Hedge funds seek an edge by using AI’s speed

{"text":[[{"start":6.9,"text":"When a merger announcement hits the wires, Sand Grove Capital Management has to be ready to act quickly. The hedge fund makes money from investing around big, corporate events such as M&A deals and is competing with others who do the same. "}],[{"start":20.8,"text":"Traditionally, reading through the complex, lengthy deal documents — which often stretch to over 100 pages — would take an investment professional over an hour. Even a quick review would take 15 to 20 minutes. But the use of AI has now reduced the process to seconds. "}],[{"start":37.7,"text":"“We think of AI as a very fast, very thorough intern who is brilliant at analysing big datasets,” says Daniel Caplan, chief executive of London-based Sand Grove. "}],[{"start":48.45,"text":"“In the first minute, you can get a good breakdown of what’s in it,” adds Caplan, whose firm uses Anthropic’s Claude, Microsoft Copilot and OpenAI’s ChatGPT. It finds Claude particularly useful for legal documents."}],[{"start":61.400000000000006,"text":"Sand Grove is just one of the hedge funds that is increasingly keen to reap the potential speed and productivity benefits offered by AI. But many remain wary of the risks of giving the fast-emerging technology too much access to their trading systems, internal data or cyber security arrangements."}],[{"start":null,"text":""}],[{"start":79.55000000000001,"text":"While many computer-driven hedge funds have been pushing the boundaries of quantitative techniques, including machine learning and artificial intelligence, for years, other firms are newer to this world. "}],[{"start":90.45000000000002,"text":"Last year, the Alternative Investment Management Association (AIMA), an industry body, found in a survey of hedge funds managing a total of around $788bn that 95 per cent of respondents were using AI, with 75 per cent saying they were using it more than before. ChatGPT and Copilot were the most popular tools."}],[{"start":111.75000000000001,"text":"The most frequent uses of AI were for general research, analysing and summarising documents, and providing summaries and minutes of meetings."}],[{"start":120.15000000000002,"text":"“At the moment there’s a high level of European deal flow,” says Caplan. “AI has definitely helped compress the time between an announcement and us initiating a view.” "}],[{"start":null,"text":""}],[{"start":129.35000000000002,"text":"The survey also found that 58 per cent of respondents expected to use AI more in their investment process over the coming year, up from 20 per cent in 2023."}],[{"start":140.40000000000003,"text":"“AI is increasingly being embedded across the internal value chain, particularly in areas like processing company filings, summarising complex legal documents and structuring large volumes of data,” says Wilson Chan, founder and chief executive of UK-based Permutable AI, which provides AI-driven financial market data."}],[{"start":159.65000000000003,"text":"“The most immediate gains have been in automating workflows typically handled by junior to mid-level staff, as well as enabling much broader scanning of potential alpha [returns in excess of a benchmark index] opportunities,” he adds. "}],[{"start":171.35000000000002,"text":"Pharo Management, an emerging markets-focused firm managing around $8.3bn in assets, is treating AI “as a strategic priority”, says Negar Laing, a director at the firm."}],[{"start":183.90000000000003,"text":"It is exploring ways to use the technology, for instance, to process information to help support portfolio managers, in software development or in its back office — the part of a hedge fund firm that performs tasks such as trade settlement and fund accounting, which proponents of AI believe can be automated."}],[{"start":202.60000000000002,"text":"Renaud Saleur, a former trader at Soros Fund Management who now runs Geneva-based hedge fund firm Anaconda Invest, says he uses Claude to write investment reports for clients and also to screen companies’ financial ratios."}],[{"start":218.00000000000003,"text":"But many funds remain wary of giving AI too much access or control of their internal systems or trading. The recent revelation that Anthropic’s new Mythos AI model broke out of a secure digital environment to contact a company researcher and publicly post details of software glitches online will only increase that caution."}],[{"start":238.20000000000002,"text":"AIMA’s survey found the top two risks that hedge funds saw from using generative AI were hallucinations — instances in which the system generates false content — cited by 64 per cent of respondents, and data security and privacy, cited by 83 per cent. Only 5 per cent of funds used AI for cyber security, while even fewer used it for finance and accounting or optimising their portfolios."}],[{"start":null,"text":""}],[{"start":263.6,"text":"Odi Lahav, chief operating officer at consultancy Bfinance, sees potential for AI in accurately summarising documents. He says AI’s accuracy is getting “better and better” and that he has not seen a hallucination for a couple of months."}],[{"start":278.40000000000003,"text":"But he adds that nothing leaves his firm without being reviewed and signed off by an investment professional."}],[{"start":285.6,"text":"“You can ask the model to interpret [documents or data] and give its views, but [taking an investment decision] is best left to an experienced professional,” he says. “You can’t rely on it completely.”"}],[{"start":297.6,"text":"Neither Anaconda nor Pharo use AI to select the securities the firm trades, for instance."}],[{"start":304.25,"text":"“Having a judgment layer that sits above AI is still very important,” says Sand Grove’s Caplan. “I can’t see how that human element will change, even in the long term.”"}],[{"start":321.8,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1777899229_2462.mp3"}