Vasant Dhar warns against ChatGPT investing advice
AI trading pioneer Vasant Dhar says chatbots can help investors think, but he would not hand them his own money untested.
By Frankie Delgado · News Reporter
4 min read
Vasant Dhar helped bring machine learning to Wall Street in 1994, but his view on ChatGPT investing advice is blunt: useful for thinking, risky as an autopilot.
In an interview with MarketWatch columnist Michael Sincere, Dhar said he would not ask a chatbot whether to buy a stock and then trade on that answer alone. Dhar is a professor at NYU Stern School of Business and the NYU Center for Data Science, and he founded SCT Capital, which MarketWatch described as one of the first hedge funds built around machine-driven trading.
His own algorithms still trade daily, according to MarketWatch. That makes his warning sharper: the issue is not whether machines can help investors, but when investors should trust them.
Should investors trust ChatGPT with money?
Dhar told MarketWatch that large language models can be helpful for long-term investment questions because they can reason through scenarios and uncertainty. He said that kind of work, such as thinking about Nvidia, BYD or the S&P 500, used to be mostly a human exercise.
The catch is consistency. Dhar said he asked ChatGPT about going long on Nvidia, BYD and the S&P 500, then asked again later and received a different response. In his view, the answer can shift depending on how the question is framed.
For people who lack the time or interest to study investments, Dhar’s advice was plain: consider an index they believe in. He told MarketWatch that investors who believe in long-term U.S. growth could buy the S&P 500 and leave it alone, while those betting on a small group of technology leaders should still do their own research.
He was even tougher on frequent trading by individuals. Dhar said retail investors should not try short-term trading unless they have a real system or algorithm, adding that trading more often can make losses arrive faster.
The AI bot that challenged a valuation guru
Dhar also discussed the Damodaran Bot, or DBOT, an AI valuation tool named for NYU finance professor Aswath Damodaran, whom MarketWatch identified as the “dean of valuation.” The bot is designed to value companies using Damodaran’s public methods and an investment thesis supplied by the user.
According to Dhar, the bot uses a free-cash-flow-to-the-firm model built around four drivers: operating margins, revenue growth, weighted cost of capital and reinvestment efficiency, measured by sales-to-capital ratio.
The appeal is speed. Dhar said Damodaran might publish one or two valuations a month, while DBOT can apply the method across the S&P 500 quickly.
One example he gave was SpaceX. Dhar told MarketWatch the company’s IPO valued it at $1.77 trillion, while Damodaran’s own estimate was about $1.2 trillion. DBOT came in around $600 billion, making it far more cautious than the human investor it was modeled after.
Dhar said the bot examined SpaceX’s launch business, Starlink and its AI business tied to xAI, and found too much uncertainty in the AI segment because of heavy capital needs and an unclear payoff.
Why AI trading could become a market risk
Dhar told MarketWatch that AI can be wrong, just as people can be wrong. His bigger concern is what happens if many investors eventually rely on the same systems and react to the same signal.
That could create herding, he said, making market shocks larger when the machine gets something wrong. Dhar said finance is not close to that point yet, though AI’s reach in markets is expanding.
He said it is still unclear whether AI will make markets steadier or more volatile. The outcome, in his view, depends on the guardrails people build around the technology.
Dhar’s career gives the warning a long runway. He told MarketWatch that he brought machine-learning methods to Morgan Stanley in 1994 after academic work and experience at ACNielsen, later built models at Deutsche Bank and spun out SCT Capital in 1998. Today, he said, his work includes deep learning, vision models and systematic commodity signals for sovereign wealth funds.
This story draws on original reporting from MarketWatch.