Machine learning also plays an increasingly important role in the development of trading strategies. In addition to classic macroeconomic indicators and correlations with other asset classes, new types of data, such as sentiment indicators based on social media posts, are increasingly being incorporated into model training. As a result, the number of data channels considered is constantly growing. This development increases the chance of finding new relationships in the market ahead of other competitors.
Typically, the data in financial applications are high-dimensional. One way to deal with the resulting so-called curse of the dimension is to pre-filter the data appropriately, possibly combined with methods of dimension reduction. The aim is to improve the signal-to-noise ratio of the data in order to reduce the susceptibility to over-fitting. Here, SCAI develops new methods and approaches, also on behalf of partners from the financial sector.