A strategy you have not measured is a belief. Journaling and backtesting are the two cheapest ways to turn beliefs into numbers — and the only reliable way to know whether a losing run means stop or continue.
Record the instrument, timeframe, direction, entry, stop, target, position size, and the percentage of the account risked. Then add the part most people skip: the reason for entry in one sentence, and whether the trade followed your rules regardless of how it ended.
That last column is the valuable one. It separates outcome from process: a rule-following loss is a cost of doing business, a rule-breaking win is a warning. Over fifty trades the split between them tells you more about your account's future than the profit-and-loss line does.
Write the rules down before testing so entry conditions cannot quietly change to fit what you already saw. Test on data you have not studied, move forward bar by bar rather than scanning a finished chart, and include realistic spread, commission and slippage.
Sample size matters more than most people accept. Thirty trades tells you almost nothing; a hundred or more across different market conditions starts to be informative. A strategy tested only in a trending month will fail the first time the market goes sideways.
Track win rate together with average win versus average loss, because neither is meaningful alone — a thirty-five percent win rate with a three-to-one payoff is a strong system. Also record the worst losing streak and the largest drawdown, since those are the numbers that decide whether you can keep following the rules.
This is the same reason Safabot publishes no accuracy percentage: a figure produced by whoever sells the tool is unverifiable. Your own journal, with your own instruments and costs, is the only win rate that means anything to your account.
At least a hundred across varied conditions before the numbers stabilise. Below thirty, the results are mostly noise.
Yes. Consistency matters far more than tooling — the same fields filled in for every trade, including the ones you would rather forget.
Backtesting replays historical data; forward testing runs the rules live on small or demo size. Forward testing catches execution problems — spread, slippage, hesitation — that a backtest cannot show.