9th International Conference on ICT for Sustainable Development

Authors - Ruthvik Reddy Gaddam, Rutvik Nelluri, Dandu Samir Varma, Sagar Basavaraju
Abstract - The movement of a stock can be known by following the news about the company from various sources. One of the largest sources of this is social media, where many discuss various stocks and how the stock price may change based on their analysis. Using these sources of information one can try to classify the movement of a stock as bullish or bearish in nature. Data consisting of tweets from StockTwits mainly about Indian stocks is selected and gathered. The tweets are mainly taken from the IT and Banking sectors as they are volatile even for a slight change in the market and more people tend to buy these stocks. Most recent research in the analysis of stock trends is centered around sentiment analysis. By using natural language processing techniques and sentiment analysis models, data from Stocktwits can be processed and used to train machine learning models to classify them and help traders make more robust decisions. From analysis, it is found that the Artificial Neural Network has performed better with an accuracy of nearly 80% which is backed up by classification done on newly gathered data from Stocktwits where 8 out of the 10 tweets were correctly classified. The Random Forest model has been trained with an accuracy of 70%. The results obtained from this analysis are found to be on par with other research work in this domain.