How to Forward Test an MT5 EA Before Going Live

You've backtested your EA over 5 years with real tick data. The results look solid — PF 1.45, max DD 7.2%, consistent OOS performance. Now what?

The next step before risking real money is a forward test on a demo account. Backtesting tells you how an EA performed on historical data. Forward testing tells you how it performs in real-time market conditions — with live prices, real spreads, and the psychological reality of watching your account balance fluctuate.

This guide walks you through the complete forward testing process.

---

Why Forward Testing Is Not Optional

Backtesting has limitations that forward testing exposes:

Limitation	What It Means
Spread assumptions	Backtest uses historical average spread; live spread varies
Slippage	Backtests assume perfect fills; live orders have slippage
Data quality	Even "real tick" data can have gaps or errors
Execution latency	Backtest is instant; live orders travel to the broker
Psychological reality	A live account moving against you feels different on a screen

A 4–8 week forward test on a demo account catches most of these gaps. If the EA behaves significantly differently from its backtest, you know before losing real money.

---

Step 1: Set Up a Demo Account on Your Live Broker

Why your live broker specifically? Because you want to test under identical conditions to where you'll trade live: same server, same symbol names, same spread profile, same execution speed.

Most brokers offer instant demo account creation:

• XM: Create at xm.com → Demo Account (note: use GOLD not XAUUSD)
• Exness: Create at exness.com → Demo
• HFM: Create at hfm.com → Demo

Match the account type to what you plan to use live — if you'll trade a Standard account, demo on Standard. If you plan to use Raw/ECN, demo on Raw/ECN.

---

Step 2: Install MT5 and Configure It Identically to Your Backtest

Download MT5 from your broker (each broker has their own build). Install it on your computer or VPS.

Key settings to match your backtest:

• Same EA parameter values as your backtest (don't change them for the forward test)
• Same RiskPercent (start with 1% or lower)
• Same MaxSpread setting
• Same chart timeframe (H1 for most XAUUSD EAs)

This is important: the forward test should use the exact same parameters as your best-performing backtest run. If you change parameters during forward testing "to improve results," you've started optimizing on live data — which is cheating.

---

Step 3: Run the EA on Demo for 4–8 Weeks

Attach the EA to the chart, enable Auto Trading, and let it run without interference.

Minimum forward test duration by strategy type:

Strategy Type	Minimum Test Period	Target Trade Count
H1 trend-following (XAUUSD)	4 weeks	20–40 trades
H1 breakout (EURUSD)	4 weeks	15–30 trades
Scalping (M5/M15)	2 weeks	100+ trades

The more trades you accumulate, the more statistically meaningful the forward test. 10 trades is not enough to draw conclusions about a strategy. 50 trades gives you reasonable signal.

---

Step 4: Track These Metrics Weekly

Open a simple spreadsheet and record these numbers at the end of each week:

Metric	What to Measure
Net P&L	Total gain/loss in USD and %
Trade count	Number of trades taken
Win rate	% of trades closed in profit
Avg win / Avg loss	In pips or USD
Max drawdown (this week)	Largest peak-to-trough during the period
Spread paid	Check order history for average open/close spread

Also note: did the EA behave as expected? Did it skip trades you expected it to take? Did it hold positions longer than the backtest suggested?

---

Step 5: Compare Forward Test Results to Backtest Expectations

After 4–8 weeks, compare your forward test numbers to the monthly averages from your backtest.

Acceptable variance:

• ±30% on monthly P&L
• ±10% on win rate
• ±20% on max drawdown

Investigate if:

• Monthly P&L is consistently below the backtest expectation (check spread and slippage)
• Win rate is significantly lower (check signal logic and execution)
• Drawdown is notably higher (check if max position size or spread filtering is working)

Abandon if:

• The EA is consistently unprofitable after 50+ trades
• Drawdown exceeds 2× the historical max
• The EA behaves erratically (multiple unexpected opens/closes)

---

Step 6: Transitioning to Live

If your forward test passes, transition gradually:

1. Start with 0.25–0.5% risk (not the 1% you forward tested)
2. Run for 2 more weeks at this reduced risk
3. If results match expectations, scale to 1% risk
4. Scale further only after 3+ profitable months

This gradual approach protects your capital while confirming the EA performs as expected on a real (non-demo) account.

---

Common Forward Testing Mistakes

Mistake 1: Stopping the forward test after a bad week One losing week proves nothing. Stick to the minimum test period unless the EA clearly malfunctions.

Mistake 2: Tweaking parameters during the test This invalidates the test. Lock the parameters and don't touch them until the test period ends.

Mistake 3: Testing on the wrong demo server If you test on a generic MT5 demo but plan to live trade on XM, your spread and execution data won't match. Always use the same broker.

Mistake 4: Not running on a VPS If your forward test PC goes offline for even a few hours, you miss trades and distort the results. Use the same VPS setup you plan to use live.

---

Ready to Forward Test?

Download the GOLD EMA ATR EA — our flagship free XAUUSD H1 EA with a 5-year backtest PF of 1.45 and max DD of 7.2%. Set it up on a demo account today and run the 4-week forward test before going live.

See our demo to live transition checklist for a full pre-live verification process.