An experiment with ChatGPT brought a teenager a return six times higher than the Russell 2000

Nathan Smith, a teenager from rural Oklahoma, entrusted ChatGPT with complete control over a $100 stock portfolio. Over four weeks, the experiment yielded impressive results: returns reached 23.8%, significantly outperforming the Russell 2000 Index and the biotechnology-focused XBI Index.

The idea came to Smith after he saw an advertisement for stock selection algorithms on YouTube. He decided to test whether artificial intelligence in the LLM format could independently manage investments in the microcap segment. According to him, he was surprised that such comprehensive experiments had hardly been conducted before.

The task was clear: to assemble a portfolio of American companies with a market capitalization of less than $300 million and maximize returns over the course of a month. Smith emphasized that ChatGPT was given complete autonomy in management: from selecting securities and position sizes to setting stop losses.

The risk profile turned out to be aggressive. The Sharpe ratio was 0.94, indicating a high level of risk, but Sortino at 2.0 showed limited drawdowns and significant growth potential. One of the key decisions was the purchase of CADL shares, which accounted for about half of the total profit.

The young man noted that AI sold CADL securities without hesitation, reflecting discipline in microcap management. According to him, even many hedge funds would not have dared to take such a step.

To manage the process, Smith created five main functions: manual transactions, portfolio processing for tracking, daily calculation of results via Yahoo Finance, and generation of comparative charts with the S&P 500 index. The algorithm updates the portfolio weekly, and Smith only executes transactions and records the results.

The teenager plans to continue the experiment until December, hoping for a longer period of analysis. “I think I’ve found my calling in life and I really hope to make it my profession,” he said.

Recall that back in February 2024, institutional investors predicted the growing role of AI in trading.