Three weeks into a stress test: +80% profit and why I am not celebrating

On August 1st I connected a live account to my Veteran Army Gold EA and published it as a signal. The setup is deliberately harsh: the Aggressive preset with ReferenceCapital at 10,000 on a 5,000 USD deposit - roughly twice the risk of the Aggressive profile on the product page, and far above the Conservative default the EA ships with. The point of the account is to show the engine under stress, on real money, at a regulated broker.

Three weeks in, the numbers look like this: +80% growth, 74 trades, profit factor 2.86, worst balance drawdown so far 11.9%. No deposits, no withdrawals, no manual trades - the zeros are on the signal page, so the growth figure means exactly what it says.

Here is the part I actually care about. The win rate is exactly 50%. Half the trades lost money. The account grew anyway, because the average winner (166 USD) is nearly three times the average loser (58 USD). That asymmetry is the whole design: the portfolio does not try to be right more often, it tries to lose small and win large. In these three weeks it already produced both tails - a streak of 15 wins and a streak of 10 losses. Both are normal. Both will happen again.

And here is why I am not celebrating. MQL5 has flagged this signal with "too much growth in the last month indicates a high risk" - an automated warning, and a correct one. Stress testing puts the expected maximum drawdown for this configuration near 40%. That drawdown has not arrived yet. Three good weeks tell you very little about a system; the interesting post will be the one I write from the bottom of the first real drawdown, because the account will still be running and the record will still be public. It will not be deleted, reset or hidden - that commitment is in the signal description and it is the only part of this experiment I consider non-negotiable.

If the balance keeps growing, risk settings will be reduced along the way - the point of the account is the engine, not the leverage - and every change will be posted on the signal page before it is made.

The signal, with the full setup disclosed: https://www.mql5.com/en/signals/2384638

The EA it runs, at its recommended (much calmer) settings: https://www.mql5.com/en/market/product/188603

Judge it over months, not weeks. I will post the next update either at a new balance milestone or from inside the first drawdown - whichever comes first.