Automated trading is controlled by a computer, i.e. a machine, but of course it cannot think itself, nor can it take decisions. It is completely free of emotions and only executes the commands given to it. The machine places orders directly on the basis of predefined quantitative (mathematical) models. An algorithm decides on the timing and volume of a buy or sell order. Parameters are defined by historical and current market data, price differences, trends and other factors.

A fully automatic trading system like this is programmed using the know-how of a large number of experts. The challenge when writing algorithms is to aggregate and analyse historical market data and to aggregate real-time prices in order to enable trading.

Thousands of sequences have been stored in order to react in real-time. The machine generates buy and sell signals autonomously, which are translated into orders on the exchange before people have time to intervene. This happens in a fraction of a second (1/4000) while the machine continues its analysis. This is something no human being can do.

Furthermore, the amount of important information interpreted and processed by the machine is much higher than if a person was involved. This further reduces the costs of the executed transactions. There is no time lag between receiving and entering changes, since the machine responds in real-time.

But human beings are not out of the picture altogether, because the systems used for algorithmic trading have been programmed and installed by people, who also set them in motion and monitor them permanently. Indeed, it would be dangerous to trust machines unconditionally with investment decisions. The key lies in the interaction between artificial intelligence and management. Human beings plus quantitative systems is the winning combination.