What Is Alpha Decay? Definition and Example
Alpha decay is the fading of an investing edge over time as more people find and trade it. Here is what the research shows and what it means for strategies you read about.
Backtest overfitting is when a strategy is tuned so closely to past data that it captures noise instead of a real pattern. Here is how to spot it, with a real example.
What is your own portfolio exposed to? Check it with Exent →
A backtest replays a rule on historical prices. Overfitting happens when the rule is shaped by that same history: the parameters, the start date, the list of assets. With enough tries, some rule will always look brilliant in the past. It is also called curve fitting or data snooping.
See how this applies to your own holdings with an Exent portfolio check.
Exent’s hypothetical backtests re‑optimize a six‑ETF mix (SPY, QQQ, IWM, TLT, GLD, AGG) on a rolling window of the prior 36 months, so each new mix is then tested on data it has not seen. Jul 2019 – Sep 2026:
| Return a year | Sharpe | Max drawdown | 2022 | |
|---|---|---|---|---|
| Max‑Sharpe optimized | 13.5% | 0.83 | −30.5% | −27.8% |
| Minimum‑variance optimized | 0.7% | −0.27 | −17.1% | −13.0% |
| SPY | 15.7% | 0.83 | −23.9% | −18.2% |
| Max‑Sharpe optimized | Minimum‑variance optimized | SPY | |
|---|---|---|---|
| Return a year | 13.5% | 0.7% | 15.7% |
| Sharpe | 0.83 | −0.27 | 0.83 |
| Max drawdown | −30.5% | −17.1% | −23.9% |
| 2022 | −27.8% | −13.0% | −18.2% |
The optimizer always picked the mix that looked best on the past three years. On new data that fit did not carry forward, which is the same gap between in‑sample and out‑of‑sample results that overfitting creates. See optimization pitfalls.
Newsletters, apps and social posts show backtests that were chosen because they look good. Ask how many versions were tried and whether the result held on data the rule never saw. See out‑of‑sample testing and permutation test.
Use few parameters, decide them before testing, hold back data the rule never sees, include costs, and count how many variations you tried.
No. A simple rule tested honestly on long, out-of-sample data is useful context. The problem is a backtest selected for looking good.
Very smooth past returns, many finely tuned settings, and results that change sharply when a setting or start date moves slightly.
Upload a screenshot of your holdings and get concentration, overlap, crash exposure and a risk-matched benchmark in about a minute.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.