Algorithmic Trading - what is it?

One of the biggest sources of confusion relating to algorithmic trading lies in its definition. Different people mean different things by the term, with many now regarding it as synonymous with any sort of automated trading. The “traditional” definition (and the one that we use in the magazine and on this site) is that algorithmic trading is about using a set of rules to finesse trade execution.

Algorithmic trading involves splitting a trade into multiple orders in order to reduce visibility and market impact, but the decision to take the main trade might or might not be automated. A fund manager might decide that a particular stock looks attractive based on his/her fundamental analysis and then instruct his/her trading desk to buy a block of stock. The traders on the desk might well use trade execution algorithms to finesse the placement of this trade.

Automated trading on the other hand involves a set of rules (a very simple example might be a pair of moving averages of different lengths crossing over) that when satisfied automatically trigger the placement of an order. A small, simple automated trade might be placed directly into the market, while a more substantial one might be handed to an execution algorithm for placement in small order slices so as to reduce market impact etc.

In brief, (in our “traditional” definition) an automated model determines whether to place a trade, while an algorithmic model determines how to place it.