Stockholm (HedgeNordic) – In HedgeNordic’s upcoming Special Report on Equities, we have looked to the future for inspiration. Our supplement on Hedge Funds 2.0 will unveil some exciting analysis on artificial intelligence, alpha in a changing world, and even MiFID II. While the shop is busy bringing the final touches to the report, we are bringing forward the views of Ola Björkmo, CEO at Stockholm-based hedge fund QQM, who accepted to share his perspectives on what new technologies mean for his strategy.
One of the reasons innovation, especially in quantitative fields like finance, is now accelerating tremendously, is the availability of data. Computers have never been so fast, and therefore able to process such large amounts of data. In parallel, the cost of data storage has fallen dramatically, and the vast extent of networks allows for ever increasing reach. For a systematic strategy such as QQM’s, this certainly makes a difference.
“At QQM we combine modern portfolio management techniques with insights from behavioural finance theory,” starts Björkmo. “To be able to analyse and trade a universe of 1000 European stocks we depend on the combination of fast computers and huge databases. Ever larger-data sets and the ease of gathering the data presents opportunities to trade new markets and new strategies.”
With the availability of high processing powers, quantitative-driven technology has taken off. Neural networks, machine learning and other self-learning systems are revolutionising fields such as automotive industry (self-driving cars) but also security and internet searches (image recognition) or even healthcare. Many are trying to take advantage of these tools in financial applications as well, but there are obvious potential pitfalls.
“One potential problem with these kinds of techniques and tools is the potential for data mining,” warns Björkmo. “When you test or optimise an infinite number of strategies, statistically you are bound to find several that have performed well, however odd they may appear. We believe in the old quant saying: ‘If you torture a database long enough, it will confess to anything’.”
Hence applying algorithms should be done wisely, and be grounded in proven financial theories. Björkmo continues: “The models QQM deploys in fund management are based on established economic principles and not solely designed to adapt to historical data. We are constantly looking for new models and techniques – models that complement existing models well and contribute to greater diversification in the fund.”
Another threat could come from the very success of these newly super-powered quantitative strategies. More data may mean more capital and more leverage, but those may come at the cost of stability. “We believe quants and systematic strategies are increasingly focused on large data sets, real-time data, AI and machine learning,” Björkmo explains. “We have been told that long and gross exposure have both reached record highs. The advent of new quant strategies has increased the risk of large market movements due to simultaneous de-leveraging or rebalancing of portfolios, triggering an August 2007-like meltdown.”
Looking ahead, Björkmo sees his strategy develop further, mainly through “more data, faster computers and self-learning systems.”