Machine learning and improving the investment decision making process
Exposing market participants’ data to algorithms exposes bias, which can creep in at many stages of an investment process due to human fallibility. Machine learning and AI can assist in understanding the evolutionary nature of markets but are there models dynamic enough to capture data-generating processes, allocate risk and position portfolios?
Sanjiv Kumar, partner and co-founder, Fort LP
Moderator: Laurence Parker-Brown, institutional content producer, Conexus Financial
- Capital markets arguable display the behaviours of adaptive expectations that produce empirical rather than universal statistics, meaning there is value in memory and gauging persistence.
- Machine learning can be deployed to capture these autocorrelations to provide an edge over human recall which is influenced by the availability heuristic.
- The recent influence of central banks has suppressed volatility to the detriment of many hedge funds meaning that those left in the industry have been able to thrive in challenging operating conditions.
What would you consider to be the appeal of quant/systematic managers to complement a portfolio?
- The liquidity profile of the strategy
- The time-tested uncorrelated return stream
- The repeatability of the quantitative approach recipe
- The strategy response to market shocks