Zurich-based Vestun, a start-up boutique manager focused on AI-based investment programmes, has opened its market agnostic systematic US equity hedge fund programme to new outside investors.
The flagship strategy, which is managed on the firm’s Hawkrel AI platform and trades liquid US stocks, is designed to autonomously adapt its portfolio and risk exposure according to prevailing market conditions. It eschews traditional systematic hedge fund approaches which employ signals derived from statistical rules and historical events, instead building a less rigid investment process around uncorrelated, domain-specific intelligence gleaned from certain datasets.
Chayan Asli, Vestun’s founder and CEO, believes that relying on signals generated from statistical rules and back-testing history is not sustainable for delivering consistent long-term market outperformance.
“Nowadays, everyone has access to the same financial datasets and machine learning models. If everyone uses the same smart systems with the same recipe for success, it will undermine the competitive advantage obtained by using computer-driven models to invest.”
Asli, who said Vestun’s fund performed “sustainably” during this year’s Covid-19 market crash with low volatility compared to benchmarks, described the crisis as a stress test for “revealing the relevancy of a successful investment strategy.”
Now, having weathered this coronavirus-fuelled market turbulence, Vestun – which has up until now managed its own capital internally – aims to attract a broad range of institutional investors to its strategy, including banks, multi-family offices and asset managers.
“Most commercially available AI applications only focus on machine learning, a subfield of AI which is best suited to recognising patterns but not understanding them,” said Vestun’s chief scientist Stephen Varey, who prior to joining Vestun spent more than 20 years developing intelligent and risk engine systems for banks and hedge funds including HSBC, JP Morgan and UBS, later serving as lead AI architect at Cognizant.
“As there is no understanding, machine learning does not explain its thinking so cannot be given autonomy in non-stationary situations,” he said of Vestun’s approach. “The intelligence of these applications can be boosted by modelling human expertise”.
