Interview with Holger Knauer, Catana Capital

Holger KnauerPrior to joining Catana Capital, Holger Knauer was a board member of EUR 1.8bn Celios Investment-AG TGV and its predecessor Prisma Investment-AG TGV since 2011 which he co-founded. Before that he was with Universal Investment for more than a decade as a deputy head of department for institutional sales and relationship management. Although big data has been a buzzword for quite some time, the finance industry seems to be late to the party once again. Except Google’s plans to launch robo-advisors at some point in the future not much seems to happen. Why are there not more fund managers building products around big data?
Holger Knauer: Because it is not so easy to do and many fund managers are still quite relaxed in their comfort zone. With volatility in markets increasing, the accelerating negative interest environment and crowded carry trades no longer working, however, this is about to change. So investors are looking for new ways of allocating and our Big Data based asset management approach is a way to do that. There are a number of large funds experimenting with big data already, but many of them are still either trying to understand what it is all about or using it as one of many components to their existing trading strategies. Catana Capital is different as we are “pure play” Big Data and A.I. Maybe just a comment what Big Data actually means: it is only to a minor degree about large amounts of data. In our view it refers to new types of data that was not available years or even sometimes months ago. This type of data comes – besides in very large amounts – in a highly unstructured form – so noise reduction and working with these information is very different to traditional data mining. You analyze more than 400,000 relevant messages per day in real time, but only adjust your portfolio every two weeks. Do you not miss out on short term opportunities and big market events?
Knauer: Actually quite the contrary. In the beginning it was also a surprise to us that the approach does not work best as a high-frequency trading strategy. We found out, when we developed our approach, that markets are actually still incredibly inefficient, and information still needs time to be fully reflected in prices. This is why we only trade approx. every two weeks for the stocks positions. The DAX index future which we a trading as well is traded only a little more frequent with approx. 3 trades per week. But why is this the case? We think the ever increasing flood of information actually makes it harder to grasp the relevant information. Please keep in mind that approx. 90% of the world’s data has been created in the last two years! This is why we are saying: we are turning “epsilon” into “alpha”, to stick with the traditional CAPM. As 70% of alpha can still not be explained by the traditional market factors we think that Big Data opens a way to cope with this 70% of “noise” or epsilon which currently can not be addressed by traditional fund managers. The first hedge fund to base its investment decisions on Tweets already shut down again. Why do you think you can do better? 
Knauer: We cover more than 25,000 securities globally in real time including also currency, commodities, etc. This is more 400.000 relevant messages regarding financial markets per day – post filtering. Basically anything which is written in the web, we can read. Whether it is a news piece or a research piece but also user-generated content, like Twitter or StockTwits, or information regarding key events, that is all captured in real time by our Research Partner Stockpulse. On top, the information is automatically compared to historical information and price developments via an artificial intelligence. This is very different from now defunct Twitter hedge fund, which only used Twitter as a platform and which you are referring to. The crawling technology covers German and English language data, but we are already looking to add other languages such as Chinese and French. Are you afraid that companies like Google or big asset managers like Blackrock will nullify your edge once they dedicate enough attention (and money) on the subject?
Knauer: I think three factors differentiate Catana’s strategy from peers. First, our predictions are not based on historical security prices, whereas most fund managers usually try to identify past patterns and then extrapolate. Instead, Catana bases its predictions on a very unique set of uncorrelated data sources which we collect. The second differentiating factor is how machine learning, comes into play and how it requires the full timeframe of the database to be able form a proper prediction. The third component is the specific trading strategy and how we are using this information. To put it in a nutshell: I am convinced usage of Big Data and AI-based trading strategy will strongly grow in the asset management space. However, developing the technology and building the required database to include enough information just takes time. At the same time I think at there are many ways of bringing Big Data to use to manage money and we are encouraging more managers to look at Big Data. Thank you for the interview.