Guides

Choosing, comparing, running, and retiring EAs — risk before returns.

Build MT5 EAs without code and take them from idea to backtest.

Test, optimise, and evaluate EAs before they touch real money.

Platform setup, EA operation, and troubleshooting on MetaTrader 5.

Data-backed experiments and analysis across EAs, symbols, and strategies.

An MT5 backtest report is a claim about one set of trades on one price history, and both halves move the numbers. We read all 23 reports published here: 14 of the 20 that publish both slices flatter the period the EA was selected on, one report covers only 303 of its own 315 trades, and the same EA re-run on a different data source went from 1.35 and a pass to 1.13 and a fail.

Assemble an Expert Advisor from blocks in a browser, compile it to a real .ex5, and test it before it ever sees money. Includes the number nobody quotes: what the 84 shipped starter templates score at their stock settings, measured from the Builder's own records.

Every EA published here closes its backtest with a profit factor above 1 — and ten of the fourteen still had a losing calendar year. Here is what the full closed-trade record says about when an EA makes money and what you have to sit through to collect it.

Comparing an EA against discretionary trading usually turns into an argument about judgement. Replaying 9,702 published trades under the stop rules a human would actually use gives a sharper answer: switching off after three losses in a row would have cost 93% of the catalogue's result.

Wire a MACD crossover Expert Advisor from blocks in a browser, compile it to a real .ex5, and read its numbers honestly. Includes the result that should worry you most: the Builder's own parameter search found a MACD setting that scored 1.14 out of sample, then collapsed to 0.74 on the window before it.

We ran parameter searches on eight of our own strategy templates and adopted none of them. The useful part is how each one failed: the biggest in-sample winner fell the furthest, one design passed a single out-of-sample check and still collapsed, and two never beat 1.0 even on the period they were fitted to. Here is the procedure, and the reading that stops you shipping the top row.