Introduction to algorithmic trading

What it is

Algorithmic trading means defining precise entry, exit and risk management rules and having a program execute them. The main advantage is not 'earning more': it is consistency, the ability to test the rules on lots of data and the absence of emotion at execution time.

Strategy families

• Trend following: you follow the move (moving averages, breakouts of highs, momentum)
• Mean reversion: you bet on a return to the mean after an excess
• Volatility breakout: entries on the break of a range
• Arbitrage and market making: they require infrastructure and costs the retail trader doesn't have

Advantages

• Verifiable, repeatable rules
• Testing on years of data in minutes
• No hesitation or revenge trading
• Ability to monitor several instruments together

Limits

• A badly executed algorithm loses money in a perfectly consistent way
• Backtests can deceive: overfitting and underestimated costs are the main risk
• Markets change and rules can stop working
• You need programming skills and continuous maintenance

Tools to get started

• Python with pandas and numpy for analysis and backtesting
• MetaTrader 5, which also lets you connect Python or write Expert Advisors in MQL5
• TradingView with Pine Script for quick prototypes of indicators and strategies
• Reliable historical data, with the same session and the same time zone for the whole test

Regulatory note

Automating your own account is allowed. If in the future you wanted to sell an algorithm, offer signals or manage other people's money, authorisation requirements may apply (for example in the EU and UK): get professional advice before doing so.