Backtesting means applying a trading strategy's exact rules to historical price data to see how it would have performed. Done properly, it's one of the few ways to get real evidence about whether a strategy has any edge at all before committing actual capital to it. Done carelessly, it can produce a strategy that looks brilliant on paper and falls apart the moment it meets live, unseen data — which is exactly what happens to a lot of "backtested 500% return" claims circulating online.
What backtesting is actually for
The honest purpose of a backtest isn't to prove a strategy will be profitable — it's to rule out strategies that clearly aren't, and to build a rough sense of a strategy's behavior: how often it wins, how large the losses tend to be, how long losing streaks typically run, and whether the logic behind it holds up across different market conditions. A backtest is evidence, not proof, and treating it as anything more than that is where most of the damage in this lesson comes from.
The pitfalls that quietly invalidate a backtest
The same data, fit two different ways
Same seven data points in both panels. The overfit line looks like a "perfect" backtest — it explains 100% of the historical data. The robust line explains less of the past but is far more likely to generalize to data it hasn't seen yet, which is the only kind of data a live trade actually faces.
Overfitting Curve-fitting
This happens when a strategy's rules are tuned so precisely to historical data that they capture noise from that specific period rather than a real, repeatable pattern. A strategy with a dozen finely-tuned parameters can often be made to show a near-perfect historical result — and will almost always perform far worse on data it hasn't seen, because it was never actually finding an edge, just memorizing the past.
Look-ahead bias
This is when a backtest accidentally uses information that wouldn't have been available at the time of the trade — for example, using a day's closing price to decide whether to enter earlier that same day. It's an easy mistake to make in backtesting software and it inflates results in a way that's invisible unless you're specifically checking for it.
Ignoring real trading costs
Spread, commission, and slippage (the difference between the price you wanted and the price you actually got) all eat into real returns. A backtest run on clean historical prices with none of these costs applied will consistently overstate how a strategy performs once it's actually being traded.
Small sample size
A strategy tested over 20 trades can look exceptional or terrible almost entirely due to chance. Meaningful conclusions generally require a large enough sample — and a long enough stretch of time, covering more than one type of market condition — before the results say much about the strategy itself rather than about the specific period tested.
Survivorship and data-quality bias
Historical data that excludes delisted instruments, uses adjusted prices inconsistently, or has gaps and errors can distort results in ways that are hard to spot without deliberately checking data quality first. A backtest is only as trustworthy as the data feeding it.
Why this connects directly to guru marketing
Claims like "this strategy returned 500% backtested over five years" are technically often true and still misleading — because an overfit strategy can be engineered to show almost any historical result you want. This is the same pattern covered in the lesson on spotting fake trading gurus: an impressive-looking number, without visibility into how it was produced or whether it holds up out of sample, tells you very little on its own.