Loading...
[Can AI usage be used for stock selection? U.S. stock backtesting shows a weekly difference of 0.64% between high and low exposure groups] According to monitoring by Beating, three economists conducted backtesting by combining real AI usage data with U.S. stock market performance. The study spans from January 2024 to April 2026 and utilizes approximately 380 trillion tokens on OpenRouter. OpenRouter is an AI model aggregation platform that allows users to access over 400 models through a single interface. The researchers synthesized weekly token usage, spending amounts, and active user growth rates into an 'AI Usage Index' and then observed the stock price reactions of each U.S. stock over the past 13 weeks. Stocks that tend to outperform the market when AI usage accelerates and underperform when AI usage slows were categorized into the high exposure group. Each week, they bought stocks in the high exposure group while shorting stocks in the low exposure group. Backtesting results showed an average weekly difference of 0.641 percentage points between the high and low exposure groups. After controlling for factors such as company size, valuation, profitability, and stock price momentum, the difference remained around 0.56 percentage points. However, the sample size is limited to 28 months, OpenRouter only covers a small portion of global AI usage, and its users are predominantly developers. Whether this method can be effective in the long term remains to be verified. [Original link]