Prediction of stock markets using artificial intelligence algorithms
Loading...
Date
item.page.authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Forecasting the stock prices for future is a major work of concern
newlineas it helps in earning massive profits for stock traders, investors and stock
newlinebrokers. There are various features affecting the market prices like demand
newlineand supply of the product, company health, sentiments of traders, as well as
newlinegovernment policies. These features result in stock market fluctuation and
newlineleads to uncertainty in the stock trend curve. This research work aims to
newlineprovide a reliable enhanced prediction model by alleviating problems like
newlinelook ahead bias, overfitting due to number of input features, and noise in the
newlinedata. This research work provides three contributions in an effort to mitigate
newlinethe volatility associated with the stock trend curve. The first contribution aims
newlineto select the best features for the prediction model to overcome the problem of
newlineoverfitting and provides a mathematical method to overcome the local minima
newlineproblem encountered during feature selection. This contribution presents a
newlinereliable and efficient model for short term prediction. A Levy Flight-based
newlinetuning for the prediction model are incorporated in the Support Vector
newlineRegression based prediction model. The primary features based on which the
newlinemachine learning model predicts the prices are selected by a tailored Levy
newlineFlightprobability distribution model instead of random modelling. The values for
newlinethe parameters of the modified Support Vector Regression model are tuned by
newlinethe Antlion algorithm which specifically selects the best values for the current
newlinestock prediction problem taken in hand. The prediction accuracy is found to
newline be higher with this metaheuristic-based approach and the tuned parameters
newlineyield high confidence value than the basic regression modelling of data.
newlineThe Covid-19 pandemic has hit the stock markets and the trends of
newlinestock markets have accelerated share prices of few companies and has also
newlinebrought freefall to certain companies. Stock markets have voluminous data
newlineand are subjected to uncertainty. The second contribution aims to build a
newlinepredictio