Document Type

Thesis

Date of Award

5-31-2021

Degree Name

Master of Science in Computer Science - (M.S.)

Department

Computer Science

First Advisor

Guiling Wang

Second Advisor

Ioannis Koutis

Third Advisor

Alberto Martin-Utrera

Abstract

Stock market prediction has attracted not only business but academia as well. It is a research topic, to which many computational methods have been proposed, but desirable and reliable performance is yet to be attained. This study proposes a new method for stock market prediction, which adopts the Gated Recurrent Unit a deep neural network and incorporates investor sentiment to improve its forecasting performance. By extracting investor sentiment from news headlines using VADER sentiment, this paper makes it possible to analyze the irrational component of stock price. Our empirical study on DJIA index proves that our prediction method provides 6% better prediction compared to baseline models. Furthermore, our empirical study helps to better understand investor sentiment and stock behaviors. Finally, this work shows the potential of deep learning in forecasting a financial time series in the presence of strong noises.

Share

COinS