Intraday Stock Trend Analysis Based on Social Sentiments and Prediction of Entry and Stoploss

Abstract

Machine Learning and Data Analytics are branches of modern technology that are fast being used in conjunction to solve multitudes of problems that inflict us in our daily life. A particular field that could benefit immensely from the capabilities of data mining and machine learning is the issue of unpredictability in the stock market environment, especially in accordance with Intraday Trading. Intraday Trading or Day Trading, as it is commonly known, involves the purchase and sale of stocks on the same day. Intraday Trading gives investors a daily profit when the investment is done intelligently. In recent times, social media has proven to be a beneficial tool in representing the emotions of the public regarding current happenings. Social media sentiment analysis is a fast growing and explorative field that is rapidly gaining news fields of applications. It involves identification and categorization of opinions implied in a text to determine if the emotion expressed is positive, negative or neutral. The system proposed, classifies user sentiments for a particular stock across various social media platforms such as Twitter, online news websites and blogs etc. Using several classification algorithms in combination with Natural Language Processing, the system strives to find a correlation between user emotions and stock performances. Also, using Logistic Regression Technique, a model is developed that takes as input user sentiments for the current day as well as historical stock data in order to classify whether the market would open as a positive one or negative. In addition to this approach, a functional logic that would assist in determining the entry, stop loss and target levels of a stock with an assured profit of a minimum of 0.5% using a plethora of technical indicators, primarily the usage of 15 minute candle charts is proposed. Also, the performance of the model is monitored to test the accuracy of the prediction and validity of the logic.