MT5 Multi-EA Backtest: Testing Multiple Expert Advisors on One Account (2026)

The MT5 Strategy Tester runs one EA at a time. Even its multi-symbol mode is not a portfolio backtest. Here is what MT5 can and cannot show you when you run several EAs on one account — and how to test the combination for real.

If you run more than one Expert Advisor on a single MetaTrader 5 account, the most important number you need — the drawdown the combination would have produced — is the one number MT5 will not give you. The Strategy Tester is built to evaluate one EA in isolation. That is exactly the wrong unit of analysis once your EAs share a balance, share margin, and draw down at the same time. This guide explains precisely where MT5 stops, why it stops there, and how to run a true shared-balance portfolio backtest from the reports you already have.

By the FXOptimize research team. Last updated 2026-06-14.

What the MT5 Strategy Tester actually does

The MT5 Strategy Tester loads exactly one Expert Advisor per run. You select an EA, a symbol, a timeframe, a modelling mode (1-minute OHLC, every tick, or every tick based on real ticks), a date range and a deposit — and it simulates that single EA against historical data. MT5 improved meaningfully on MT4 here: it supports a multi-currency / multi-symbol mode, a genuine tick-level model, and a faster optimization engine. Those are real upgrades.

But the multi-symbol capability is frequently misunderstood. It lets one EA open trades across several instruments inside a single test. It does not let you load two different EAs into the same test and have them trade together off one balance. There is no "add a second Expert Advisor" button, because the tester's model is one-program-per-run by design.

Why even multi-symbol mode is not a portfolio backtest

When traders discover MT5 can trade multiple symbols, they often assume that solves the portfolio problem. It does not, for a simple reason: a portfolio is not "several symbols", it is several strategies sharing one balance. The thing that makes a portfolio behave differently from its parts is the interaction between the EAs, and a single-EA multi-symbol test has no second EA to interact with.

The workaround most people reach for is to backtest each EA separately, export the reports, and add the equity curves together in a spreadsheet. That is the single most common — and most misleading — way to "combine" EAs. It breaks in three places:

• Dynamic lot sizing is balance-dependent. Any EA that risks a percentage of equity (or scales lots with balance) sizes each trade off the account balance at that moment. On a shared account, every other EA's open trades and closed P&L change that balance. Summed independent curves assume each EA traded its own private account, so the position sizes — and therefore the returns — are simply wrong.
• Drawdowns stack. Two EAs that each show a 12% maximum drawdown alone can punch well past 25% together if they tend to lose in the same conditions. Adding curves hides this entirely, because the peak-to-trough of a sum is not the sum of the peak-to-troughs.
• Margin and stop-outs are shared. A combined book can hit a margin call or a prop-firm daily-loss limit during a simultaneous drawdown that neither EA would have triggered on its own. End-of-bar equity sums never reveal the intraday excursion that actually breaches the limit.

The three things a real multi-EA backtest must model

To know what a basket of EAs would actually have done on one MT5 account, the simulation has to reproduce the account, not the EAs:

• 1. Chronological event replay on one balance. Every entry and exit from every EA is merged into a single time-ordered stream and applied to one shared balance, so each trade is sized and settled against the real running equity at its timestamp.
• 2. Dynamic lot scaling off shared equity. Fixed-lot and balance-scaled EAs are detected and handled differently, so a percent-risk EA grows and shrinks its size as the combined balance moves — not as its solo balance would have.
• 3. Worst-case intraday drawdown. Using daily high/low data rather than close-to-close equity, so the metric reflects the heat that would actually have hit margin or a daily-loss rule — the number that decides whether a prop-firm challenge survives.

Only once those three are in place do correlation, Monte Carlo stress testing and Pareto optimization mean anything — because they all operate on the combined equity curve, not the individual ones.

How to run a real MT5 multi-EA backtest

You do not need to re-run anything inside MT5. You already produced the raw material when you backtested each EA. The workflow is:

• In MT5, backtest each EA as you normally would, then right-click the tester report and save it as an HTML/XML report (one file per EA, per symbol if the EA is single-symbol).
• Load all of those reports into a portfolio simulator that does shared-balance, chronological replay — not curve summing.
• Read the combined result: portfolio drawdown, the correlation matrix (which EAs truly diversify versus secretly duplicate each other), and the mixes that sit on the efficient frontier.

Common mistakes when combining MT5 EAs

• Summing equity curves in Excel. The fastest way to a portfolio that looks great and trades nothing like reality. Covered above — it gets both return and drawdown wrong.
• Ignoring correlation. Two profitable EAs that are 0.8 correlated barely diversify; you are effectively running one strategy at double size. The pairs that help are the near-zero and negative ones — see how drawdowns stack.
• Trusting close-to-close drawdown. End-of-bar equity understates the real intraday excursion. For prop-firm accounts this is the difference between passing and breaching — see the FTMO daily drawdown mechanics.
• Optimizing on one history. A single backtest window is one sample. Monte Carlo trade-shuffling and walk-forward validation tell you whether the edge survives reordering and out-of-sample periods.

Frequently asked questions

Can the MT5 Strategy Tester backtest multiple EAs at once?
No. It loads exactly one Expert Advisor per test. Multi-symbol mode lets a single EA trade several instruments, but it cannot run two different EAs together on one shared balance — which is what a portfolio backtest requires.

Why can I not just add up the equity curves of separate MT5 backtests?
Because dynamic-lot EAs size positions off the running account balance. On a shared account every EA changes the balance the others trade from, so summing independent curves overstates returns and hides how drawdowns stack when several EAs lose together.

How do I run a real multi-EA portfolio backtest from MT5 reports?
Export each EA as an MT5 backtest report, then combine them in a simulator that replays all trades chronologically on one shared balance with dynamic lot scaling. FXOptimize does this in the browser for free, and your files never leave your machine.

Related guides

• How to Run Multiple EAs on One MT4/MT5 Account Without Blowing Up
• MT4 Backtest Portfolio Analysis: Why the Strategy Tester Falls Short
• How to Build an EA Portfolio in 2026: The Complete Guide