What is an Expert Advisor (EA)?

An Expert Advisor is software that runs inside MetaTrader and executes a trading strategy automatically, following the entry and exit rules it was given. It has no judgement of its own and executes a bad rule as faithfully as a good one.

What does out-of-sample (OOS) mean?

Out-of-sample data is a slice of history that a strategy was never tuned on. Running a tuned strategy against it checks whether the rules found a real pattern or just memorised noise in the data they were fitted to.

Drawdown is the decline from an account peak to its next low, expressed as a percentage. If an account reaches 10,000 and falls to 8,000 before recovering, that is a 20 percent drawdown. It is the standard measure of how much loss a system puts you through.

Algorithmic Trading Glossary | Rubotics Academy

Expert Advisor (EA)

A program that places trades for you inside MetaTrader, following rules you set.

An Expert Advisor is a piece of software that runs inside the MetaTrader platform and executes a trading strategy automatically. It watches the market, applies the entry and exit rules it was given, and places the orders without you clicking anything. An EA does exactly what its rules say, nothing more. It has no judgement and no opinion about the market.

Why it matters

An EA removes hesitation and inconsistency from execution. It also executes a bad rule just as faithfully as a good one, which is why validating the rules first is the whole job.

Backtest

Running a strategy against historical price data to see how it would have behaved.

A backtest replays past market data through a strategy and records what it would have done. It is a simulation, not a result: the trades never happened, and the fill prices, spreads and slippage are estimates. A backtest tells you whether a set of rules was ever coherent. It does not tell you what the strategy will do next.

Why it matters

A backtest is the cheapest way to reject a bad idea. Treating it as a forecast rather than a filter is the single most common mistake in automation.

Out-of-sample (OOS)

Testing a strategy on data it was never tuned on, to check the rules were not just curve-fitted.

When you tune a strategy you use one slice of history (the in-sample data). Out-of-sample data is a separate slice the tuning never touched. Running the tuned strategy on that untouched slice checks whether it found something real or just memorised the noise in the first slice. If a strategy only works in-sample, it learned the past instead of learning the market.

Why it matters

Any strategy can be made to look excellent on the data it was fitted to. Out-of-sample testing is what separates a genuine pattern from an expensive coincidence.

Drawdown

The drop from an account peak down to its next low, measured as a percentage.

Drawdown measures decline: if an account reaches 10,000 and then falls to 8,000 before recovering, that is a 20 percent drawdown. It is the standard measure of how much pain a system puts you through on the way to wherever it ends up. Maximum drawdown is the worst such drop over a period. Floating drawdown counts open positions that are currently down, not just closed ones.

Why it matters

Drawdown, not profit, is what decides whether a system is survivable. It sets your position size, and it is the number a prop firm will fail you on.

Prop firm

A company that funds traders with its own capital, under strict rules, for a share of the results.

A proprietary trading firm gives a trader access to a funded account instead of the trader risking their own capital. Access is normally earned by passing an evaluation, and it comes with hard rules: a maximum daily loss, a maximum overall drawdown, and often minimum trading days. Breaking any rule ends the account regardless of how profitable it was.

Why it matters

Prop rules are constraints your system must be engineered around from the start. A strategy that ignores the daily-loss limit will eventually breach it, whatever its long-run numbers look like.

Heatmap

A colour grid showing how a strategy performed across many parameter combinations at once.

A heatmap plots one strategy parameter against another and colours each cell by how that combination performed. Instead of reading one result, you see the whole landscape. What you are looking for is a broad stable region where neighbouring settings all behave similarly, not an isolated bright cell surrounded by poor ones.

Why it matters

A single lucky setting sitting next to bad neighbours is fragile: a small shift in the market lands you in a bad cell. A wide stable plateau is what robustness actually looks like on a chart.

Expert Advisor (EA)

Backtest

Out-of-sample (OOS)

Drawdown

Prop firm

Heatmap

Ready for the structured version?