Summary

• Introducing StockGPT, an autoregressive model utilizing generative AI to forecast stock returns based on a vast dataset of U.S. stock returns.

• Key Features:

  • StockGPT predicts stock returns by treating return series as sequences of tokens and learning patterns predictive of future returns through its attention mechanism.

  • On a test sample, the model’s predictions resulted in an annual return of 119% with a Sharpe ratio of 6.5, spanning various market factors and trends.

    Importance

• StockGPT has potential to redefine investment strategies by automating and improving stock return predictions, thus aiding investors in making more informed decisions.

• The model demonstrates the applicability and effectiveness of generative AI in financial fields, paving the way for further innovations in AI-driven financial analysis and decision-making.