Enhancing Automated Trading with Sentiment Analysis: Leveraging Large Language Models for Stock Market Predictions

Аннотация

This study explores the use of Large Language Models (LLMs) for automating investment strategies through sentiment analysis of financial news, social media, and market data. By fine-tuning models like GPT-3 on financial datasets, sentiment indicators are extracted and integrated with traditional machine learning algorithms to predict stock price movements. A comparative analysis of various models, including LLM-based, traditional machine learning models, and hybrid approaches, was conducted. The results reveal that the hybrid model, combining LLM-generated sentiment with machine learning algorithms, outperforms other models in terms of both prediction accuracy and financial performance. The hybrid approach achieved an accuracy of 77.4%, cumulative The American Journal of Engineering and Technology 185 https://www.theamericanjournals.com/index.php/tajet The American Journal of Engineering and Technology returns of 17.2%, and a Sharpe ratio of 1.20, research is to develop a robust automated trading demonstrating its potential for real-world trading strategy that integrates sentiment analysis with market applications. These findings highlight the importance of data, ultimately enhancing the ability of investors to sentiment data in enhancing market predictions and make informed decisions based on real-time sentiment provide a promising framework for automating shifts. investment strategies. However, challenges such as Literature Review ambiguity in sentiment classification and the need for Sentiment analysis has emerged as a crucial tool for model adaptation to changing market conditions understanding investor behavior and its influence on remain. Future research should focus on improving financial markets. In traditional finance, market sentiment analysis accuracy and incorporating movements were often seen as driven by tangible reinforcement learning for real-time trading. economic factors such as earnings reports and macroeconomic indicators. However, recent studies Keywords: Large Language Models (LLMs), sentiment have highlighted the importance of psychological analysis, financial markets, automated investment factors and sentiment in shaping market trends (Fama, strategies, hybrid models, machine learning, prediction 1970). Investor sentiment, which refers to the overall accuracy, stock price movements, back testing, Sharpe mood of market participants, can often drive market ratio. prices independently of fundamentals (Shiller, 2000). Sentiment analysis, therefore, plays a critical role in understanding market behavior by capturing this Introduction: In recent years, the use of artificial intangible yet influential aspect of trading. intelligence (AI) and machine learning (ML) has The application of sentiment analysis in financial revolutionized the financial industry, particularly in the markets has evolved significantly with advancements in development of automated trading strategies. One of Natural Language Processing (NLP) and machine the most promising areas of AI in finance is sentiment learning. Early sentiment analysis methods involved analysis, which involves extracting insights from textual rule-based systems that analyzed text to classify data to predict market movements. With the advent of sentiment as positive, negative, or neutral. More Large Language Models (LLMs) like GPT-3, it has recently, machine learning models have been become possible to analyze vast amounts of employed to improve the accuracy of sentiment unstructured data from diverse sources such as extraction (Pang & Lee, 2008). These models are financial news, social media, and blogs. These insights trained on labeled datasets of financial news articles, can be leveraged to automate investment strategies social media posts, and other relevant textual data to that respond to changes in market sentiment in real identify sentiment indicators that correlate with stock time. price movements (Chen, 2013). Financial markets are influenced by a wide range of One of the most significant breakthroughs in sentiment factors, including macroeconomic indicators, corporate analysis came with the development of deep learning earnings, geopolitical events, and investor sentiment. models, particularly Recurrent Neural Networks (RNNs) Traditional methods of stock price prediction typically and Long Short-Term Memory (LSTM) [1,2] networks, rely on structured data such as historical stock prices, which can process sequential data like text (Hochreiter trading volume, and other financial metrics. However, & Schmidhuber, 1997). These models have shown unstructured data—especially from news articles and promise in capturing the temporal dependencies in social media platforms—has become increasingly financial sentiment, where market sentiment at one important in shaping market sentiment, which can, in time point can influence future market movements. turn, influence asset prices. The combination of Recent studies have applied LSTMs to predict stock str