Technical Analysis: the prediction of stock prices using patterns in historical stock charts is nothing short of art. Quantitative analysts spend most of their time pouring over patterns to predict the future direction and price, to maximize profits or mitigate risk. Some analysts use these signals along with Fundamental Analysis and market news, to make trading decisions for short and long term trading strategies. Algorithm Trading is the process of using computerized mathematical models, based on certain patterns, rules and factors. With the exponential growth in Algorithm trading, currently capturing ~80% of market volume and the growth in computing power, sophisticated mathematical models is the order of the day. These strategies have grown over the period from simple Excel rule-based programs to Artificial Intelligence using deep neural networks.

Market news pertaining to the stock or macro economic factors, while having an impact on the stock price and important for trading decisions was always elusive to factor in quickly, due to it being voluminous and its inherent subjectivity in nature. Natural Language Processing (NLP) of news can provide a powerful signal that can augment trading decisions. This paper aims to explore the sensitivity intensity of relevant historical news through VADER NLP and use it as an additional signal to assess the excess predictive power over and above historical time-series stock price prediction, using neural network algorithms. This model is developed using Python and Tensorflow Keras neural network libraries.

KEYWORDS: Natural Language Processing, Sentiment Analysis, Algorithm Trading, AI/ML, Stock Trading.