Application and Modeling of LLM in Quantitative Trading Using Deep Learning Strategies

After more than 100 years of development, with the breakthrough of computer technology, deep learning and big data industry, the quantitative trading market has gradually matured, and more and more investors have begun to use quantitative trading to invest. Quantitative trading automatically executes transactions through written programs, eliminating the interference of human subjective factors on transaction execution. But the threshold for quantitative trading is high, requiring researchers to have a deep understanding of mathematics, statistics, finance, and computer technology. The newly emerged Large Language Model (LLM) can help users get started to a certain extent, by giving the general framework of the code, so that users can have a preliminary understanding of the countermeasures faster and more accurately. In terms of model training and testing, this paper adopts the CSI 300 index obtained from tushare platform to study the results of daily data, weekly data and monthly data after training. This project trained a stock price prediction model using long short-term memory (LSTM) methods. Then, the backtest model was established with the classic double moving average strategy in quantitative trading, and the backtrader platform was used to visualize the return results simulated by the backtest. Finally, we discussed the risks of using LLM codes to execute quantitative trading.