Citadel's CTO says AI won't generate lasting alpha for hedge funds
Hedge fund Citadel is no stranger to the use of artificial intelligence. But its chief technology officer (CTO) does not subscribe to visions of the future in which AI displaces human beings who make investment decisions, or even humans who run quant trading strategies.
Get Morning Coffee ☕ in your inbox. Sign up here.
Speaking at the Milken Conference yesterday, Citadel CTO Umesh Subramanian, said that even if you were able to develop a quantitative trading AI that could be trusted to back test its own decisions "with reasonable confidence," it would not be a source of enduring alpha.
In a world where the AI was making the trading decisions, “everyone would know what it would do," said Subramanian. In that world, he said the alpha would live at the "frontier" where people were trying to innovate. And, “that is where human beings come in.”
Amidst the AI hype, Subramanian suggested AI is a tool rather than a revolution and that it's not even that new. He noted that machine learning has been used by Citadel's quant traders for 10 years: “Google came out with tensor flow and within like a week we put that to use to make money.”
Instead of actually making trading decisions, Subramanian said AI's real use is helping discretionary traders (humans) handle the huge "information surface." A human portfolio manager covering a universe of 50-80 stocks is "bombarded with news, filings, transcripts, research articles." AI therefore facilitates "productivity plus" whereby Citadel's portfolio managers (PMs) are given the information they need to make decisions more quickly. Subramanian said they interact with Citadel's AI tools via a chatbot: "You can talk to it. Human beings prefer to ask rather than click, click, click."
While AI has definite strengths like data assimilation and analysis of unstructured data, particularly alternative data, it also has weaknesses. It cannot be used in regulatory areas or for functions like trade matching, where exactitude is required and there's no margin for error. AI is most appropriate in situations where, "There is a probabilistic range of acceptable answers and the speed of getting that answer more right than wrong quickly is important," said Subramanian.
This means that to the extent that Citadel hires AI-focused professionals, it will only do so "at the small margin." AI is like any other "tool in the toolbox", Subramanian added. It's strength isn't predicting the future, but allowing people to see the present more clearly. "We are hiring data scientists and data professionals to embed in various businesses and to optimize the various workflows.” – But Citadel did exactly the same for cloud professionals a few years ago. "We got that talent from inside. We also learnt it from inside, and then let that go," said Subramanian, implying that pure AI roles may not endure.
Instead of focusing primarily on AI skills, Subramanian said Citadel therefore focuses on hiring people with "curiosity and problem-solving ability" and that this hasn't changed. The firm does encourage its curious humans to use AI, though. Ken Griffin himself uses it. "Ken Griffin uses ChatGPT," quipped Subramanian.
Have a confidential story, tip, or comment you’d like to share? Contact: +44 7537 182250 (SMS, WhatsApp or voicemail). Telegram: @SarahButcher. Click here to fill in our anonymous form, or email editortips@efinancialcareers.com. Signal also available.
Bear with us if you leave a comment at the bottom of this article: all our comments are moderated by human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. Eventually it will – unless it’s offensive or libellous (in which case it won’t.)