Backtesting means applying written rules to past charts and recording what would have happened. Done carefully, it filters out ideas that clearly do not work. Done carelessly, it produces confidence in ideas that do not work. This walkthrough is a hands-on, manual version; tools and theory are covered in other guides. Past results never guarantee future ones.
TL;DR -- Key Takeaways
- Write the rules first; a backtest of vague rules proves nothing.
- Record every signal, including the ones that lose, with the real payout.
- Avoid look-ahead: decide each trade using only what was visible at that moment.
- A sample that is too small or too narrow is not evidence.
- A passed backtest earns a forward test on a demo account, not real money.
1. Step 1: Write the Rules
Use the indicator-based entry rules template or an equivalent. The rule must name the market, timeframe, trigger, filters, expiry and stake. If a rule contains judgement words such as 'strong' or 'clean', replace them with measurable conditions.
2. Step 2: Choose the Data
Take a stretch of history from the market and timeframe you plan to trade, and include more than one condition: trending, ranging and quieter periods. Choosing only the most favourable months biases the result. Decide the date range before you start and write it down. Do not mix timeframes unless the rule says so.
3. Step 3: Scroll Forward One Candle at a Time
Cover the right side of the chart, or use a replay feature if your platform has one, and reveal candles one by one. At each closed candle ask whether the trigger has occurred. If yes, note the entry price, time and expiry before revealing the result. This protects you from look-ahead bias, where later candles unconsciously influence your decision.
4. Step 4: Record Everything
Create a sheet with the columns: date, time, direction, entry price, expiry, price at expiry, payout, result, rule followed. Use the real payout you would have received, not a round number. The strategy journal template has a ready layout. Include trades that the filter blocked in a separate column so you can judge whether the filter helps.
5. Step 5: Add Costs and Realism
Deduct any fees that apply to you. Allow for the fact that live entries are not always at the exact price you see, which matters most on short expiries; see execution and slippage mistakes. A result that only works if entries are perfect is fragile.
6. Step 6: Summarise Honestly
Count trades, wins, losses and the win rate. Compare the win rate with the break-even rate, which is 1 divided by (1 plus the payout as a decimal); use the payout calculator. Look at the longest losing run with the losing streak risk calculator and ask whether you could follow the rule through that run. Then see whether results hold in each sub-period, not just in total.
7. Step 7: Check for Common Biases
- Overfitting: adjusting the rule until the past looks good; see overfitting a strategy.
- Small sample: too few trades to separate skill from luck; see sample size and statistical significance.
- Survivorship and cherry-picking: testing only on assets or dates that suit the rule.
- Look-ahead: using information that was not available at the entry.
8. Step 8: Move to Forward Testing
If the backtest holds up, run the unchanged rule on a demo account in real time for a set number of trades and compare. For the wider development process and tools, see backtesting and strategy development and strategy backtesting tools.
Frequently Asked Questions
How many trades should a backtest include?
There is no fixed number. Small samples are unreliable, so use as many trades as the history reasonably allows, across different conditions.
Can I backtest in a spreadsheet?
Yes. Record each manual trade in a sheet and calculate the win rate and results from it.
What does a good backtest prove?
It shows the rule did not clearly fail on past data. It does not prove the rule will work in future.
Should I optimise the settings to get the best result?
Be careful. Optimising on the same data often produces overfitting. Decide the settings before testing and check them on separate data.
Related Reading
- Backtesting Your Own Strategy -- an advanced view
- Strategy Journal Template -- recording trades
- Common Strategy Flaws -- what to check
Compare brokers: