ANALYSIS AND SHORT-TERM PREDICTION OF STOCK MARKET PRICES USING VAR MODELS OF TIME SERIES

Forecasting stock assets prices remains a vital area in financial analytics, driving ongoing research into more effective predictive techniques. This study explores whether the price trends of a given stock can be forecasted based on the price movements of other related stocks. Our approach seeks to uncover inter-stock relationships and predict price trends by analyzing these connections. Leveraging neural networks, we identify clusters of companies exhibiting similar price movement patterns. Subsequently, we employ a Vector Auto Regression (VAR) model to generate impulse response functions, projecting potential price fluctuations over the following days. Our experiments focus on refining both the accuracy and computational efficiency of these predictions. Additionally, we developed an interactive dashboard that visually displays forecasts of stock price changes, enabling users to observe anticipated trends in companies’ stock prices. This work contributes to the growing field of predictive modeling by enhancing understanding of interdependencies among stocks and providing accessible predictive insights for financial decision-making.