Automated Trading Glossary — EA, MT5, Backtest, VPS

Automated trading on MetaTrader sits at the intersection of several disciplines — strategy design, statistical validation, software development, operational hygiene, and risk management — and most retail attempts fail because they engage only one or two of those layers. This cluster collects the vocabulary you need across the full lifecycle from initial EA concept to sustained live operation, with realistic framing about where the failure modes typically occur.

The entry point is the EA architecture. Every Expert Advisor, regardless of complexity, decomposes into four layers: signal (when to enter), filter (when not to enter — news filter, regime filter, session filter), risk (how much to size each trade — position-size cap, daily loss limit, drawdown trigger), execution (how to submit and manage orders — market execution, fill rate, slippage tolerance). Understanding which layer is responsible for any observed behaviour is the foundational debugging skill. Most retail discussion focuses on the signal layer because it is the most marketing-visible, but live performance is typically dominated by execution and risk discipline.

The MetaTrader 5 ecosystem provides the runtime. MT5 hosts the EA, MetaEditor compiles the MQL5 source into the .ex5 binary, the Strategy Tester replays historical data through the strategy logic for backtest evaluation, the Expert Journal logs operational events and errors, Market Watch governs which symbols receive live ticks. The MQL5 Marketplace is the highest-trust distribution channel for commercial EAs; vendor-transparency signals (versioned changelogs, documented defaults, identifiable principals) discriminate higher- from lower-quality offerings within it.

Validation is the layer where most strategies fail. A clean backtest is necessary but emphatically not sufficient evidence of edge — almost any rule set, with enough parameters and enough optimisation, can be made to look attractive on any specific history. The discipline of out-of-sample testing (evaluating frozen parameters on data the optimiser never saw), walk-forward analysis (rolling-window in-sample/out-of-sample alternation), look-ahead-bias avoidance, and tick-data-quality awareness separates strategies with real edge from curve-fitted illusions. Modelling quality below 99% on execution-sensitive strategies invalidates the backtest entirely.

Deployment introduces operational layers absent in backtest. A persistent VPS is the always-on requirement; sub-10ms ping and selective co-location matter for latency-sensitive strategies; magic-number management distinguishes one EA's trades from another's on multi-EA accounts; .set files codify the production parameter configuration; default values are starting points requiring per-broker re-optimisation. Demo-account-parity matters because many brokers run demos with better conditions than live, inflating EA test results and producing nasty surprises on switchover.

Long-run operation requires risk discipline. Pre-commitment to capital-scaling rules and EA-retirement criteria — written down before deployment — prevents the emotional decisions that destroy retail accounts. Drawdown triggers, daily loss limits, and maximum drawdown cutoffs convert defensive responses into mechanical action. Review cadence — weekly tactical, monthly strategic — catches gradual degradation before it compounds. Realised expectancy compared to backtest expectancy is the empirical evidence of how much edge has survived live deployment.

The craft, ultimately, is operational rather than predictive: edge is small, fragile, and decays; defence is what compounds over decades.

All 87 terms in this cluster

Expert Advisor (EA)

MT5 (MetaTrader 5)

MQL (MetaQuotes Language)

Walk-Forward Test

Overfitting / Curve-Fitting

Strategy Tester

Magic Number (EA Identifier)

Expert Advisor (EA)

MetaTrader 4 (MT4)

MetaTrader 5 (MT5)

Backtesting

Forward Testing

Demo Account

VPS (Virtual Private Server)

Forex Indicator

Script (MetaTrader)

Timeframe

Scalping

Day Trading

Swing Trading

Trend Following

Grid Trading

Martingale

Hedging

Range Trading

Breakout Strategy

Mean Reversion

News Trading

Copy Trading

Algorithmic Trading

High-Frequency Trading (HFT)

Price Action Trading

Currency Correlation

.set File (MetaTrader EA Preset)

.tpl File (MetaTrader Chart Template)

MQL5 Symbol Mapping

Pattern Recognition (in Trading)

Supervised Learning (in Trading)

Walk-Forward Optimization

Regime Detection

Ensemble Model

Reinforcement Learning (in Trading)

Walk-Forward Analysis

In-Sample Period

Out-of-Sample Period

Out-of-Sample Test

Optimisation

Genetic Algorithm

Curve-Fitting

Default Value

Demo Account Parity

Signal (EA layer)

Signal Parameter

News Filter

Regime Filter

Market Regime

MetaEditor

MQL5 Compiler

MQL5 Marketplace

.ex5 File

Expert Journal

Market Watch

Arbitrage EA

Grid EA

Scalper EA

Trend-Following EA

Correlation (EA portfolio)

Overfitting

Hedging Account

Netting Account

PAMM

Prop Firm (funded account)

Versioned Changelog

Multi-Strategy EA

Strategy Class

Neural Network

ML Filter

Retraining Cadence

Concept Drift