a live signal's decay profile is changing.