{"text":[[{"start":8.05,"text":"Robinhood is launching a feature that will enable clients to use artificial intelligence chatbots for share trading, underscoring the growing “arms race” among brokerage platforms to offer retail investors the technology. "}],[{"start":21.6,"text":"The California-based company said on Wednesday that traders with new “agentic trading accounts” would be free to use AI programs, including Anthropic’s Claude Code and OpenAI’s ChatGPT to build portfolios, automate trading strategies and place orders. "}],[{"start":37.5,"text":"The announcement from Robinhood, which is popular among day traders, comes two months after rival platform eToro introduced agentic trading on its app and Tiger Global-backed Public rolled out agents that investors can prompt to execute trades. "}],[{"start":54.2,"text":"“There is going to be an arms race this year among all of these brokerages to offer agentic trading to their users,” said Stephen Mackintosh, chief investment officer at Sui Group Holdings, which has backed a start-up called Nof1, which is building an AI model for financial markets."}],[{"start":70.7,"text":"“It’s a trillion-dollar opportunity,” he added. "}],[{"start":73.75,"text":"AI chatbots have been used to inform retail traders’ investment decisions since the launch of ChatGPT in 2022, though there is scant academic literature about whether AI-generated recommendations lead to better or worse returns."}],[{"start":88.8,"text":"Last October, Nof1 gave six leading large language models $10,000 to trade on Hyperliquid, a crypto platform for trading derivatives. Models were given a single goal: to maximise profit using mid-to-low frequency trading, where decisions are made on the order of hours. "}],[{"start":104.94999999999999,"text":"The experiment found that models finished in profit only six times across 32 sets of results and showed significant differences in bias, including some that had a penchant for short-term bets."}],[{"start":116.49999999999999,"text":"The group also found that models were “highly sensitive to seemingly trivial prompt changes”."}],[{"start":122.74999999999999,"text":"Successful frontier models will also need the traditional datasets that trading firms such as Jane Street, Hudson River Trading and Citadel Securities use to train their models, Mackintosh said. "}],[{"start":133.7,"text":"“Out of the box LLMs are not suited to generating [returns] — they’re just generic probabilistic models,” he warned. "}],[{"start":142.04999999999998,"text":"Amanda Fischer, policy director at investor advocacy group Better Markets, said: “The typical retail investor does not benefit from more trading, but Robinhood’s fee income certainly benefits.”"}],[{"start":154.04999999999998,"text":"Robinhood on Wednesday warned that “AI-driven strategies may perform poorly under certain market conditions, move quickly, and may be difficult to monitor or stop in real time”."}],[{"start":171.2,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1779945358_2182.mp3"}