For a hypercompetitive industry deadset on finding an edge over peers, hedge funds have been slow to integrate artificial intelligence into their most important functions.
But, as AI advances continue to impress and sometimes even scare companies and governments, the industry is realizing it can potentially replace its most valuable commodity — investing talent. A set of former hedge funders from firms like Bridgewater have started companies to do just that.
Asset managers, thanks in part to their significant financial resources, have been at the forefront of developing AI and machine-learning algorithms internally, but the use of these technologies has mostly been to make the least sexy parts of hedge funds — regulatory requirements, legal review, tech support — more efficient. In the last few months, LLMs and AI agents have become increasingly sophisticated, giving hedge funds greater confidence that they can better automate research, analysis, and other work done by humans.
Billionaire Citadel founder Ken Griffin, once skeptical of AI’s ability to beat the market, told an audience at a Stanford talk that he was “fairly depressed” by how good some of his firm’s new agents were. A project that once took an employee with a master’s or Ph.D. weeks could now be done in hours, he said.
As a result, the technology might level the playing field between funds with near-unlimited resources and up-and-comers trying to carve out their own niche, and turn the industry’s status quo on its head.
The hedge funders turned AI founders
With more information to process and investors moving faster than ever, the last 15 years have been tough for smaller hedge funds.
“The proliferation of alt data and advanced risk models have rewarded scale, but we believe LLMs are a democratizing force, empowering them to process all of that information,” said Ian McInnis, founder of Y-Combinator-backed startup WithAI.
Ian McInnis
McInnis, a former Bridgewater analyst, isn’t alone in seeing the opportunity to help integrate AI into the investment process. In the last few months, a crop of companies with founders hailing from funds like Schonfeld, Jain Global, and more have emerged to build tools and technologies that don’t just assist hedge funds’ investment staffers — they can replace them.
Macro Technologies, cofounded by former Schonfeld and Citadel Securities macro researcher Jaime Villa, wants to “automate” repeatable tasks that funds’ macro analysts do. Serona Data, led by former Jain Global executive Cameron McKendrick, finds investing signals in healthcare data, a needle-in-the-haystack type of exercise once left to humans who could detect nuance.
These former hedge funders believe there’s plenty of demand for investment-talent-replacing AI from the industry. WithAI already counts four different funds as clients, and McKendrick said his firm developed its first dataset in conjunction with a large multistrategy manager.
It’s a sea change for the industry even compared to last year. At a London conference last October, quant executives were complimentary of AI, but believed the human on the other side of the screen was where investing edge lay.
Now, Bridgewater has partnered with Mira Murati’s Thinking Machine Labs to create an LLM that can “interpret text with expert-level taste and judgment” so humans don’t need to sift through financial documents.
A talent war reprieve?
The advancement and adoption of this technology might help cool the raging talent war that has driven up the costs of running a hedge fund.
The biggest expense for hedge funds is talent, and the most expensive talent is the people making the investment decisions — the portfolio managers and analysts poring through spreadsheets, tweaking models, and interviewing corporate executives.
If some of this work can be outsourced or automated, then firms’ costs should drop. For example, Claire Brown, the founder of stock-picking fund Aristides Capital, posted on X that “90% of the answers that Claude generates in 5 minutes are better than the vast majority of people could generate in an hour.”
In June, Brown wrote that “the latest version of Claude is as good as a junior analyst” who would get paid more than $100,000 a year; Claude’s monthly subscription is a fraction of that.
“I’m pretty sure there’s a massive ROIC there,” she wrote, referring to the acronym for return on invested capital.
Jaime Villa
Jan Szilagyi, CEO of AI platform Reflexivity and the former co-CIO at Lombard Odier, said he expects AI to pick stocks on its own soon as it learns from investors’ practices. Villa, the macro researcher, said his firm’s mission is to “convert the most valuable workflows in asset management” into something an AI can replicate.
For Stephen Wu, these advances could help him scale. Wu founded Carthage Capital, a $50 million hedge fund that blends quantitative and fundamental techniques to pick stocks in 2023. His firm was the first investment manager to join Carnegie Mellon’s VentureBridge program, and Wu hopes the program will help him on the “AI side.”
He wants to replicate his thought process to trade more stocks, he told Business Insider in an interview. Right now, he can trade between 10 and 20 stocks. With the right AI agents, he could see his fund trading hundreds of stocks.
“There are many different areas where intuition can be replaced,” he said.
