A journal helps answer whether you followed your process, while a backtest explores how defined rules would have behaved on a historical sample. Neither can guarantee future performance, and both are only as reliable as their records, assumptions, and interpretation.
1. Record the plan before the outcome is known
For each trade, capture the date and time, instrument, market conditions, setup name, entry rationale, entry and stop, target, planned risk, position size, and the condition that would invalidate the idea. Save a chart or screenshot when useful. Recording the plan in advance makes it easier to distinguish the original decision from explanations added after the result.
2. Record execution and costs separately
After the trade, record actual entry and exit, fees, spread, funding where applicable, slippage, realized profit or loss, and any deviation from the plan. A setup may look attractive before costs but perform differently after realistic execution expenses. Do not replace missing values with assumptions without labelling them.
3. Compare outcomes in units of initial risk (R)
R is a common way to normalize results against the initial planned risk. If a trade was planned to risk $20 and closed with a $40 gain before costs, the result is +2R gross. If it lost $20, the result is −1R gross. Keep gross and net results distinct and define how partial exits, commissions, and changes to stops are treated.
4. Review more than the win rate
Useful measures can include total net return, average win and loss, expectancy, profit factor, maximum drawdown, number of trades, time in market, and performance across different market conditions. A high win rate alone does not establish profitability if losing trades are much larger than winners. Small samples can be misleading, so note the number of observations and uncertainty.
5. Define backtest rules before running them
Write precise entry, exit, stop, target, sizing, and filtering rules before examining the results. Specify the timeframe, data source, fees, spread, slippage assumptions, and how missing or illiquid periods are handled. If rules are repeatedly changed to fit the same historical sample, the result becomes increasingly vulnerable to overfitting.
6. Avoid common testing errors
- Look-ahead bias: using information that would not have been available at the time of the simulated decision.
- Survivorship or selection bias: testing only symbols or periods that are known to have performed well.
- Unrealistic fills: assuming every order executes at the exact historical price without spread, slippage, or liquidity limits.
- Overfitting: tuning many parameters until one sample looks good, then treating that result as general.
- Data leakage: allowing information from the test period to influence rules supposedly selected beforehand.
7. Separate development from validation
Use one sample to develop the idea and a separate, later or otherwise untouched sample to check whether the rules remain plausible. Forward testing or paper trading can reveal operational issues that historical simulations miss. Even successful validation does not eliminate the possibility of future regime changes or losses.
8. Turn the review into a process improvement
Review a consistent group of trades rather than changing rules after every single outcome. Look for repeatable execution errors, conditions where the strategy struggles, and whether real results differ from the assumptions. Record any rule change and evaluate it on new data instead of rewriting the history of the original test.
How TRADEX fits in
TRADEX includes a Journal Trading workflow and Backtest Setup tools to help organize trade records and review R-based outcomes. The available tools do not remove the need to verify inputs, understand their calculation assumptions, and account for costs. See the TRADEX methodology and risk management guide.
Important: This guide is educational and not financial advice. Historical, simulated, and backtested results do not guarantee future results. Trading forex, leveraged products, and crypto assets involves substantial risk.