Backtest Tools

Who This Page Is For

• System traders building rule-based strategies for their own accounts or prop evaluations.
• Discretionary traders who want to test core ideas before going live with size.
• Investors who want a clear framework to judge whether a system has been tested properly.
• Anyone tired of “curve-fit” backtests that collapse as soon as real money is at risk.

What Backtests Are Really For

• Check if an idea has any edge at all once costs, slippage and realistic rules are included.
• Estimate key numbers: win-rate, average R per trade, drawdown depth and length.
• See how a system behaves in different regimes: trends, ranges, shocks, low liquidity.
• Give you the confidence to follow rules when live trading feels uncomfortable.

A good backtest does not guarantee profits. It gives you a realistic picture of how bad things can get and whether you can survive that path.

1. Define the Idea

• Describe the behaviour you want to capture: breakout, mean-reversion, trend, regime switch.
• Specify markets and sessions: for example ES and NQ US RTH, major FX in London + NY overlap, or crypto 24/7.
• Write rules in plain language before touching any software.

2. Build a Testable Spec

• Entry rules, exit rules and filters (indicators, time-of-day, news, volatility).
• Risk rules: stop distance, target logic, trailing rules, daily loss and position limits.
• Data granularity: tick, 1-minute, 5-minute, daily, etc., based on how the system trades.

3. Configure the Backtest Engine

• Set correct contract specifications, lot sizes and tick values for each instrument.
• Apply realistic commissions and a slippage assumption that matches how you actually trade.
• Choose the historical window – usually several years covering different volatility regimes.

4. Run & Validate

• Check raw trades: are entries/exits happening exactly where your rules say they should?
• Look for obvious bugs: trades on holidays, orders during forbidden times, position size jumps.
• Only when the behaviour is correct do you look at the performance numbers.

Key Metrics to Track

• Net profit in R: total reward relative to average risk per trade.
• Max drawdown: depth and length of equity curve pullbacks.
• Win-rate & payoff ratio: how often you win and the average size of winners vs losers.
• Trade frequency: number of trades per month and per instrument.
• Exposure: time in the market, overnight vs intraday, reaction to gaps.

Behaviour You Want to See

• Equity curve that survives different market regimes without catastrophic collapses.
• Drawdowns that are painful but survivable with your real account size.
• Performance that is not entirely driven by one short period or one market.
• Reasonable sensitivity when you slightly change parameters such as indicator lengths.

If a small parameter change destroys the entire backtest, the system is probably curve-fit.

Out-of-Sample Logic

• Split history into at least two parts: one to build and one to verify.
• Develop parameters using the first segment; then lock rules and test on the second.
• Accept that performance usually drops when you move to out-of-sample – that is normal.

Walk-Forward Ideas

• Roll the calibration window forward over time, re-optimising on recent data.
• Track the combined equity curve of each walk-forward step as if it were live trading.
• Use simple, robust parameter ranges rather than chasing the single “best” value.

Simple Sanity Checks

• Reverse the strategy (go long where it used to be short) – it should not produce a great equity curve.
• Randomise entries slightly in time or price; edge should weaken but not vanish completely.
• Run tests on closely related markets to see whether behaviour is at least somewhat consistent.

These checks will not make a bad idea good, but they expose fragile, over-fitted systems before they damage live accounts.