Analysis of Daily Illini College Football Predictions: Accuracy of Forecasters and Betting Line

Zhou, Yayu

Permalink

https://hdl.handle.net/2142/90257 Copy

Description

Title

Analysis of Daily Illini College Football Predictions: Accuracy of Forecasters and Betting Line

Author(s)

Zhou, Yayu

Contributor(s)

Hildebrand, A.J.

Issue Date

2016

Keyword(s)

• probabilities
• betting line
• college football
• prediction
• accuracy

Abstract

Each week during college football season, the Daily Illini (DI), the college newspaper at the University of Illinois, publishes predictions by sports columnists on selected college football games for the upcoming weekend. How accurate are these predictions when compared to the betting line (i.e., always predicting the favorite team to win), or to random predictions (i.e., predicting by a coin toss)? Are there significant differences in accuracies between games involving Illinois (or Big Ten teams) and games not involving Illinois? Are there significant differences in accuracies between individual forecasters? To answer these questions, we collected and analyzed Daily Illini college football predictions over the 6 year period 2008--2013. This data set consists of 1740 individual predictions made by 29 forecasters and covering 346 games. Our main conclusions are: (1) The accuracy of DI forecasters during this period was significantly lower than that of the betting line. (2) The accuracy of DI forecasters was significantly higher than that of random predictions. (3) The DI forecasters were significantly more accurate in predicting Illinois (or Big Ten) games than in predicting games not involving Illinois (or Big Ten) teams.

Type of Resource

text

Language

en

Permalink

http://hdl.handle.net/2142/90257

Copyright and License Information

Copyright 2016 Yaya Zhou

Owning Collections