Stock Market Prediction With Risk Analysis Using Two ml Module

Abstract

Attempts to predict movements in stock prices have historically proven difficult for academic researchers. Seminal publications in the literature have shown that the seemingly random movement patterns of stock price time series may be anticipated with a high degree of accuracy, contrary to the claims of proponents of the efficient market theory. Such risk-adjusted prediction models require appropriate variable selection, variable transformation processes, and model parameter adjustment. This study proposes a methodology for predicting stock prices using a combination of statistical analysis and machine learning that is both predictable and accurate in its risk analysis. We utilize five-minutely daily stock price data from a major company listed on India's National Stock Exchange (NSE). When building and training forecasting models, the granular data is aggregated into various time slots throughout the day. We propose that agglomerative model development, combining statistical and machine learning approaches, may successfully learn from unpredictable and erratic movement patterns in stock price data while mitigating risk. Using this effective learning, models can be trained to be robust and low-risk, increasing their utility for predicting stock movement patterns and short-term stock prices. Regression and classification models are built using statistical and machine-learning techniques. A large amount of data on these models' efficacy has been provided and thoroughly examined.