Most EA marketing sells a feeling: switch it on, walk away, get paid. That is not how trading robots work, and believing it is how most people lose money with them. Read this before you buy anything, from anyone, including us.
Every EA has losing months
An Expert Advisor is a set of rules that places trades for you. The rules do not change when the market does. A breakout system earns in active, trending sessions and gives some back in quiet, choppy ones. A mean-reversion system does the opposite. No rule set wins in every kind of market, so every honest EA has losing weeks and losing months.
The question is not whether it will lose, but how much, how often and for how long. That is why every figure on this site starts with drawdown.
Win rate hides the size of the losses
A high win rate is the easiest number to sell and one of the least useful on its own. What matters is the win rate together with the size of the average win and the average loss.
9 wins × $10 = +$90
1 loss × $100 = −$100
Every 10 trades: −$10, at a 90% win rate
This is not a made-up extreme. In one of our Lab tests, a gold EA’s entry signal won 96.8% of its trades and still lost $516 over the test. Read that verdict. Look at profit factor (gross profit divided by gross loss) and expectancy (the average result per trade) instead.
Backtests can be tuned to look perfect
A backtest runs the rules over past prices. With enough inputs to adjust, almost any rule set can be made to look brilliant on some stretch of history. This is curve fitting: the settings end up describing the past instead of a real edge.
Signs to look for:
- A smooth equity line with almost no drawdown.
- Results for one short window, with nothing before or after it.
- Many inputs with oddly precise values.
- A result that collapses when you move the start date or nudge one setting.
- Tests on one-minute bars or generated ticks instead of real tick data.
The cure is out-of-sample data: a period the settings never saw. If a vendor does not show one, ask why. See how the Lab tests it.
Your risk setting decides your drawdown
The same EA can be calm or brutal depending on how much you risk on each trade. Double the risk and you roughly double the drawdown. Losses also hurt more than gains help, because you recover from a smaller balance:
| Drawdown | Gain needed to break even |
|---|---|
| 10% | 11.1% |
| 20% | 25% |
| 30% | 42.9% |
| 50% | 100% |
| 70% | 233.3% |
A 50% drawdown needs a 100% gain just to get back to the start. Decide the largest drawdown you can live with first, then set the risk per trade to match.
What an EA can and can’t do
It can
- Follow its rules exactly, every time, without fear or hesitation.
- Trade sessions you cannot sit through.
- Put a stop-loss and a sized position on every order.
- Be tested on years of data before you risk money.
It can’t
- Adapt when the market changes character.
- Guarantee a profit, or even a profitable year.
- Protect you from a risk setting that is too high.
- Decide for you when to switch it off.
Before you buy: seven checks
Before you pay for any EA, from anyone, look for all seven:
- A live signal, verified by a third party, running for months rather than weeks.
- Results on an out-of-sample period, not only the showcase window.
- The maximum drawdown, and the risk per trade that produced it.
- A hard stop-loss on every trade.
- No martingale or grid, unless its worst case is sized and shown.
- Real costs in the tests: spread, commission and swap.
- A risk setting at which you could live through the worst drawdown twice.
If an EA clears all seven, it is worth a closer look. It is still not a promise.
Education only; nothing here is investment advice. Trading forex and CFDs on margin carries a high risk of losing money and is not suitable for everyone.