Many people associate artificial intelligence (AI) with robots and self-driving cars. At the same time, the availability of huge volumes of mass data and the rapid increase in processing power also mean that AI is predestined for use in portfolio management. Through machine learning, key characteristics are detected over time, allowing the system to find patterns that even the most seasoned analyst would struggle to detect.
Application areas for AI in the investment segment include:
- enterprise valuations
- predictions of future “winner stocks”
- portfolio optimisation
- detection of critical text passages
- identification of similar firms
Dr. Hendrik Leber shares developments regarding the four fund members
of the ACATIS AI-family at the 4th NextGen AI Aware Event
ACATIS has been doing research in the area of AI for about four years, with the aim of using it for portfolio management. At the beginning programmes were used for text analysis purposes, which can search a report for specific key words.
Today, ACATIS works mainly with Deep Learning models, an approach from the area of machine-based learning. This type of artificial intelligence can be compared to a good analyst with years of experience. During the course of an analyst’s career, they can get to know a lot of companies and gain insight into them. As time progresses, they can develop a knack for detecting patterns in companies’ figures and balance sheets. Over time, they learn what features are important, with their experience helping to quickly and better contextualise new situations.
Deep Learning models work in a similar fashion. They learn to independently detect patterns in balance sheets, which they then apply to new data. The more data that is available to the system, the better it can learn and gain “experience”.
At the same time, as the volumes of data increase, so do the demands on processing performance. The two big advantages of Deep Learning models as compared to an analyst are their much greater capacity and emotional detachment. The system can also find patterns that humans would not be able to detect. In addition, it makes decisions strictly on self-generated rules, leaving the emotional aspect aside.
The first practical application of AI at ACATIS took place in October 2016, when the already existing ACATIS Global Value Total Return was populated solely with titles that were pre-selected on the basis of artificial intelligence. The programme that was used in this context was a Convolutional Neural Network (CNN), which had the task of identifying those stocks that were supposed to out-perform in the short, medium and long term. The CNN was trained on the basis of the ACATIS fundamental database, which has been developed for 15 years and which contains financial data going back to 1989. The CNN delivered 1,300 stock recommendations, to which the classic ACATIS Value filters were subsequently applied: valuation, hit statistics, trend norm filter, forecast programme. This was followed by a manual check at the end of the process. The result - the current portfolio of 50 equally weighted titles. The management of the financial rate of investment (0-110%) is not affected by the title selection. In ACATIS Global Value Total Return, ACATIS navigates the route and AI is sitting in the passenger seat.
In 2017, ACATIS launched two equity funds that invest globally in individual stocks, as well as an external mandate whose share selection and portfolio composition process is entirely controlled by artificial intelligence. The external mandate FIM Tekoäly is a copy of ACATIS AI Global Equities (same stocks and weighting). The fund is marketed exclusively in northern Europe out of Finland. One of these is ACATIS AI Global Equities. ACATIS provided the specifications and data for the investment strategy. The resulting stock selection, weighting and reorganisation processes are based on Deep Learning models that are inspired by the way the human brain works.
The model that is used for ACATIS AI Global Equities is based on the work conducted by Prof. Dr. Jürgen Schmidhuber, co-inventor of the long-short term memory (LSTM) neurons and a pioneer in the field of Deep Learning. The applied Deep Learning models can store previously learnt patterns and events, and can remember them in due course. The self-learning model progressively adjusts to the market environment and pursues a long-term horizon. The models were developed by AI specialist NNAISENSE for ACATIS. They look for the correlations on their own, while the neurons in the neural network specialise in detecting certain details. The entire model is the result of the interaction between all of the neurons. The models are based on fundamental data such as revenues, EBIT, and profit. The fundamental data also comes from the extensive ACATIS company database. Several end-to-end optimised sub-models are used within the applied model. The fund manager only ensures that the portfolio design is implemented in the fund.
In May 2018, ACATIS AI BUZZ US Equities was launched, whose fund composition is based on the BUZZ NextGen AI US Sentiment Leaders Index. Using artificial intelligence (AI), online comments on individual US stocks from relevant sources such as social media, news portals, blog posts or other discussion forums are evaluated. AI identifies companies with the highest positive investor sentiment from a wealth of opinions expressed. With monthly rebalancing, investments are always made in the top 75 most positive companies. Around 350 US companies currently belong to the BUZZ universe and are analysed continuously with regard to investor sentiment.
Sentiment measurement 2.0
- Momentum in comments is a precusory for momentum
- Revolutionary new sentiment measurement in fund management using artificial intelligence
- Individual portfolio strategies based on investor sentiment measurement by using social media and online forums
"Using artificial intelligence and big data in the pursuit of investor sentiment"
ACATIS AI BUZZ US Equities presented by Stefan Riße
Our collaboration with Jamie Wise
We are working with Jamie Wise, founder and CEO of BUZZ Indexes from Toronto. BUZZ Indexes creates its own individual portfolio strategies based on investor sentiment measurement by analysing social media and online forums. BUZZ Indexes uses artificially intelligent supervised learning models to filter and edit the data.
"Searching for Investor Sentiment Across Online Platforms with Artificial Intelligence"
More information in the following video.
ACATIS is also in the process of testing other types of artificial intelligence models.
Anyone interested in testing the method used by artificial intelligence can do so by visiting https://playground.tensorflow.org.
As a 50/50 joint venture, ACATIS and NNAISENSE founded Quantenstein GmbH in 2016. The purpose of this joint venture is to develop artificial intelligence models in the area of long-term value investing. We understand this as automated, self-learning investment strategies.
Since autumn 2018, Quantenstein is a 100% subsidiary of ACATIS Investment.
Quantenstein can be used to design tailored and performance-optimised investment portfolios for a defined investment universe and specified restrictions (e.g. dividend yield, holding period, drawdown). Quantenstein integrates stock selection and portfolio design in one single process (end-to-end architecture).
It searches the ACATIS fundamental database to find the best medium- to long-term stocks.