{"text":[[{"start":8.7,"text":"One of the biggest names in quantitative investing has warned against handing over investment decisions to artificial intelligence, arguing that some corners of the hedge fund industry have gone too far in adopting the technology. "}],[{"start":20.9,"text":"Martin Lueck, who was one of the three founders of quant pioneer AHL and is co-founder and president of hedge fund Aspect Capital, said that being able to understand exactly why its computer models were recommending trades was crucial for his firm. "}],[{"start":35.3,"text":"“My starting place is I’m not going to put my name on or my company’s reputation on something [where] . . . I have no idea why it took these positions,” said Lueck, whose firm manages $9bn in assets, in an interview with the FT."}],[{"start":50.599999999999994,"text":"“I need to have some kind of hypothesis I can hook my hat [on]. If I’m investing my own money in it I want to know what it’s doing,” he added."}],[{"start":59.55,"text":"Lueck’s comments come as hedge funds and proprietary trading firms deepen their use of AI and machine learning techniques. Billionaire Cliff Asness, founder of US quant giant AQR, last year told the FT it was “surrendering more to the machines”. "}],[{"start":76.19999999999999,"text":"Asness admitted his fund was latching on to patterns that his quant researchers sometimes could not explain, and was using machine learning to decide how much money to put into trades. "}],[{"start":86.79999999999998,"text":"“It’s been easier that this has been a very good period for us after a very bad period,” said Asness at the time. “Odds are it will be a little harder to explain [to investors] in a bad period, but we think it’s clearly worth it.”"}],[{"start":100.29999999999998,"text":"Lueck told the FT he had listened to Asness and found him “persuasive”, but added that one of the key reasons in 1995 that he left Man Group, which had by then acquired AHL, was to give investors more information about what was driving the trading models. "}],[{"start":116.69999999999999,"text":"“In those days it was a black box,” he said. “There was no transparency on what these models were doing and the whole [quant] world was opaque.”"}],[{"start":126.6,"text":"Quantitative investing has traditionally used in-depth research, often drawing on academic papers, to try to identify and explain market patterns that can then be traded by computer algorithms. Such programmes typically have rules determining how much money to invest and when to adjust leverage based on market risk. "}],[{"start":145.4,"text":"Lueck was one of the early pioneers of making money by following market trends — a strand of quant investing that aims to profit from investor herding into or out of trades — when he co-founded AHL alongside Michael Adam and David Harding. "}],[{"start":159.35,"text":"But the advent of more computing power and ever more sophisticated AI and machine learning techniques has tempted hedge funds to delegate more responsibility to machines. "}],[{"start":169.75,"text":"Lueck added there are great opportunities in using large language models to assist quantitative researchers in tasks such as organising data, running tests and preparing presentations to colleagues. "}],[{"start":181.3,"text":"But he added: “I still want the researcher to be thinking about what is it that I’m actually researching rather than here is some data, find me some relationships.”"}],[{"start":198.5,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1776923429_9092.mp3"}