Q&A: Grant Fuller, Irithmics
Grant Fuller, CEO of UK AI fintech Irithmics, discusses developments in artificial intelligence and how hedge funds are using AI in their portfolio construction.
AW: Grant, artificial intelligence has actually been around for decades. Tell us why it’s only in the past few years that it has become more front and centre of the finance industry.
GF: Many of the building blocks of modern AI and machine learning have indeed been around for decades: from logistic regression in the 1950s, artificial neural networks in the 1970s, random forests and reinforcement learning in the 1990s. Two main developments have catapulted AI, allowing it to advance at a spectacular pace. Firstly, the ability to process, store and make use of large volumes of data, and secondly the advance of computer infrastructure, specifically parallel computing and GPUs. These two advances meant computer scientists and mathematicians have developed more advanced and elaborate algorithms to process data, discovering patterns and relationships which we weren’t able to appreciate before. This, in turn, means machines are able to accomplish things we’d only thought possible in science fiction: photorealistic images of human faces, hotel rooms, cars or emotional and evocative music composed by AI. The technology is getting better all the time; AI’s been shown to outperform qualified doctors at diagnosing cancers, and qualified lawyers at identifying issues with clauses in agreements.
AW: Hedge funds have always been quick adopters of technology that helps in the hunt for alpha. What are you seeing specifically with regards to how hedge funds are using AI?
GF: Hedge funds have always sought to differentiate themselves, either through their strategy - how they capitalise on market opportunities and mitigate risks - or how they research or execute on that strategy. While AI is topical now, the buildings blocks have been around for years, and many fund managers have taken full advantage of the technology – some with stunning effectiveness. There’s a growing trend looking to unlock the potential of new data sources (so-called alternative data), and managers are using AI and machine learning techniques to help evaluate these new data.
We’re also seeing the technology being used in ways other than trading and portfolio management. Managers now profile and screen investors, helping them understand what investors are looking for and how to more effectively demonstrate their investment case, identifying the opportunities and risks of the fund as the investor is likely to perceive them. So important are these advances in investor screening and fund development they were the subject of Fidelity International’s investment conference in Hong Kong and Singapore last October.
AW: One way that your firm, Irithmics, helps its clients is by analysing risk in an equity portfolio. What is the application of AI here?
GF: Irithmics uses deep neural networks to understand the behaviour of global institutional investors and fund managers: how they build portfolios and how they mitigate risks. This is important because, for the first time, we’re able to measure and monitor the risk to a portfolio due to the behaviour of other market participants. Like value-at-risk, fund managers can now begin to understand the impact on their P&L related to the action of other portfolio managers, fund managers can begin to see where other portfolio managers are likely to be making strategic or tactical allocations or when a particular stock is increasingly vulnerable to selling or shorting.
One exciting feature of Irithmics’ work with deep learning investor behaviour has been on target prices. These are typically associated with brokers or research analysts, but state-of-the-art deep learning now makes it possible to imply target prices from the behaviour and portfolios of the buy-side. Fund managers can now estimate what the portfolios of other investors imply about their assessment of a company’s performance.
AW: It seems that AI is being used by scaremongers who think it will have a significant dislocation effect on jobs. What’s your take on that with respect to hedge funds?
GF: Humankind’s ability to understand our world and create tools to help us manipulate it are attributes which define us as a species – it’s what we do, and it inevitably shapes our development. The first person to put a fence around a cow or plant seeds in the ground started changing the role of the hunter-gatherer. It happened with the steam train and electricity, it’s happening now. In the early 1800s, some English workers were so concerned with machines taking their jobs (especially in cotton and woollen mills) they formed bands to sabotage and destroy them – they became known as the Luddites. There is no doubt that technology changes jobs, after all, that’s why it was invented! A recent Harvard and MIT report stated that AI was unlikely to take the jobs of Americans, but colleagues [and competitors] who knew how to use AI would. I think this is very true; AI will provide a disproportionate advantage to those who are able to use it.
The hedge fund industry already uses AI extensively, either directly or indirectly. Managers themselves, Prime Brokers, Fund Administrators, Auditors, Exchanges and Regulators are all finding applications for the technology. The technology will impact how hedge funds operate and how they are evaluated by investors.
AW: Finally, are we at the beginning, middle or end game of the AI zeitgeist and what else do you expect to see AI offer hedge funds in the next five years?
GF: Niels Bohr said predictions were difficult, especially when they involved the future. With Bohr’s warning ringing in my ears, I think it’s safe to say AI and our applications of the technology are in its infancy. We haven’t even scratched the surface.
Hedge funds will use the technology to help tap new sources of alpha, improve operational efficiencies, attract and service investors. The technology will begin to help fund management companies themselves run better businesses.
Investors will increasingly use the technology to screen and monitor funds and their strategies – we’re already helping investors understand fund strategies, and this is only going to become more widespread. AI is already able to evaluate individual fund strategies, but this is going to become much more significant for investors.
Now for the fun speculative prediction: it’s interesting to think about how new products can be created, and we’re already seeing the embryo of these ideas. Larger managers with many fund products are looking at of these can be wrapped to form structured products of funds. Pushing the speculation even further, you can see how downside protection products might be created, and there are a couple of interesting academic projects looking at this.
Grant Fuller is CEO of Irithmics
© The Sortino Group Ltd
All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency or other Reprographic Rights Organisation, without the written permission of the publisher. For more information about reprints from AlphaWeek, click here.