What a backtest can and can’t tell you
A backtest measures how a set of rules would have performed on historical data. It’s genuinely useful for sizing expectations and comparing variations — for example, “what if I’d taken profit later?”
What it can’t do is guarantee the future. Markets change, and a rule that fit the past may not fit what’s next. Treat a backtest as evidence, not a promise.
The metrics to judge
Look beyond total return:
- Expectancy: average profit per trade — the core measure of edge.
- Win rate and risk-to-reward together — neither means much alone.
- Maximum drawdown: the worst peak-to-trough stretch you’d have endured.
- Sample size: dozens of trades minimum; a handful is noise.
A strategy with a smaller return but a shallow drawdown is often more tradeable than a high-return, high-drawdown one you’d abandon at the bottom.
Avoid overfitting — and test on your own trades
Overfitting is tuning a strategy so tightly to past data that it captures noise instead of signal. The tell is a backtest that looks perfect but falls apart live. Guard against it with simple rules, enough data, and out-of-sample testing.
The most honest backtest uses your own history. CoinCrypTick’s what-if optimizer replays your real trades and shows what smarter exits, stops or timing would have returned — a counterfactual on data you actually traded, not a curve-fit on someone else’s.