Manual backtesting

You can run a manual backtest by replaying past charts candle by candle, log every trade consistently and avoid the most common biases.

Test on the past

Before risking money on a set of rules, there is a cheaper question to ask: how would these rules have done in the past?

A backtest applies a rule set to historical data to see what it would have produced. Programmers do this automatically, but you can also do it by hand, with a chart, a spreadsheet and discipline. A manual backtest is slow, which is part of its value: you see every trade as it would have looked at the time.

One candle at a time

Pick a period. Choose a market, a timeframe and a stretch of history, ideally including trends, ranges and at least one volatile phase.
Hide the future. Many charting platforms have a replay mode that shows past candles one at a time. Without it, cover the right side of the chart.
Apply the rules. At each new candle, check your written setup, trigger, exit and management rules exactly as written.
Log every trade. Date, direction, entry, stop, target, exit, result in R, and a short note. Include costs, such as the typical spread.
Evaluate. Only after the whole period: win rate, average win and loss, expectancy, drawdown and longest losing streak.

The biggest danger is look-ahead bias: using information that was not available at the time of the decision. Seeing where the price went, even for a moment, makes every rule look better than it is.

Honest vs. fooling yourself

The same backtest can be honest or flattering, depending on a few habits:

Rules written first. If a rule changes halfway, the test restarts from the beginning with the new rule.
Every signal logged. A setup that "looked bad" but met the rules counts. If you would really skip it, write that into the rules.
Costs included. Spreads and commissions are part of every trade.
Enough trades. Twenty trades say little. A hundred or more, across different market phases, start to say something.

A practical target for a first manual test: 100 trades on one market and timeframe. It takes several evenings, and it teaches you the rules better than any amount of reading.

Common mistakes

Peeking at the right side of the chart. Hide the future every time.
Skipping signals that look unattractive. If the rules triggered, the trade counts.
Changing rules mid-test without restarting. Results from different rule sets cannot be mixed.
Testing only one kind of market. A rule tested only in a strong trend may fail in a range.

The takeaway

A manual backtest applies your written rules to past charts, one candle at a time, without seeing the future. Log every trade, including the losers.

Check yourself

Quick quiz

4 questions. Instant feedback, nothing is saved.

1Tap the step that prevents you from seeing the outcome before deciding.

Tap the chart.

Replaying the chart candle by candle, with the future hidden, keeps you from judging setups by what happened next.
2You notice the next candles before deciding on a setup. What bias is that?

Using information that was not available at the time makes backtest results look better than live results can be.
3Why log signals you would 'never have taken'?

Either the rule changes for every future test, or the trade counts.
4What is a reasonable minimum before drawing conclusions from a manual backtest?

Small samples are dominated by luck. Different market phases show whether the rules only work in one kind of market.

of 4 correct.

Sources

Back to Trading Styles & Strategy Development