Computer Successfully Predicts Football Playoffs in 7 Out of 8 Games

Share Article

Coolstandings.com, a web site that projects playoff scenarios using simulation, has successfully predicted the outcome of 7 out of 8 football playoff games, tops among a panel of 22 experts.

First baseball, now football. The computer model at http://www.coolstandings.com has correctly predicted the outcome in 7 of the first 8 pro-football playoff games, including 4 upsets, after successfully predicting the outcome in 5 of 7 baseball playoff series.

After two rounds of the football playoffs, their simulation is tops among 22 experts from ESPN, Sports Illustrated, CBS SportsLine, Fox Sports, the Sporting News, and Football.com. Although Joe Theismann and Merril Hoge of ESPN matched the number of correct picks as the simulation, only coolstandings.com correctly predicted the final 4 teams remaining in the playoffs.

Co-founders Greg Agami and Sean Walsh launched coolstandings.com in August of 2005, projecting playoff chances for all major league baseball teams. Their simulation builds upon principles from the field of sabermetrics, or the objective analysis of baseball using statistics, which was pioneered by Bill James. Following the success of their baseball projections in both the regular season and playoffs, Agami and Walsh expanded the site to model both football and basketball. Daryl Morey and other statisticians had already extended some of the sabermetric principles to these sports.

"Modeling football is a lot harder than modeling baseball, since there are only 16 games in a season," explains Agami. For the regular season, the computer simulation uses simple team statistics such as points scored and points allowed to predict individual game results, then simulates the rest of the season millions of times to determine the probability that each team has of making the playoffs. Game location, recent performance, and strength of remaining schedule are also taken into account.

The projections during the regular season were remarkably good, as the site correctly predicted playoff spots for all 12 playoff teams 8 weeks before they clinched a spot, on average. The only team labeled a "playoff lock" that failed to make the playoffs was Philadelphia, given such a standing at the beginning of the season. However, this label was rescinded even prior to injuries and chemistry issues proving the downfall of that team.

To predict the playoffs, the simulation uses the same team statistics but can calculate the exact probabilities of a given team winning each round of the playoffs, due to the limited number of playoff games. So far the simulation has picked 4 upsets correctly, including Pittsburgh over Indianapolis, a game where Indianapolis was an overwhelming 10 point favorite and picked by most experts.

The one game that coolstandings.com failed to forecast was the New England win over Jacksonville in the first round. The simulation had Jacksonville slightly favored, while most experts picked New England to win. Agami believes the simulation put too much emphasis on a meaningless New England loss in the last game of the season, and says the site plans to optimize its algorithm to improve its accuracy.

"We can't complain much when we're 7-1, though!" says Walsh.

So if you want to know whether or not to make travel plans to Detroit for the championship game this year, you may want to check this site out - it's a fun way of following your team. To see the latest predictions and find out more about how the simulation works, visit http://www.coolstandings.com.

About coolstandings.com

Coolstandings.com, Inc. projects the playoff picture for baseball, football, and basketball, calculating the chance of each team making the playoffs - updated daily - by simulating the remainder of the season millions of times. The site has been featured by ESPN the Magazine for its baseball projections.

For more information, please visit http://www.coolstandings.com.

Contact:

Greg Agami

coolstandings.com, Inc.

866-829-4553

http://www.coolstandings.com

# # #

Share article on social media or email:

View article via:

Pdf Print

Contact Author

Greg Agami
Visit website