AI Science - Q&A

According to a broad scientific consensus we witness one of the largest scientific revolutions in mankind’s history. Artificial Intelligence is about to improve every aspect of our daily lives thanks to the outstanding work of responsible engineers and scientists. This scientific revolution also gives rise to a paradigm shift in the asset management industry toward a more mature, industrialised, digitized, systematic, and scientific way of constructing investment portfolios using the support of AI. This cutting-edge technology cannot only be employed to enhance returns, but also be used to manage risks. With the power of AI, entirely new ways of creating scenarios, simulating correlation matrices and (stress) testing portfolios are now available to professional investors.  Thus, future regulatory requirements could be met, and sound risk management practices could be implemented. Examples  are packaged retail investment and insurance products performance scenarios and European Securities and Markets Authority stress testing. As a consequence, similar to the obligation of minimising the total expense ratio (TER), of including ESG criteria and diversity industry standards, incumbent regulators such as the SEC, BaFin or FINMA may soon require professional investors to prove that they have thoroughly assessed the implementation of scientifically superior methodologies to assess product risks. AI-powered investment strategies are therefore poised for becoming rather an integral part of an investors’ investment style than just a niche asset class.

Taxonomy

Taxonomy

AI-powered investment strategies can be classified into five Machine Learning approaches employing three different sets of data. A blank (white) field means that this technology cannot be applied to the respective data set. The more colourful a matrix field, the stronger the relationship between the ML-approach and the data sets.

It is important to note that all five ML-approaches might apply Deep Learning Artificial Neural Networks or Convolutional Neural Networks as a machine learning architecture to analyse the data. Some of these Deep Learning algorithms have evolved from simple multi-layer perceptrons to capture time dynamics through Recurrent Neural Networks. In particular, Gated Neuron Designs, which allow for capturing long-term dependencies in financial time series, text or speech analysis (e.g., Long Short-Term Memory (LSTM)), have gained popularity in order to explain complex, time-dependent relationships.

Most of the successful AI-powered investment strategies are specialised in one or a combination of the five ML-approaches applying one or more of the three data sets. Combining several investment strategies applying different ML-approaches in an investor’s portfolio delivers enhanced diversification effects.