As fund management continues to evolve, one of the main areas of innovation and investment has been in the field of artificial intelligence.
Today’s stock markets move incredibly fast, presenting fund managers with a torrent of data to gather, decipher and act on.
Investors face the practical problem of how they apply their intelligence and insight to that data without being overwhelmed by it, trying to avoid an almost paralysis of decision-making.
When markets turn volatile that challenge is even greater. That is where artificial intelligence comes in.
Humans now lag far behind computers in their ability to analyse market data quickly.
Humans will also struggle at making decisions in periods when emotions are running high, especially when events, like conflicts or disasters go beyond market implications. So using dispassionate, rational, calculating computer processing power to help us cope with that data overload seems an obvious solution.
AI can help investors, at a minimum, on three specific aspects: gathering and interpreting the relevant data; building a framework to determine the best time to trade, or type of trade to take; and continually evolving (or learning) with every event.
For a fund, an AI-informed model can do the data analysis of perhaps 10 fund managers.
By building AI models and programmes in advance, fund managers can establish what their trading strategy is going to be under a range of scenarios (including tail events).
Without AI, that kind of scenario planning would be practically impossible. It means that even in the most difficult of market conditions, AI can provide you with an appropriate game plan.
Some practical ways that we’ve used AI include identifying when to hedge out equity risk in periods of high market volatility, or using momentum trades in periods of low volatility, or knowing when to transition from a high growth portfolio to one with a greater emphasis on value or income.
Using AI will also allow you the opportunity to test and challenge planned trades, helping to identify and remove in advance some of the human biases that tend to lead to underperformance – for example, holding on to a loved stock too long, or trading at the wrong time.
However, for AI to work properly and deliver the outcomes expected, it’s important to know the difference between good AI and bad AI.
Good AI and machine-learning (just like a star fund manager) will be able to adapt to information it has never seen before and still deliver the right outcomes.
The model you build has to be immaculate, as well as have the ability to evolve and adapt appropriately.
In an ideal world, this model would be highly transparent to investors, a type of ‘glass box’, to allow them to know how their money is being managed.
