Stock Price Prediction using HFTSF Algorithm

Abstract

Forecasting is still a potential area of research, particularly in the stock market. Any forecasting model must overcome the subjective nature of the factors that affect market oscillation. Current fuzzy models have made an effort throughout the years to improve financial market forecasting accuracy. The fuzzy returns of the phenomena under study contribute to reducing the subjective nature of the financial market, particularly with respect to the effect of human emotions. These are based on large part on fuzzy sets. Fuzzy sets, on the other hand, may not fully satisfy or characterize the ambiguity of the data since they are unable to depict the level of neutrality of time series. Existing fuzzy inference systems’ reliance on a univariate framework is another important and crucial shortcoming. However, the time series that are part of a prediction problem frequently interact with one another. Given these factors, it is important to create a hybrid fuzzy system for a time series prediction issue that is built on fresh fuzzy sets and a collection of fuzzy logic relations. In this context, this research suggests a hybrid fuzzy time-series forecasting model (HFTSF) on the Standard & Poor Bombay Stock Exchange Information Technology (S& P BSE IT) index, for the prediction of time-series data. This model boosts the chances of getting better forecasts. The validation techniques such as root mean square error, mean square error, and mean absolute error were used in terms of validating the predicting outcomes.