Fredrik Giertz is senior manager – quantitative strategies at the Third Swedish National Pension Fund. AP3 is a 35 billion EUR pension fund managing part of the buffer capital within the Swedish public pension system with investments across several asset classes, styles and approaches. Fredrik has a background in mathematics and physics from the Royal Institute of Technology in Stockholm and joined AP3 in 2014.
Fredrik is a part of AP3’s internal Alpha Asset Management team and manages two main portfolios, Alternative Risk Premia and Quantitative Alpha Strategies.
Quantitative Alpha Strategies
During the last two years, AP3 has developed and launched a new tier of strategies aiming to extract alpha across several asset classes using advanced machine learning algorithms. The aim of these strategies is to identify complex repeating patterns in asset pricing often overlooked by other market participants. The focus of the strategies is on the self-learning algorithms rather than the use of alternative data sources, something that differs from the typical approach. The strategies build upon algorithms developed by Fredrik and are both developed and coded from scratch by Fredrik and his team. The strategies use a combination of supervised and unsupervised ML methods to mitigate many of the problems associated with the use readily available ML algorithms on noisy and dynamically changing data, the typical problem setting for financial data. To make it possible to use machine learning strategies in practice, Fredrik has spent a lot effort on both the organizational buy-in as well as the technical infrastructure required.
Alternative Risk Premia
AP3 has invested in ARP strategies since 2011. The allocation was initiated using external providers but is today almost fully managed internally using strategies both researched and executed in-house. Within ARP, economic rationale lies at the core of the investment hypothesis, e.g. how different information drives risk premia. To understand the impact of information in asset pricing, as well as the dynamic relationships between data, Fredrik has used Machine Learning as a very important part of both the research step and the ongoing allocation process. This has in turn led to a portfolio with more balanced ARP strategies and better diversification.