Trading robots: how they work and how not to get scammed
A robot (EA, Expert Advisor) is a program that trades on its own following rules. Neither magic nor a curse: it automates a strategy, which can be good or garbage. The problem is that 95% of the ones sold are the latter.
Why so much garbage gets sold
Building a robot with a spectacular backtest is easy: just over-optimise it to the past until the curve is perfect. Building one with a real edge that survives the future is extremely hard. And here's the trick: if you truly had a robot making 200% a year, you wouldn't sell it for €50 — you'd use it quietly. Whoever sells it makes money from the sale, not the trading.
The 7 red flags
- Profit promises: "500% a year", "no losses", "guaranteed gains". Trading guarantees nothing. Run.
- They only show the backtest, not a verified live account (Myfxbook/FX Blue with a long history).
- Backtest at 90% modelling quality (synthetic ticks) instead of real ticks. Here's why that lies.
- A perfect equity curve, a straight line up with no drawdowns. That doesn't exist live.
- It uses martingale or grid with no stop-loss: it grinds small wins for months… until one strong move wipes the whole account.
- Few people can see the code (encrypted .ex5) and they give no explanation of the logic.
- Urgency and scarcity: "only 10 copies", "offer ends today". A sales tactic, not a trading one.
Before putting a real euro into ANY robot —bought or your own— measure it with real ticks, tail Monte Carlo, walk-forward and parameter sensitivity. If you haven't measured it, you don't know what you have. That's exactly what we do in the validation service.
Does that mean all robots are bad?
No. An EA with solid logic, honestly validated and used in the right regime can be an excellent tool (it trades without emotion, 24h, no fatigue). The problem isn't automation — it's the lack of honest validation. Separate wheat from chaff by measuring, not believing.