El Duderino -> RE: Around the NFL (News) - 2013 Season (12/8/2015 12:59:59 PM)
|
quote:
ORIGINAL: JC2015 This is why that 538 is bogusquote:
Panthers unimpressed by ridiculous 'worst 11-0 team ever' label Second, 538 acknowledges an inherent built-in preseason bias to its projection system: Add in the fact that Carolina wasn't expected to be anything special before the season began, and it isn't hard to see why our Elo ratings2 still doubt that the Panthers are even the NFL's best team this season. The preseason rankings definitely got everything right, what with the Cowboys at No. 6, Eagles at No. 7, Ravens at No. 8 and Saints at No. 10. You should absolutely keep considering your preseason opionions as something relevant, computer. Third, the 10th-best team on this list (2009 Saints) actually beat the 7th-best team on this list (2009 Colts). Fourth, the previous "worst 11-0 team" before the Panthers came along? The 1972 Miami Dolphins. Seems logical. http://www.cbssports.com/nfl/eye-on-football/25396353/panthers-unimpressed-by-ridiculous-worst-11-0-team-ever-label Their ranking system is based on Elo. Elo was designed to rank individual chess players. Applying it to a team sport like football is ridiculous for a bunch of reasons: 1) A chess player gets better or worse based on nothing more than himself. A football team has far more moving parts, with some players improving and others getting worse. 2) There is no seasonal variation in chess players, while the composition of a team changes each season, and even within the season. Consequently, you have teams starting the season with their prior season's Elo score. That's essentially an assumption that the team will be the same in September as it was the prior January. When you start with a crap assumption, you're going to get a crap result. 3) Chess is a game of pure skill. There are no funny bounces, no officiating gaffes, no execution issues. A bad chess player isn't ever going to beat a grandmaster on a fluke, but we know flukes happen all the time in football.
|
|
|
|