Development of Multivariate Stock Prediction System Using N-Hits and N-Beats

The capital market serves as a pivotal hub within a nation's financial ecosystem, facilitating the exchange of stocks and securities. It assumes a paramount role in propelling the country's economic growth and development. Profitable decisions in the capital market are frequently encountered; nevertheless, the intricacy lies in the challenge of accurately forecasting unpredictable fluctuations in stock prices. The analysis and forecasting of stock price movements have emerged as a highly sought-after area of research. The forecasting of stock price movements can be effectively categorized into two primary domains: technical analysis and fundamental analysis. Time series analysis, commonly referred to as technical analysis in the realm of stock market analysis, involves the meticulous examination of a stock's price movements over a specific period. The forecasting technique employed in this study involves the analysis of Multivariate time series data using the advanced Neural Hierarchical Interpolation for Time Series Forecasting (N-HiTS) methodology. Based on the findings of the conducted research, this methodology exhibits commendable predictive efficacy in both the short and medium time horizons. Moreover, it demonstrates a notable ability to accurately forecast long-term stock patterns.