Published 2026-10-14
Alpha decay
Alpha decay is the fading of a signal's predictive power over time — after it is discovered, after it gets crowded, or simply because the edge was never stable.
What it is
Alpha decay is the fading of a signal’s edge over time.
An effect that was strong in the 1990s may be weak today — because it was arbitraged away, crowded out, or was never as stable as the backtest suggested.
Why it matters for a backtest
A backtest averages over the whole test window. If the edge decayed inside that window, the average hides it.
| If you only report | You miss |
|---|---|
| Full-period Sharpe | Whether the edge is alive at the end |
| Total return | Whether it is fading |
| A single number | The shape of the decay |
The end of the sample is the most relevant part — and the easiest to hide behind an average.
How to check
Ask for the result by sub-period, not just in total. A signal whose edge is concentrated in the distant past is a historical artefact, not a live strategy.