What backtesting means in practice
At its core, a backtest answers a simple question: if I had followed these rules in the past, what would have happened? The rules can be discretionary guidelines you practise bar by bar, or precise coded logic an engine can run automatically.
The value is not nostalgia. It is evidence. You want enough structure to learn whether an idea has any historical support before you risk real capital.
How a backtest works
You define rules for entries, exits, position sizing and risk. The software then steps through historical data, simulates the trades those rules would have generated, and records the results: equity, drawdowns, trade list and summary stats.
In manual tools, you are the decision engine: you replay charts, place simulated orders, and journal outcomes. In automated tools, the engine applies your coded or no-code rules across thousands of bars without you clicking each trade.
- Choose a market, timeframe and history window
- Define entry, exit and risk rules as clearly as you can
- Simulate trades forward through history (no peeking ahead)
- Review metrics, trade quality and whether the idea survives costs
Manual vs automated backtesting
Manual backtesting means replaying charts and recording trades by hand or in a replay journal. It is slower, but excellent for chart-based traders who need to practise execution, emotion and discretionary reads.
Automated backtesting uses code or a rule builder to test many trades quickly. It is better for systematic rules, optimisation and large samples. Many traders use both: automate the rule research, then manually practise the discretionary parts of their process.
Key metrics to review
No single number tells the whole story. A high win rate with terrible reward-to-risk can still lose money. A strong net profit with an unbearable drawdown can still be untradeable for a human.
Look at several metrics together, and always read them next to the assumptions that produced them (costs, fills, sample size).
- Net profit and compound annual growth rate (CAGR)
- Maximum drawdown and how long drawdowns lasted
- Win rate, average win and average loss
- Profit factor and expectancy per trade
- Risk-adjusted returns such as Sharpe or Sortino
- Number of trades and whether they span different market regimes
What good historical data looks like
A backtest is only as good as its data. Check how far history goes, whether intraday or tick detail matches your strategy, and whether costs are modelled. For stocks, ask about splits, dividends and delisted symbols.
More history is not automatically better if the market structure changed, but too little history makes it easy to fool yourself with a short lucky window.
What a backtest cannot tell you
A backtest describes the past under specific assumptions. It cannot promise that the same edge will persist. Markets change, liquidity changes, and your live psychology is not in the simulation.
Execution also differs from the model: slippage, partial fills, latency and platform quirks appear in live trading. Treat backtests as evidence, not guarantees, and follow them with forward testing on demo or small size.
A sensible next step
If you are new, start by defining one simple strategy in writing, backtest it manually or with a beginner-friendly tool, and review the metrics above without optimising yet. Learn the failure modes first.
When you are ready to choose software, use a clear framework: realism, real cost, market coverage, interface, reliability, core tools, extras and support. Compare options side by side, then pick the tool you will actually open every week.