Modern societies are characterized by digitalization, through which a vast amount of data can be obtained and processed. The Latin word digital (digitus = finger) is still used in medicine as well as in colloquial language to refer to the finger. In EDP, however, it means gradually, broken down into individual steps. This is in contrast to analog (smooth, continuous). In a digital clock, the digits move step by step; with an analog clock, this happens fluidly. The jerky switching of seconds/minutes is based on a calculation method (algorithm) that proceeds in precisely defined steps. Algorithms are mathematical sequences that control all forms of data processing.

 

So spoke Algorismi

The Persian mathematician and astronomer al-Chwarizmi, latinized Algorismi around 825, opened his discourse on the Arabic-Indian number system and written arithmetics with this expression. The term algebra is derived from the title of his book, the expression algorithm from his name. In the “Astronomical Tables of Sindhind” there is a table with the values of the sine angle function.

Since algorithms are the basis of programming and are independent of the programming language used, they have a special significance. In the example of the digital clock, they must ensure that the same time is displayed after the jumps as in the analog clock. Algorithms are subject to the following criteria: executability, determinability, unambiguity, finiteness, terminability (both in terms of preconditions and continuation).

 

Securities trading

Algorithmic trading (also called automated or high frequency trading) accounts for a good half of all transactions on exchanges today. Here, high-performance computers trade with programmed algorithms either with human interaction or independently. The principle of high-speed trading is based on the fact that orders can be placed on several exchanges and can be divided into partial orders. The most important parameters of the algorithms of these trading systems concern time, price and quantity.

The advantages of algorithmic trading are obvious. By separating the time and place of trading, minimal price differences can be exploited. In the case of high volumes, the tiny profits per individual transaction add up to respectable returns. Arbitrage of price differences on different trading venues or of different instruments on the same trading venue (static arbitrage), spread trading of price inefficiencies of futures contracts or mean reversion are the main areas of application.

 

Investment advice and management

Under the term Robo-Advisor, attempts have recently been made to introduce automated advice and processing in investment advice. This is based on recommendations of an algorithm-based program without human participation. The advantages for the banks are manifold. On the one hand, they can optimize processes, on the other, they can save money. Computer-supported digital advice means less expenditure per customer and a certain standardization, although the solution found is still customer-oriented (tailor-made). The disadvantages for customers are twofold: little or no personal contact and below-average performance so far. People usually play a more important role than assumed, and the difficulty of finding algorithms for a longer period of time is underestimated.

Asset management using an algorithmic computer program is the holy grail of the modern financial world. However, the mysterious object that promises its owner earthly and heavenly happiness in medieval poetry can only be found by the person predestined for it. As in the Middle Ages, the search for it, even with the use of algorithms, will not be successful.

 

Assessment

Algorithms certainly have their place in the financial world. Where analog methods were often costly and inaccurate, today clicks are enough to ensure the preparation of numerous data and facts. The danger is that we hardly recognize the forest for the trees. We even delegate thought processes and rely too much on the computer result instead of on our critical judgement.

All algorithms that reduce time, i.e. minimize latency, are literally “worth their weight in gold” in areas such as securities trading. Where time and price have to be taken into account, they are less successful because important bases for decision-making are poorly or not at all programmable (Covid-19). Moreover their composition and importance are constantly changing. To include performance is almost impossible, as it can only be measured after a certain time.


Christian Wagner
Financial Advisor