Key takeaways
- A backtest is a history lesson, not a forecast.
- Overfitting and survivorship bias flatter results.
- Compare against a simple baseline and look at the worst stretch.
A backtest replays a rule, for example “hold these five stocks and rebalance yearly”, over past data. It is useful for seeing how a rule behaved, including how bad it got at its worst.
Ways backtests mislead
- Overfitting: tuning a rule until it fits the past perfectly, which tells you little about the future.
- Survivorship bias: testing only companies that still exist today.
- Cherry-picked periods: a start date that flatters the result.
- Ignoring costs, taxes and the difficulty of sticking to the plan in a downturn.
What to look at instead of the headline return
- The largest peak-to-trough fall.
- How it compared with a plain index or simple buy-and-hold.
- How results changed with different start dates.
If a result only works with one narrow set of settings, treat it with suspicion.
Past results are not a promise. A backtest tells you what happened, not what will.
For informational and research purposes only. Not financial advice. Past performance doesn't guarantee future results.