Trading practice · Manual backtesting
A repeatable workflow for manual forex backtesting
Manual backtesting is a process for testing your own decisions against historical data without looking ahead. Set the question and costs first, reveal prices in order, keep a decision log, and review the evidence with the replay limitations in mind.
Method
A six-step test you can repeat
Write one question and one rule
State what you want to learn in a sentence. Define the setup, entry trigger, invalidation or exit, and the information you are allowed to use at each decision.
Choose the data and assumptions
Select an available forex instrument and date range. Record the starting capital and the spread, fee and slippage assumptions you will use; avoid changing them halfway through a run.
Reveal bars in chronological order
Start replay and use pause or step controls before each decision. Base each choice only on the prices revealed so far, not on the completed chart or what you remember happens next.
Record every decision, including no trade
Note the timestamp, setup, action, position size, planned exit and reason. A skipped setup is useful evidence too; log it instead of silently removing it from the sample.
Finish the run and inspect the review
Compare your notes with fills, costs, net P&L and the equity or drawdown curves. Look for rule breaks and execution patterns before deciding whether the rule needs another test.
Repeat on a separate sample
Keep the rule fixed for the next run and use a different period where possible. Separate the data used to develop an idea from the data used to assess it.
How can you make a manual backtest repeatable?
A short test card makes it easier to distinguish a clear rule from a story added after the chart is visible.
- Market and sample: record the instrument, date range and data source shown in the selector.
- Setup and trigger: describe the observable conditions using only information available by that time.
- Risk and exit: write down how size is chosen, what invalidates the idea and how an open trade is closed.
- Costs: note the configured spread, fee and slippage assumptions before starting.
Reduce look-ahead and selection bias
A replay can help you reveal history in sequence, but the exercise is only as disciplined as the information you let yourself use.
- Pause before a decision and keep future candles hidden until you have recorded the choice.
- Do not discard losing attempts or no-trade cases. Define when a run ends before seeing its result.
- Avoid changing a rule after each outcome and then reporting only the best variation; treat changes as new hypotheses.
- Use another date window for a follow-up check instead of repeatedly tuning to the same sample.
Keep a compact decision journal
Write the reason at the time of the decision, then compare it with what the review shows after the run.
- Useful fields: date/time, instrument, setup, visible evidence, action or no-trade, size, planned stop/exit, costs, exit reason and rule-following note.
- Review process quality separately from the result of one trade. A positive simulated outcome can still come from a rule break.
- When signed-in saving succeeds, the product can reopen the review from Dashboard History; keep your own test definition with the session notes.
Scope
Interpret results cautiously
- The replay uses one-minute OHLC data and a synthetic intraminute price path, not tick-by-tick history.
- Historical touches and simulated fills do not establish that a live order would fill at the same price.
- Preset forex coverage and date ranges can change; check the selector, and treat imported custom data according to its stated UTC handling.
- A manually reviewed sample can be small or biased. It is not a reliable forecast or evidence of future profitability.
- Manual practice is distinct from automated Strategy Replay, which requires Pro.