A backtest applies a system's exact rules to historical data to produce the list of trades it would have taken, and the statistics of those trades, so an edge can be measured rather than assumed.
Target audience: Traders with written rules who want to measure an edge before risking money, and avoid the classic backtesting traps.
A backtest applies a system's exact rules to historical data to produce the list of trades it would have taken, and the statistics of those trades, so an edge can be measured rather than assumed.
Backtesting is how a written system becomes evidence. Done correctly it tells you the expectancy, the drawdowns, and the trade frequency you should expect. Done carelessly it produces a beautiful equity curve that has nothing to do with the future, which is worse than no test at all because it breeds false confidence.
Feed the rules bar by bar through history and you get a trade list: entry time and price, exit time and price, and the result in R. From that list come the statistics that matter: expectancy per trade, win rate, average win and loss, the equity curve, and the maximum peak-to-trough drawdown. The equity curve is seductive but secondary; the per-trade expectancy and the drawdown are what tell you whether the edge is real and survivable.
Look-ahead is using information at a decision point that would not have been available yet. The classic forms: deciding on a bar's close but filling at that bar's open; using the day's high or low before the day is over; or using a indicator value that is only known after the fact. The rule that prevents it is strict: at the moment of any decision, the system may use only data that had already printed. If your test cannot reproduce the exact information set you would have had live, its results are fiction.
Survivorship bias is testing today's surviving instruments and ignoring those that were delisted, which flatters results because the failures are missing. Test the universe as it existed at each date. Two more quiet biases: using revised data that was not available in real time, and ignoring trades you could not actually have entered (illiquid, halted, or limit-locked markets). A backtest is only as honest as the data and the fills it assumes.
A breakout system signals on the close of the 30-bar-high bar. If the backtest fills at that close, it captures a price that, live, you would already have missed. Fill instead at the next bar's open, or at a stop one tick above the level with assumed slippage. On a fast instrument this difference alone can turn a +0.3R expectancy into a negative one. The edge lives or dies in the fill assumption, not the entry rule.
Deciding on a bar's close but filling at the same bar's open or price (look-ahead).
Testing only instruments that still exist today (survivorship bias).
Assuming perfect fills with no spread, slippage, or commission.
Optimising on the entire history and reporting that same history as the result.
Falling in love with the equity curve while ignoring drawdown and trade count.
Myth
That a great backtest guarantees a great future.
Reality
No paired reality note provided.
Myth
That more historical profit means a more robust system.
Reality
No paired reality note provided.
Myth
That the prices in your data are exactly the prices you could have traded.
Reality
No paired reality note provided.
Take any backtest (yours or a published one) and check it against the three bias traps.
This lesson is educational content only and is not financial advice. Trading involves substantial risk. A tested process improves decision quality and survivability; it does not predict the market or guarantee any outcome. Trade only with risk you can afford to lose.