Investigating the Effect of Sentiment in Daily News on Stock Price Volatility using Deep Learning Models

In the contemporary dynamic financial environment, volatility in the stock price is a crucial issue to predict and understand in the portfolio of investors, traders, and policymakers. Volatility has various financial applications and the prediction of the volatility of financial assets can be beneficial. A widely used indicator of risk in the market is the Stock sector volatility. This article introduces a deep learning solution, which explores the sentiment of daily news as a driver of stock price volatility through GRU model. GRU model is utilized because it is more efficient and it is more effective in temporal relations learning of sequential data. The model is performing good since in the experiment, the experimental data produced a high digit R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.9924 and low values of RMSE, MSE, and MAPE that deduce that the model is of good predictive abilities. A further comparative study on such RNN and DAE-LSTM models can further determine the superiority of GRU to reflect an appropriate relationship in the existence of sentiment indicator and volatility in the market. The results suggest the future of the synergistic algorithm of DL and sentiment analysis in systematic and live monetary forecasting.