Ecosystem Academy · Process
How to Backtest a Trading Strategy (Without Fooling Yourself)
Backtesting is how you find out whether your idea is an edge or a hunch — before a funded account's drawdown rules find out for you. Here's the honest, step-by-step process.
On this page
1. Write the rules down first
A backtest is only as good as the rules behind it. "Buy when it looks bullish" is untestable. Every rule must be binary — answerable yes/no from the chart with no judgement calls:
- Entry: the exact condition that triggers a trade (e.g. "price closes back above VWAP after touching it, RSI between 40–55").
- Stop loss: the exact price logic (e.g. "below the pullback low").
- Take profit: fixed R-multiple, structure target, or trailing rule — pick one and write it down.
- Position size: fixed fraction of account risk per trade (0.5–1%).
- Filters: sessions, news windows, max trades per day.
2. Collect a real sample
Scroll the chart back, hide the future, and step forward candle by candle, logging every trade your rules produce — wins and losses, with no skipping the ugly ones. A spreadsheet is enough: date, market, entry, stop, target, result in R.
Aim for at least 100 trades, and make sure the sample covers different conditions — trends, ranges, quiet weeks, volatile weeks. A strategy tested only in a bull run is a coin flip everywhere else.
3. The numbers that matter
Expectancy
Average R per trade: (win rate × avg win) − (loss rate × avg loss). Positive = edge. This is the one number that decides everything.
Max drawdown
The worst peak-to-trough losing stretch, in R. Compare it to your funded account's drawdown rules — if the backtest shows 8R of drawdown and your daily limit is 5%, the strategy needs resizing.
Win rate
Meaningless alone — a 40% win rate with 3R winners beats a 70% win rate with 0.5R winners. Always read it next to average win/loss.
Losing streaks
Count the longest run of consecutive losses. You will live through it again live — decide now whether you'll keep taking the signals.
4. Overfitting: the silent killer
Overfitting is tuning your parameters until the backtest looks perfect — and discovering live that you memorised history instead of finding an edge. Defences:
- Keep rules few and simple. Every extra condition is another chance to curve-fit.
- Split your data. Build the strategy on one period, then test it unchanged on a different period (out-of-sample). If performance collapses, it was fit, not found.
- Be suspicious of perfection. A 90% win rate with tiny drawdown in a backtest is almost always a mistake or a fit, not a discovery.
5. From backtest to funded account
- Backtest until you have 100+ logged trades and positive expectancy.
- Forward-test on a demo for 2–4 weeks — live spreads, live fills, live emotions. Expect results to be somewhat worse than the backtest; that's normal.
- Map the numbers to the rulebook. Your strategy's max drawdown and losing streaks must fit comfortably inside the funded account's daily and max drawdown limits, with room to spare.
- Then take the evaluation. With Pay After Passing, even this step is risk-free — you only pay the fee once you've passed.
6. FAQ
Manual or automated backtesting?
Manual (bar-by-bar replay) is slower but teaches you the strategy's feel and is honest about discretion. Automated testing is faster and covers more data, but only works for fully mechanical rules. Most traders benefit from doing both.
How many trades make a backtest valid?
100 is a working minimum; 200–300 is far better. The sample must also span different market conditions, not just one good month.
My backtest is great but live results are worse. Why?
Usually slippage, spread, skipped signals, or hesitation — the human factor. That's exactly what the forward-test stage is for. If live results are dramatically worse, re-check for overfitting.
Take your proven strategy to a funded account
Pay After Passing: run the evaluation with zero upfront cost and pay the fee only after you've passed.