Trying to beat analytical models with LLMs in the stock market
A trading-agent experiment that started with FINSABER, wrapped ARIMA-style signals in Mercury 2, and ended with the analytical baseline winning.
A trading-agent experiment that started with FINSABER, wrapped ARIMA-style signals in Mercury 2, and ended with the analytical baseline winning.
Adjusting for population helps Japan a little. Germany is why scale cannot explain the gap with Northern Europe.