AI in Asset Management – Trends and opportunities for the intelligent investor

Artificial Intelligence, the science of designing machines that replicate human knowledge and behaviour, has undergone exponential development throughout the last decade. The increase of computational power predicted by Moore’s law and the access to large amounts of data allowed machine learning models to deliver on their promises. The growth of the field has been so tremendous that some experts predict that by 2030 between 10% and 15% of the global GDP will be based on AI, and that the point in which AI will surpass human intelligence (a.k.a. the “AI singularity”) will occur in about 20 to 30 years from now.

Such growth and predictions will nonetheless depend on the flexibility of the market to adapt to this new revolution, on the regulatory framework imposed around technology, and on the trust and willingness of humans towards machines to perform part of their jobs. 

A particularly interesting case refers to how AI will shape the financial industry. Applications of AI and machine learning in finance include, for example, the use of supervised learning models for credit ratings and factor investing, or the application of unsupervised learning techniques for clustering investment behaviour and detecting potential anomalies.  

The pace of adoption of AI in the financial industry is, however, somewhat slower than in other industries, such as IT & Telecommunications, Retail, Healthcare, or Manufacturing. While this may be in part due to the use cases of machine learning that were first envisioned (e.g. image recognition), financial managers ought to understand the potential business applications of AI in order to adapt to such technological development and survive in this innovative environment. 

At LPA we have identified some of the most relevant applications of machine learning in financial markets and developed an expertise around four main areas: Trading Execution, Portfolio Optimization, Risk Management, and Derivative Pricing. This line of research has allowed us to develop both consulting services and software products that provide asset managers with increased revenues, lower costs, and reduced risks.

Can machines therefore think and help us make better investment decisions?

We believe the answer is yes. For example, we have used neural networks and clustering techniques to build stable and optimal portfolios with low concentration risk, reinforcement learning to reduce transaction costs and market impact when trading large or illiquid orders in the market, and generative adversarial networks to deliver more accurate tail risk estimation and hence better manage large, unexpected, drawdowns.

It is now time for the financial industry to claim the throne of innovation in artificial intelligence. It is time to show that traders and portfolio managers can greatly benefit from machine learning. It is time to invest with artificial intelligence.