Glossary

Definitions of algorithmic trading, AI, and platform terms used across the site.

A

Algorithmic trading— **Algorithmic trading** (or *algo trading*) executes orders automatically based on pre-defined rules — typically technic

Backtest— A **backtest** runs a trading strategy against historical price data to estimate its past performance. Backtests are eas

ECN broker— An **ECN broker** routes client orders directly to a network of liquidity providers (banks, hedge funds) rather than int
Expert Advisor— An **Expert Advisor (EA)** is an automated trading program for the MetaTrader 4 or 5 platforms, written in MQL4 or MQL5.

In-sample period— The **in-sample period** is the slice of historical data used to fit or optimize strategy parameters. Performance here i

Leverage— **Leverage** lets you control a position larger than your margin. 1:100 leverage means $1,000 margin controls $100,000 n
Lot size— **Lot size** describes order volume: 1 standard lot = 100,000 units of base currency; 0.1 mini lot = 10,000; 0.01 micro

Machine learning— **Machine learning (ML)** in trading uses statistical models to predict price moves, classify regimes, or detect anomali
Market maker— A **market maker (MM) broker** takes the opposite side of client trades. Their P&L can be adversarial to yours, which hi
Maximum drawdown— **Maximum drawdown** is the largest peak-to-trough decline in account equity during a backtest or live period. Critical
Mean reversion— **Mean reversion** strategies bet that prices oscillating around a long-run average will return to that average. Classic
MetaTrader 5— **MetaTrader 5 (MT5)** is the multi-asset trading platform from MetaQuotes, supporting forex, stocks, futures, and crypt
Momentum strategy— **Momentum strategies** bet that recent price strength persists for some period. Trend-following EAs, breakout systems,
MQL5— **MQL5** is the C++-derived programming language used to write Expert Advisors, custom indicators, and scripts for MetaT

Optimization— In algo trading, **optimization** sweeps strategy parameter ranges (stops, take-profits, indicator lookbacks) against hi
Out-of-sample period— The **out-of-sample period** is the slice of historical data that was NOT used for parameter fitting. Out-of-sample perf
Overfitting— **Overfitting** (or *curve-fitting*) occurs when a strategy's parameters are tuned so tightly to historical data that th

Pip— A **pip** (percentage in point) is the smallest standardized price movement in a forex pair — typically the fourth decim

Reinforcement learning— **Reinforcement learning (RL)** trains an agent to take actions (buy/sell/hold) that maximize a cumulative reward (P&L).

Sharpe ratio— The **Sharpe ratio** measures risk-adjusted return: (mean return − risk-free rate) ÷ standard deviation of returns. Valu
Slippage— **Slippage** is the difference between the price you intended to trade at and the price you actually got. Tight stops an
Sortino ratio— The **Sortino ratio** is like the Sharpe ratio but penalizes only downside volatility (negative returns) rather than all
Spread— The **spread** is the gap between the broker's bid and ask prices. Scalping strategies are particularly spread-sensitive
Stop Loss— A **Stop Loss (SL)** is a standing order to close a position at a predefined adverse price, capping the worst-case loss.
STP broker— An **STP (Straight-Through Processing) broker** passes client orders directly to liquidity providers without a dealing d
Strategy Tester— MT5's **Strategy Tester** is a built-in backtesting environment that replays historical tick or M1 OHLC data against an

Take Profit— A **Take Profit (TP)** is a standing order to close a position at a predefined favorable price, locking in a gain. TP pl
Trailing Stop— A **Trailing Stop** is a dynamic stop loss that moves toward the current price as the trade goes favorable, locking in u

VPS— A **Virtual Private Server (VPS)** is a remote machine that runs your MT5 + EA 24/7 with low latency to the broker. Esse

Walk-forward analysis— **Walk-forward analysis** optimizes strategy parameters on an in-sample window, then tests them on the next out-of-sampl