Explainable Financial Forecasting Using Collaborative Artificial Intelligence Agents

Abstract

The stock market is highly dynamic and difficult to predict due to continuous fluctuations influenced by economic conditions, investor sentiment, company performance, and global events. Traditional stock prediction systems mainly depend on single-agent models, which often suffer from limited analytical capability and reduced prediction reliability during volatile market conditions. This paper proposes a Multi-Agent AI-Based Stock Market Analysis and Investment outcome that integrates multiple intelligent agents, machine learning algorithms, deep learning models, and sentiment analysis techniques for improved stock market forecasting. The system combines market analysis, sentiment analysis, risk evaluation, technical analysis, and decision making agents to provide Buy, Hold, and Sell recommendations. Historical stock data is collected using Yahoo Finance APIs and processed using Random Forest, XGBoost, and Long Short Term Memory (LSTM) models for trend prediction. The paper achieved 89.42% accuracy using Random Forest, 92.15% accuracy using XGBoost, and 93.28% prediction performance using the LSTM model. The multi-agent framework also demonstrated higher analytical efficiency with decision agent accuracy of 95% and sentiment agent accuracy of 91%.