STOCK PREDICTION AND SIMULATION OF TRADE USING SUPPORT VECTOR REGRESSION

Jay Borade, Ayush Khanvilkar, Rahul Kanvinde, Akshay Narkhede, Anuj Odedra

Abstract: Up’s and Down’s in share market are always unpredictable. Commercial banks offer their customers market predictions based on the sentiment and market news of the given day. These predictions are relevant for a short period of time. Commercial banks cater to a large demography hence they have to limit their prediction services. Investment banks have used predictive models which use past market data to predict stock prices and market indexes. Common people cannot afford the services provided by the investment bank. Candlestick is widely used in the trading community for analysis. But candlestick chart looks different for various time frames and they make it difficult to manage risks. The proposed system aims at helping traders make sound financial decisions. It simulates trading thereby helping new users understand the application.It will be of assistance to beginners so that they can learn how to trade without losing any capital. The system uses machine learning techniques and also lets the user view sentiment about the stock in real time. Both mathematical predictions and sentiments are used as parameters for making a financial decision. The proposed system is able to achieve prediction accuracy of up to 95%[1]

Keywords: Machine Learning, scikit-learn, share market, stock prediction, support vector regression

DOI: https://doi.org/10.15623/ijret.2018.0704009