Algorithmic trading is also known as high frequency trading or, in financial jargon, algotrading. It is an investment strategy based on economic models and computer software that can take positions in a market. With the dematerialization of trading orders and the increasing computing power of computers, this type of trading is booming. We will explain to you as precisely as possible how algorithmic trading works.
Algorithmic trading is an investment method based on a mathematical model that proposes decisions to take positions instead of the usual operator. This method began to be developed in the 1980s thanks to the dematerialization of stock market orders. Today, it takes only 13 milliseconds to make a round trip by cable between the NASDAQ and Chicago stock exchanges and more than 70% of American stock market transactions use this trading method.
The latest algorithmic trading software also offers decision making strategies in that the computer can react instantly to the slightest price change and is therefore faster than a human trader.
One of the prerogatives of using this type of trading is to use it on a market with high liquidity and on classic products such as stocks, currencies or interest rate products.
There are two types of activities in algorithmic trading with assisted trading and automated trading.
The former are in some ways the basic version of this method since the mathematical models used simply provide trading suggestions that investors are free to use or not when placing their own orders.
Automated trading is rarer and more complex in the sense that it trades directly 24 hours a day and according to set strategies. They work on the principle of truncating large orders into a series of marketable lots in order to reduce brokerage fees by placing orders directly through a machine.
Of course, this mode of trading is regulated as MiFID 2 has published various rules and definitions that apply to it. Thus and since 2016, companies that transmit at least two orders per second on one instrument and more platform or 4 messages on several instruments and on one platform are considered as high frequency trading operators.
It goes without saying that algorithmic trading is not for novices. Indeed, trading algorithms are available to programmers who define their working instructions in order to follow a precise and well thought-out investment strategy. However, they are designed to work on different operating systems.
The transactions made by these instruments are of course encrypted. The machines launch very powerful calculations because they can cross-check different sources simultaneously, taking into account market history, volatility and other such elements.
Regularly, algorithmic trading is accused of manipulating prices. This is why the Autorité des Marchés Financiers or AMF, in June 2016, requested a penalty of one million euros against the high-frequency trader Getco, which was accused of having manipulated the price of no less than 7 CAC 40 stocks.
Previously, in 2015, Euronext and Vitu, an American trader specializing in this type of trading, were convicted on the same grounds.
Price manipulation is made possible by the extremely fast calculation speed of these machines and by techniques such as spoofing, which consists of flooding the market with orders and then cancelling them in order to feign strong demand. This makes it possible, for example, to sell securities whose price you set yourself.
Thus, in order to reduce the risk of manipulation, the European authorities have asked high-frequency traders to keep their algorithms for 5 years and to provide detailed information with 34 criteria on their transactions.
In conclusion, algorithmic trading is not for everyone and even less for individual investors who occasionally trade the markets.