How Investtech uses artificial intelligence

Published 4 July 2023

Artificial intelligence (AI) has experienced a significant increase in popularity and interest in recent years. Technological advancements in computing power and data processing have opened the door to larger and more complex AI applications. This has made it possible to process enormous amounts of data faster and more efficiently than before, leading to the development of advanced algorithms and models within the field of AI. The growing availability of data has also played a crucial role in the popularity of AI. AI technologies can analyse and extract insights from these datasets, thereby uncovering patterns and trends that were previously impossible to detect.

Investtech, however, has been utilising AI for a long time. The company was founded in 1997 based on the results of an artificial intelligence research project conducted at the University of Oslo a few years earlier. Even back then, the algorithms could automatically recognise important patterns in stock prices, such as rectangle formations and head-and-shoulder formations, more objectively and accurately than humans could. Such pattern recognition is a crucial part of technical stock analysis.

In the following years, we developed algorithms for recognising general trends, support and resistance, and other key elements in visual technical analysis. These automated chart analyses remain the core of our analysis systems to this day.

Around the turn of the millennium, we developed our first language engine and digital analyst. Based on input from chart analyses, this robot conducted an analysis of historical price movements, what it revealed about investor optimism, and what it indicated about future price developments. The result was a complete analysis text, a technical score, and a positive or negative recommendation.

Investors have paid higher prices over time to buy Apple and the stock is in a rising trend channel in the medium long term. This signals increasing optimism among investors and indicates continued rise. The stock has given a positive signal from the inverse head and shoulders formation by a break up through the resistance at 157. Further rise to 193 or more is signaled. There is no resistance in the price chart and further rise is indicated. In case of a negative reaction, the stock has support at approximately 175 dollar. The stock is overall assessed as technically positive for the medium long term. Recommendation: Positive. (Automatic analysis based on close 29 June 2023.)

Investtech's research on Nordic data showed good consistency between the theory of technical analysis and empirical observations in the markets. However, some effects turned out to be contrary to what the theory suggested. This is critical and could lead to unfavorable investment results. Other effects varied greatly depending on other parameters, such as stock volatility and risk aspects. New acquired knowledge of this kind has been continuously programmed into the analysis robot's artificial intelligence since the 2010s.

A fantastic result of the algorithms is that they provide numerical values on how good the stocks are. This enables sorting and generating ranking lists, and is an invaluable tool for our customers and analysts when selecting stocks. At the same time, the system provides clear warnings for stocks that may be tempting to buy but should definitely be avoided based on historical development of similar stocks.

It is an important principle for Investtech that recommendations are given in easily understandable language and with references to statistical basis. This is different from many other AI systems that often function as "black boxes", where recommendations and analyses are given without understanding why and without references to the underlying statistics.

In 2023, we launched health check for stock portfolios. Here, the individual characteristics of stocks are used in combination with portfolio aspects such as covariance and risk diversification. This provides an assessment of the portfolio's robustness and upside potential, somewhat like what a highly qualified portfolio analyst could provide. The health check provides a score for the portfolio as a whole, evaluates eight different portfolio characteristics, and offers specific suggestions for improvements at the stock level, and is another example of AI in practice.

At Investtech, we have more than 25 years of research and experience in using AI for stock analysis and selection. The systems are comprehensive, objective, and independent of all external parties.

The analysis robot will never be able to consider all factors in the markets, and one's own expertise in specific sectors and stocks can be important. The markets will also change over time, especially when new technology is adopted, and we have no guarantee of how the results will be in the future. Therefore, monitoring, research, and further development are important to us, and our goal is to always be among the leading companies in automatic and quantitative stock analysis.