What Is Algorithmic Trading?

What is algorithmic trading?

Algorithmic trading is a method of executing trades using predefined rules and automated systems. The rules specify what to trade, when to enter, when to exit, and how much to risk, and software applies them without further human input.

The word "algorithmic" does more work than it needs to. An algorithm here is just an explicit set of conditions. "Buy when the 9-period moving average crosses above the 21-period, exit when it crosses back, risk 2% per trade" is an algorithm. It is not a prediction and it is not artificial intelligence. It is a rule, written down precisely enough that a machine can follow it.

That precision is the whole point. A rule written down can be tested against history, applied identically at 09:15 and 15:15, and reviewed afterwards to see what actually happened. A decision made in the moment cannot be any of those things.

How does an automated strategy actually place an order?

Four stages: the platform reads market data, evaluates your rules against it, decides whether a condition is met, and if so sends an order to your broker through an API you have authorised.

The order goes to your own broker account. The platform never holds your money. It holds permission to place orders on your behalf, which you grant and can revoke. That distinction matters more than any technical detail on this page.

In practice the loop runs on every new candle. Most evaluations produce no action at all, which is normal: a strategy that trades constantly is usually a strategy with no filter.

The four stages, and what can go wrong at each

Is algorithmic trading only for institutions?

No. Broker APIs and hosted platforms have made rule-based execution available to individual traders, though the operational burden is real and the ones who do well treat it as engineering rather than a shortcut.

What institutions still have is infrastructure: colocated servers, direct market access, dedicated data feeds and teams whose job is monitoring. An individual competing on latency will lose. An individual applying a sound rule set consistently over months is playing a different game, and that one is winnable.

What algorithmic trading does not do

It does not predict prices, guarantee returns, or turn a strategy with no edge into a profitable one. Automation makes execution consistent. That is a real benefit, and it is the only one it provides on its own.

This is worth being blunt about, because the category attracts claims that do not survive contact with a live market. If a set of rules loses money on average, automating it means losing money faster and more reliably. The edge has to come from the rules.

What actually separates strategies that survive

• Tested over enough history to include conditions the author did not anticipate
• Costs modelled honestly, brokerage, slippage and statutory charges, not just price differences
• Position sizing defined as a rule rather than chosen per trade
• An exit for every entry, including the case where the exit path fails
• Paper-traded before live, long enough to see it behave in real conditions

Where do you start?

Write one strategy you can state in a sentence, backtest it over several years, and paper-trade it before committing capital. The first strategy is for learning the process, not for making money.

On Stretus that path is: describe the strategy in plain language, let the platform assemble it into a structured object with explicit entries, exits, stops and sizing, run it against historical data, then deploy it to the paper simulator. Live execution through a broker connection is a separate, deliberate step you take afterwards.