EA Strategy Correlation Test: Before Adding a Second EA

Most traders add a second EA the way they add a second monitor. It feels like more coverage for the same desk. Then a bad week arrives, both systems are leaning the same way at the same hour, and the account takes a hit that neither backtest showed on its own. An EA strategy correlation test is the check that catches this before it costs you an account or a payout, and it belongs before you run two systems side by side, not after.

What is an EA strategy correlation test?

An EA strategy correlation test compares the period-by-period returns of two or more automated systems to see whether they win and lose at the same time. You align their trade histories over an identical date range, convert them into daily returns, and measure how closely those returns move together. If two EAs consistently move in step, you do not have two strategies. You have one strategy running at double size, and the account is carrying a concentration you never chose.

The test is about co-movement, not about whether the strategies look similar in a description. Two systems can use completely different logic and still produce nearly the same return series, because both are paid by the same market condition.

Why can two profitable EAs still be one risk?

Two profitable EAs are still one risk whenever their returns move together, because being profitable and being independent are separate properties and only the second one reduces risk. Both systems can be sound and still share a hidden common factor: dollar direction, a volatility expansion, one specific trading session, the same reaction to scheduled news, or the same behaviour when spreads widen.

The other shared factor is your account itself. Margin, equity and any drawdown rule are pooled. Two EAs that each survive their own worst stretch comfortably can, when those stretches land on the same days, produce a combined equity dip deeper than either system's own history suggests. Diversification only works when the losing periods sit in different places on the calendar.

Do different instruments mean my EAs are uncorrelated?

No. Instrument names tell you very little about correlated risk, and two systems on different symbols can still be the same bet. Gold and the commodity currencies often respond to the same dollar move. EURUSD and GBPUSD share an entire leg of the quote. Two indices in the same region are frequently one trade wearing two tickers.

The reverse is also true. Two EAs on the same symbol, one mean-reverting inside the range and one trading breakouts out of it, can be far less correlated than two trend systems spread across different pairs. Correlation lives in the return series, not the symbol list. Judge by the curves and the timestamps.

How do I compare two EA balance curves?

Export the closed trade history for each system over the same dates and broker conditions, turn each into a daily return series, and compare the series rather than the headline curves. Two rising balance lines always look alike. The information is in the day-to-day changes, not in the level.

• Use returns, not balance. Difference the curve so you are comparing what each system did each day.
• Normalise for size. Express results as a percentage of the account or in R multiples, so the system trading larger lots does not dominate.
• Align the calendar precisely. Same start and end, same server time zone, and non-trading days recorded as flat rather than skipped.
• Use a rolling window as well as the full period. Correlation drifts. Systems that look independent in quiet conditions often converge under stress.
• Inspect the worst days directly. Rank each system's largest losing days and read off what the other system did on those exact dates.

How do I check whether two EAs trade at the same time?

Compare entry timestamps and open exposure, not just outcomes. Build a picture of when each system is in the market, then find the windows where both hold positions in the same direction on related instruments.

• Plot entries by hour of day for each EA and look for a shared session.
• Count how often both systems hold open positions simultaneously, and in which direction.
• Net the exposure by currency rather than by pair, so two positions that are really one dollar bet show up as one.
• Measure how often both open within a short window of each other, which usually points to a shared trigger such as a session open or a scheduled release.

Why is correlated drawdown the real danger on a prop account?

Correlated drawdown is the real danger because prop firm rules are measured at the account level, and the rule does not care that a loss arrived from two separate systems. A daily loss limit, a static or trailing maximum drawdown, or an equity-based threshold sees one number. Correlated EAs stack their bad days into that single number instead of spreading them out.

Rule types vary by firm and change over time, so check your current terms directly rather than trusting a figure quoted anywhere else, including here. What matters is the mechanism: some firms measure on closed balance, some on equity including floating positions, some reset daily and some trail the peak. Each treats simultaneous drawdown differently. No correlation test promises a challenge pass. It removes one avoidable way to fail.

How do I set portfolio-level exposure limits across EAs?

Decide the risk budget for the whole account first, then divide it between systems, instead of sizing each EA in isolation and hoping the total behaves. The sum of individually sensible settings is not automatically a sensible portfolio.

• Set a maximum acceptable loss for the account over a single day and treat it as the budget every system draws from.
• Cap net exposure per currency and per risk theme, not per symbol.
• Cap the number of concurrent open trades across all systems, not just within each one.
• Decide in advance which system is disabled first if the budget is breached, and write it down.
• Resize everything whenever you add a system. A third EA changes the correct size for the first two.

What should I test before adding a second EA?

Run the candidate alongside your existing system on a demo account at the same broker, over a period that includes at least one stressed market, before it touches a funded account. Then judge the combined result, not the individual ones.

1. Build a common-period return series for both systems from closed trade history.
2. Measure correlation over the full period and over a rolling window.
3. Overlay the drawdown periods and confirm the deep parts do not line up.
4. Model the combined curve at the sizes you actually intend to run.
5. Verify margin usage and execution when both are active at once.
6. Check the operational layer. Two EAs on one terminal, one feed and one server share failure modes as well as returns.

The JPTC EA Hub for MT4 and MT5 is a one-time purchase at 797 euro, with the Pro version at 1,497 euro and the bundle at 2,499 euro, all including VAT, and a 14-day refund window provided the software has not traded. You run it on your own account at your own broker. JPTC holds no funds and has no withdrawal access. We also publish free forex and gold signals on Telegram, funded by our partner brokers. If you copy trades between accounts, the copier works across many platforms including MT4, MT5, cTrader, DXtrade and TradingView, so this correlation work carries over wherever each system runs. Our results page is there so you can apply the same checks to our systems.

If you are about to add a second system, start with the method. Our research notes cover how we align trade histories, measure rolling correlation and stress the combined curve, so you can run the same test on whatever you already trade before you turn anything on.