Backtesting intermediate 10 min read

How to run a proper backtest in MT5

A step-by-step guide to MT5's Strategy Tester. Covers tick data download, modelling quality settings, optimisation pitfalls, and how to read the backtest report without being misled by overfitted results.

Published · Reviewed

MT5’s Strategy Tester is more powerful than MT4’s — it supports real tick data, multi-currency testing, and built-in optimisation. It is also easy to misuse in ways that produce misleading results. This guide covers a correctly configured backtest from start to finish.

Step 1 — Download tick data

Strategy Tester defaults to 1-minute OHLC bars. This produces a “modelling quality” figure of 25-35%, meaning the tester is guessing at intrabar price movement. For any strategy that uses intrabar stops or trailing stops, this is not adequate.

To get real tick data: open MT5, open a chart of the pair you want to test, right-click on the chart → Tick Data Suite (if installed) or use Tools → History Center to bulk-download. Alternatively, select “Every tick based on real ticks” in Strategy Tester settings — this uses the tick data your broker has stored.

Target: modelling quality ≥ 99%. EAs in this catalogue are backtested at 99% with real spread data.

Step 2 — Configure Strategy Tester settings

Open Strategy Tester (View → Strategy Tester or Ctrl+R). Key settings:

  • Expert: select your EA from the dropdown
  • Symbol: the pair (use the exact symbol name your broker uses, e.g. “EURUSD” not “EUR/USD”)
  • Period: the timeframe the EA is designed for (H4 for swing EAs, M1 for scalpers)
  • Model: “Every tick based on real ticks” — not “Open prices only” (for indicators), not “OHLC on M1”
  • Date: use at least 3 years; 5 years preferred for statistical significance
  • Deposit: match the EA’s recommended minimum deposit
  • Spread: “Current spread” is acceptable; tick “Use real spread history” if available

Step 3 — Run and interpret the report

After the backtest completes, open the Report tab. Key metrics to read critically:

Do not focus on: total profit, profit factor alone, number of trades in isolation.

Do focus on:

  • Maximum drawdown (%): the drawdown the strategy produced in this test period. Compare to your risk tolerance.
  • Recovery factor: total profit / max drawdown. Higher is better; below 2.0 signals poor edge quality.
  • Expected payoff (= expectancy): average profit per trade in currency. Positive and above your average spread cost = genuine edge.
  • Sharpe ratio: appears in some broker-labelled Strategy Tester versions; target above 1.5.

Step 4 — Common backtest mistakes

Mistake 1 — Test period too short: testing on 6 months includes only one market regime. A trend EA will look excellent if tested on a 6-month trending market. Test over at least one full cycle (trending + ranging + volatile + calm).

Mistake 2 — Optimised parameters on full history: if you run optimisation and select the best parameters from the same data period, the results are biased. Always reserve a forward-test period (the last 12 months) that was not included in optimisation — this is walk-forward analysis.

Mistake 3 — High modelling quality but wrong spread: a 99% modelling quality test with 0 spread is meaningless. Always test with realistic spread data, especially for scalping EAs where spread is the primary cost.

Mistake 4 — Ignoring the drawdown period: look at the equity curve chart, not just the final profit. A smooth equity curve with low drawdown is more deployable than a higher total profit with a volatile equity curve that most traders would abandon mid-drawdown.