Stacking Two EAs on the Same Market Is Not Diversification

A buyer who is happy with one gold EA often buys a second one and runs both on XAUUSD in the same account, calling it diversification. It usually is not. Two trend followers on the same symbol are one bet taken twice: their drawdowns arrive on the same days, the combined risk is roughly the sum of the two, and the smoothing you were hoping for is close to zero.

Most gold EAs on the marketplace react to the same thing: a move that has already started. When gold falls 40 dollars in an hour, every trend follower with an open long loses at once. Correlation between two such products is high not because the sellers copied each other but because the input is the same price series and the mechanism is the same reaction to it. Running both does not halve your drawdown; it doubles your exposure to the same day.

Either different markets that do not move together, or different mechanisms that lose at different times. A trend follower and a mean-reversion strategy on the same symbol are closer to opposites: one loses in the ranges where the other earns, and the other loses on the breakout where the first earns. Neither pairing is guaranteed, but at least the losses have a reason to arrive on different dates. Two trend followers on gold have none.

Beyond correlation, there are mechanical problems that appear only when EAs share an account. Magic numbers can collide, and then one EA starts closing or trailing the other's positions. Margin usage doubles, so the lot the first EA computed at a 3,000-dollar balance no longer fits once the second EA is holding its own positions. Opposite signals turn into a locked hedge: one long, one short, net exposure zero, and swap charged on both sides every night until they resolve. And a news filter belongs to the EA that has one; the other EA trades straight through the release the first was told to avoid.

Run each backtest separately over the same period, same broker data, same balance. Export the daily equity from each report and turn it into a daily profit and loss series. Add the two series day by day, then look at two numbers in the sum: the worst single day and the worst drawdown. If the combined drawdown is close to the sum of the two individual drawdowns, the strategies are correlated and you have built one bigger position. As a rough illustration: two EAs each with a 1,000-dollar worst drawdown that combine to 1,900 are the same bet; if they combine to 1,200, they are worth running together.

Lay the monthly results of both EAs side by side and count the months where both were negative. If nearly every red month for the first is a red month for the second, adding the second increases the size of the bad months and does little for the good ones. The thirty-second version of this test is to compare the equity charts and see whether the dips line up. They usually do.

Add a second EA only if it loses in different months than the first. If you cannot show that from the backtests, you are increasing risk, not spreading it, and the cleaner way to increase risk is to raise the preset on the EA you already own. That at least keeps one set of rules, one magic number and one news filter in charge of the account.

Our three best-selling EAs: GOLD NEURON (AI) https://www.mql5.com/en/market/product/187329 · ATLAS PORTFOLIO https://www.mql5.com/en/market/product/182751