Forecasting Stock Market Returns and Volatility Using Time Series Analysis
Date
Authors
Major Professor
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Authors
Research Projects
Organizational Units
The Department of Finance seeks to provide knowledge of the descriptive, behavioral, and analytical background of financial management, in preparation for positions in sales management, marketing research, retail, etc.
History
The Department of Finance was formed in 1984 in the College of Business Administration (later College of Business).
Dates of Existence
1984–present
Related Units
Journal Issue
Is Version Of
Versions
Series
The Honors project is potentially the most valuable component of an Honors education. Typically Honors students choose to do their projects in their area of study, but some will pick a topic of interest unrelated to their major.
The Honors Program requires that the project be presented at a poster presentation event. Poster presentations are held each semester. Most students present during their senior year, but may do so earlier if their honors project has been completed.
This site presents project descriptions and selected posters for Honors projects completed since the Fall 2015 semester.
Department
Abstract
Stock price forecasting is a popular and important topic in financial and academic studies, and time series analysis is an advanced set of tools that is useful for this type of data. This study analyzes a time series of S&P 500 index returns data; and employs time series methods and ARMA modelling to predict S&P 500 index mean returns. Index return volatility is also analyzed and forecasted using GARCH modelling techniques. Multiple competing models are fitted and tested using Eviews software. Weekly observations of S&P 500 index returns from 2010 to 2016 are used to fit several conditional variance models, and the most preferred model is selected in order to forecast variance. The two most preferred conditional mean and conditional variance models are used to forecast values for the next ten weekly observations, and the forecasted values are compared with the actual data. The conditional variance results are of particular usefulness in risk modelling applications.