Signal and Alpha Decay
Signal decay is how fast a signal's predictive power fades; alpha decay is how edges shrink over years. Learn both, the evidence and how traders adapt.
"Decay" means two related things in trading research. Signal decay describes how quickly a signal's predictive power fades after it is observed: a news signal might matter for minutes, a value signal for months. Alpha decay describes how an edge shrinks over the years as more traders discover and exploit it. Understanding both helps traders choose holding periods, control costs and accept that every strategy needs ongoing research to stay profitable.
Signal decay: the time dimension#
After a signal fires, prices adjust as the market absorbs the information. Measuring returns at different horizons after the signal shows its decay profile.
| Signal type | Typical decay horizon |
|---|---|
| Order flow and microstructure | Seconds to minutes. See High-Frequency Trading |
| News and earnings surprises | Hours to weeks. See Earnings Reactions and Post-Earnings Drift |
| Short term reversal | Days to a month. See Short and Long-Term Reversal |
| Momentum | Months. See Momentum Factor |
| Value and quality | Many months to years. See Value Factor |
Alpha decay: the long term dimension#
Edges shrink as capital chases them. Evidence includes:
- Publication effects: McLean and Pontiff (2016) studied 97 anomalies from academic papers and found their returns were about 26% lower out of sample before publication and about 58% lower after publication, consistent with both overfitting and investors trading on the published findings.
- Crowding: popular quant factors have shown periods of weaker returns as assets in factor products grew. See Factor Timing, Crowding and Crashes.
- Technology: faster execution and better data made many short term inefficiencies disappear.
Why alpha decays#
| Cause | Mechanism |
|---|---|
| Competition | More traders exploiting the same edge push prices to fair value faster |
| Publication and education | Ideas spread through papers, books and forums |
| Market structure changes | New rules, venues or participants change behaviour |
| Data availability | Data that was once rare becomes common. See Alternative Data Explained |
| Regime changes | Conditions that created the edge disappear. See Structural Breaks and Regime Changes |
| Original overfitting | Some "edges" never existed |
Recognising decay in your strategy#
- Rolling performance trending down while volatility is unchanged.
- Shorter signal decay curves: the market reacts faster than before.
- Rising costs and crowding indicators.
- Lower information coefficients over time.
Distinguishing decay from a normal drawdown requires comparing performance with expected ranges over a meaningful period. See Monitoring Positions, P&L and Risk and The Strategy Lifecycle.
Adapting to decay#
- Keep researching: develop new signals continually.
- Diversify signals so no single decaying edge dominates. See Combining Signals.
- Improve execution to capture more of a fast decaying signal.
- Seek less crowded niches: smaller markets, new data, longer horizons.
- Size by capacity: stay small enough not to erode your own edge. See Alpha Capacity and Crowding.
Decay and costs#
Fast decaying signals must be traded quickly and often, which raises costs. If a signal loses most of its value within a day, slow execution can capture little of it. Measuring how much of the theoretical return survives after realistic delays and costs is part of judging any signal. See Costs and Slippage in Backtests.
Frequently asked questions#
What is signal decay?#
How quickly a trading signal's predictive power fades after it is observed, which determines the best holding period.
What is alpha decay?#
The long term shrinking of a trading edge as more traders discover and exploit it, or as market conditions change.
Why do published trading strategies stop working?#
Because some were overfit to begin with and others become crowded as investors trade on them, reducing returns.
Next, learn to blend several signals in Combining Signals.
3 quick questions on this lesson. Get them all right to finish it.
Turn on JavaScript to take the quiz.
Mentioned in
- Why Strategies FailResearch and Backtesting
- What Is Alpha?Research and Backtesting
- Size FactorResearch and Backtesting
- Quant Trading Learning PathStart Here
- Recency BiasTrading Psychology
- Quantitative TradingStrategies and Styles