AI model empowers quantitative investment in financial market: research based on alternative data and traditional financial data

This document is used to obtain stock trading signals from textual data and traditional financial data, and conduct backtesting of investment trading in actual markets. The first step is data collection: divided into three parts: text data acquisition, traditional financial data acquisition, and text data cleaning; Step 2, Bert_tune: used to obtain the sentiment orientation corresponding to the cleaned text data in the first step; Step three, a composite indicator constructed by ChatGPT: used to obtain stock trading signals from traditional financial data; Step 4, Trading Strategy Backtesting: Three trading strategies: ChatGPT constructed trading strategy, text data trading strategy, and "text+chatGPT" composite trading strategy