Last week Kevin Endler wrote about our presentation at the 5th NextGen Artificial Intelligence (AI) Aware Event (Report by Kevin Endler). We presented the latest updates of the ACATIS AI Global Equities fund, our first fund that is 100% managed by a machine learning model. In particular, we showcased the new stock selection model, a Graph Neural Network, which eventually leads to a portfolio consisting of 50 companies. Looking at our top ten holdings an eye-catching characteristic pops up: the AIG misses well-known large-cap companies but focuses on small and mid-cap companies, that even seasoned portfolio managers might be less-familiar with. A style analysis reveals: This finding holds for the entire fund. Compared to our benchmark, the MSCI World Index, the ACATIS AI Global Equities tends to small cap companies. Besides it concentrates on value factors (~2/3) followed by growth (~1/3).
Would a portfolio manager be concerned with these holdings? Let’s look at an example: ResMed.
The US company ResMed develops and sells medical devices that diagnose, treat, and manage sleep disorders. It has an annual two digit sales growth over the last 10 years and a stable EBIT-margin of 24%. We say: Sleep well with ResMed. Since rebalancing its share has climbed by 25%; it starts getting expensive and in our next rebalancing in a few days, the machine is likely to tell us to pocket the gains.