tag:blogger.com,1999:blog-38600807.post865088581211533518..comments2023-11-05T04:16:44.937-05:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Next Year's WinsUnknownnoreply@blogger.comBlogger8125tag:blogger.com,1999:blog-38600807.post-46984995689716387002007-11-04T11:12:00.000-05:002007-11-04T11:12:00.000-05:00Hi Shephan-1-Penalties do account for a small but ...Hi Shephan-<BR/><BR/>1-Penalties do account for a small but measurable part of winning and losing. One standard deviation increase in penalty yds will cost a team 0.4 wins on average in a season.<BR/><BR/>2-Fumbles are a difficult subject, as are turnovers in general. They are relatively rare events and are hard to predict. But they are even harder to ignore. I'm still trying to find the right balance. <BR/><BR/>See my post here for what I've learned so far: http://www.bbnflstats.com/2007/07/what-makes-teams-win-1.html<BR/><BR/>Part 2 of the article discusses penalties and turnovers.<BR/><BR/>Regarding a regression of 2005 efficiency stats onto 2006 wins, see the post titled "Leading Indicators." If I recall, penalties was one of the things that carried over from year-to-year. Turnovers, however, had a really interesting effect. The more above average a team was one year, the more below average it would be the next. Still not sure how to interpret that.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-58360009378340973412007-11-03T22:51:00.000-04:002007-11-03T22:51:00.000-04:00I found your site and have been digging through th...I found your site and have been digging through the archives. I love the analysis here and have a few thoughts:<BR/><BR/>1 - Does the penalties variable really account for a good deal of the error in the model? You use significant variables such as offensive and defensive rushing and passing yards that it seems unusual to include penalties into the regression.<BR/><BR/>2 - I agree with the inclusion of the turnover variable in the model. Teams with fumbles recovered as opposed to forced fumbles may tend to follow through on a play and never give up - both good signs of a well coached and successful team. I suppose that I have to read your post on fumbles from April... <BR/><BR/>It would be interesting to see a model where 2006s actual wins are included in a regression as the dependent variable with 2005s offensive and defensive efficiency stats.<BR/><BR/>But keep up the great work! I already love the site.Stephanhttps://www.blogger.com/profile/04678119741730362796noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-75217789634929659252007-07-20T11:00:00.000-04:002007-07-20T11:00:00.000-04:00Turnovers most definitely are a repeatable endurin...Turnovers most definitely are a repeatable enduring skill of a team. For example, offensive fumbles correlate year-to-year at 0.31, which is not terribly strong but still considerable and statistically significant. Further, we would expect "within-year" correlations to be even higher. Therefore, their predictive power is <BR/><BR/>I partially agree with you. There is an enormous amount of luck involved when turnovers occur, but that's true of every play in any sport. But part of what the statistical models do is devine what component of the variable is sustainable (the fitted value) and what component is luck (the residual value).<BR/><BR/>One technique I used to reduce the noise of luck in turnover stats is to use forced fumble stats for team defenses instead of fumbles or fumbles recovered (see my posts on fumbles from April). The forced fumble stat is most definitely a repeatable skill, but who recovers fumbles usually is random. Forced fumbles are a better predictor of future fumble recoveries than past fumble recoveries themselves.<BR/><BR/>Additionally, many defensive play calls are designed as interception plays. They are crafted to sucker a QB to throw the ball where a defender is lurking to pick off the pass. Teams with great safeties can allow their cornerbacks to gamble and "short the pass routes" for interceptions.<BR/><BR/>And I think it goes without saying that some QBs, and some offensive schemes, are better than others at preventing interceptions. If it's not a skill, then someone please explain why Peyton Manning has been able to throw 10 or fewer interceptions each of the past 4 years.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-59478619140226408132007-07-20T05:29:00.000-04:002007-07-20T05:29:00.000-04:00You seem to think that "turnover efficiency" is so...You seem to think that "turnover efficiency" is some sustainable attribute of a team. Most turnovers are luck, and thus should not be a large factor in assessing a team's true strength for predictive purposes.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-64273080217364876292007-07-12T15:05:00.000-04:002007-07-12T15:05:00.000-04:00Thanks, however the only trend or pattern I'm prop...Thanks, however the only trend or pattern I'm proposing is that last year's performance is the best starting point for estimating this year's performance.<BR/><BR/>I'm not suggesting teams are on multi-year upswings or downswings.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-37883023511349306462007-07-12T09:13:00.000-04:002007-07-12T09:13:00.000-04:00Great stuff. Not many people out there can fathom ...Great stuff. Not many people out there can fathom the idea that there are meausurable and predictable trends in sports.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-58392413960851151822007-07-01T14:14:00.000-04:002007-07-01T14:14:00.000-04:00Very interesting. Keep up the good workVery interesting. Keep up the good workAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-51719249202035407462007-06-27T15:44:00.000-04:002007-06-27T15:44:00.000-04:00This is great stuff. Found you through FO. Will ...This is great stuff. Found you through FO. Will try to keep checking in periodically.Anonymousnoreply@blogger.com