Artificial intelligence on the rise
German investment ﬁrms Acatis Investment and BayernInvest have come together to launch the BayernInvest Acatis KI Aktien Global Fonds, which claims to be the first fund entirely controlled by artificial intelligence.
Self-driving cars are launching and so are self-managing funds. Many people associate artificial intelligence (AI) with robots and movies. At the same time the availability of huge volumes of mass data and the rapid increase in processing power also makes it possible to apply AI to portfolio management.
Into this brave new world, BayernInvest Kapitalverwaltungsgesellschaft and Acatis Investment have jointly launched the BayernInvest Acatis KI Aktien Global-Fonds, the first global equity fund on the German market that is fully managed by AI. Key characteristics of AI programmes include autodidactic learning and the independent detection of patterns. Stock selection, weighting and reorganisation activities are based on so-called deep learning models. The fund manager no longer intervenes in the portfolio decision-making process. The self learning model progressively adjusts to the market environment and pursues a long-term horizon.
While BayernInvest has determined the main factors of the investment strategy, the individual title selection and their weighting is determined by Acatis’ algorithms. The fund is intended to be successful in the long term by automatically adapting to the market environment. Hendrik Leber, managing partner and portfolio manager at Acatis, sees strong market interest for such a product. From an investment universe of approximately 4,000 global shares, up to 50 are selected to achieve an outperformance of at least 3% annually against the MSCI World Index.
Volker van Rüth, speaking for BayernInvest’s management, says the performance test of the fund impressed: the investment target was exceeded clearly in a walk-ahead test in nine out of 11 years, and recovery from drawdowns was faster than for the MSCI World Index. The deep learning models for the fund were developed by Quantenstein, a joint venture of Acatis and NNAISENSE.
The model was constructed on the basis of millions of observations of fundamental equity data. Given broader AI developments, Rüth expects the technology will also occupy a crucial position within the field of asset management. Acatis has been doing research into AI for about four years, aiming to use it for portfolio management. Early efforts focused on programmes for text analysis, such as searching reports for specific keywords. Subsequently, the focus has been to apply lessons from machine-based learning, a type of artificial intelligence that can be compared to a good analyst with years of experience, who knows and has gained insight into many companies, and who over time has developed an ability to detect patterns in company figures and balance sheets. Over time such an analyst learns what features are important, and uses experience to quickly and better contextualise new situations.
Deep learning models work in a similar fashion, with the algorithms designed to enable computers to independently learn and 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”.
‘A difficult step’
Leber says that because of the new nature of how this fund works, the development partners are also looking for a particular type of investor to come on board. “Initially, we would like to address educated institutional investors who will provide between €2m and €5m each as part of a small group of 20 “godfathers”. They will accompany the development of the fund in the first couple of years. The institutional share class is open. Retail will be opened up next year.
We believe for most investors this is a difficult step, and it will take time to get educated about the workings of such a machine operated fund.” However, he adds: “This is the future – AI-powered funds will dominate the market in a few years. I see rapid growth and a lot of interest in related products.”