|
Activity Number:
|
447
- Essential Uses of Statistics in Football
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 2, 2017 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Sports
|
|
Abstract #322234
|
View Presentation
|
|
Title:
|
A Statistical Analysis of the NFL Draft: Valuing Draft Picks and Predicting Future Player Success
|
|
Author(s):
|
Robert Nicholas Citrone* and Sam Ventura
|
|
Companies:
|
Carnegie Mellon University and Carnegie Mellon University
|
|
Keywords:
|
Football ;
NFL Draft ;
Quarterback ;
Regression
|
|
Abstract:
|
The objective of our research is to identify trends and make predictions NFL teams can use to improve draft pick success. Our research includes two sub-problems with differing methodology: assessing positional value by draft round and predicting individual player success from College and NFL Combine statistics. To assess positional value we use local nonparametric regression to model the approximate value (AV) of each pick conditional on position. Our data set contains information on all draft picks between 1999 and 2013. Our results show significant differences in relative positional value throughout the draft, quarterbacks are the most valuable first-round selections while defensive ends and linemen offer the best relative value late in the draft. The second part of our research, predicting individual players success using college football & NFL combine data, utilizes log-transformed linear regression with AV per season as the response variable. The model has significant predictive power, and crowned Carson Wentz the top quarterback prospect in the 2016 NFL Draft. Height and wonderlic score were two of the most useful predictors of NFL success.
|
Authors who are presenting talks have a * after their name.
