How to backtest a trading strategy (no code), step by step
Backtesting is simple to do and easy to do badly. Here's the honest version — five steps that produce a number you can actually trust, with no programming.
What is backtesting?
Backtesting runs a set of trading rules over historical data to estimate how they would have performed. Done well, it's the cheapest way to reject a bad idea before it costs you money. Done badly, it's a machine for producing false confidence.
The five steps
1. Define your rules
State exactly when you enter, when you exit, and how you manage risk — for example, "enter long when RSI(14) crosses below 30, exit when it crosses above 70, stop at 2×ATR." Vague rules can't be tested.
2. Choose your data and costs
Use a clean, representative dataset and — this is the part most people skip — model commissions and slippage. A strategy that's profitable at zero cost and a loser after realistic costs is a loser.
3. Run the backtest
Simulate the rules bar by bar to produce an equity curve, a drawdown curve and a full trade list. The first number you see should be the one you could actually have had — after costs.
4. Read the right metrics
Net profit alone is misleading. Look at the Sharpe ratio (risk-adjusted return), maximum drawdown (the worst pain), profit factor and the month-by-month returns calendar.
5. Validate out-of-sample
This is what separates a hobbyist from a quant. Confirm the edge holds on data the optimizer never saw, using Walk-Forward analysis, Monte-Carlo simulation and a Deflated-Sharpe check. See the full method in what is overfitting.
The mistakes that ruin backtests
- Overfitting — tuning until the past looks perfect. Detect it here.
- Look-ahead bias — using information that wasn't available at the time.
- Ignoring costs — commissions and slippage quietly turn winners into losers.
- Survivorship bias — testing only on instruments that still exist today.
- Tiny samples — a handful of trades proves nothing.
The no-code way
QUANTHEON Lab lets you build the whole strategy as visual blocks — or describe it in plain English — and runs an honest engine with realistic costs, stops and sizing, then validates it for overfitting in one click. No code, and every run comes with a reproducibility receipt.
FAQ
Can I backtest a trading strategy without coding?
Yes — build it as visual blocks or in plain English on a no-code platform like QUANTHEON Lab and run it with realistic costs, no programming needed.
What makes a backtest realistic?
Modelling commissions and slippage, realistic fills, bar-by-bar stops and sizing, and validating the result out-of-sample.
How much historical data do I need?
Enough to cover several market regimes and produce a meaningful number of trades — a few dozen trades across only a calm market won't generalise.
Related: What is overfitting? · Walk-Forward analysis · Monte-Carlo simulation · Pine Script → no-code