What is algorithmic trading?

What it removes

The largest single source of retail underperformance is not strategy selection, it is inconsistent execution of a plan — moving a stop because the trade feels wrong, skipping a signal after two losses, taking an unplanned trade out of boredom. An algorithm does none of these.

It also trades at times a human cannot. A gold strategy trading the London–New York overlap is trading in the middle of the night for a large part of the world, and consistency across those sessions is not achievable manually.

What it does not remove

A bad rule is now applied perfectly and tirelessly, which is worse than applying it inconsistently. Automation amplifies whatever the rules are, and most of the failure modes in retail automated trading are consequences of that amplification — see why do most trading bots fail.

It also does not remove the human from the loop, only from the trade. Choosing risk per trade, deciding whether the account is large enough, and deciding whether a 15% decline is a normal drawdown or a broken system are all still yours.

How retail automation actually works

It is not high-frequency trading and it does not require artificial intelligence. Most of it is a modest set of conditional rules running inside a desktop platform, which in the MetaTrader world is an Expert Advisor.

The program is attached to a chart, runs on each tick or each new bar, and can place orders without confirmation. It runs on your machine, not the broker's, which is the fact that determines the entire operational picture around it.