Openers and Bye Weeks: Which Coaches Do the Most
Football commentators regularly talk up the value of the bye week or a big early season game for the opportunity to add extra preparation from a coach staff. This week I dug into the data to see how much of an effect bye weeks and openers had on team performance and which coaches are the best and worst at using the bonus time to their advantage.
Methodology
As usual, I looked at all FBS games from 2003-2012. If a team played an FCS opponent as an opener or after a bye week it wasn’t included but it wasn’t treated as a bye for the next week’s game, either. I compared how each teams EV+ (points better/worse than an average team would have done, opponent adjusted) was in openers and post-bye versus how they did overall for that season. I then assigned those numbers to the head coach and looked at how head coaches have done, under the assumption that any strengths or weaknesses under these conditions would be more coach than program. So Brady Hoke is evaluated from Ball State, San Diego State and Michigan.
General Findings
Over nearly 1500 post bye week games evaluated, a small benefit did emerge. The average team performs 1 point better post bye week than in regular weeks. 53% of teams performed better than their expected based on full season performance. The data closely matches a normally distributed outcome with an average benefit of 1 point and a standard deviation of 11.5 points.
Distribution of points versus average for post bye week games
Openers were about a wash. The typical team performs about 0.2 points worse than expected in openers. Openers feature a lot more variables than just extra preparation time. The standard deviation for opening games is the highest of any week during the regular season (but lower than bowl games). That variance is pretty low however. Teams have the most deviation from their season average in week one (11.9 points) but the low point has a deviation within 1 point (11.0) that occurs during week one. So teams are most likely to have an outlier game in week one or for their bowl but overall, most weeks have a pretty similar level of deviation.
B1G Coaches
To see how current Big Ten coaches have done, I looked at their track records for both openers and after a bye week to see who has done the most and least in each situation. The bubbles are color coded based on the team and all of the reds are team coded because there are too many red teams in the Big Ten.
Positive numbers are good and bubble size indicates sample size
Mark Dantonio and Kirk Ferentz have both been able to start the year off strong with strong opening performances. New Wisconsin coach Gary Andersen along with small sample size guys Bill O’Brien and Curtis Kyle Flood both have the best results after a bye week. Coach Hoke’s openers have been mildly below average but his bye weeks have been the most productive of any coach with a larger number of games. Urban Meyer has seen his results after bye weeks on the other end with his squads playing 3.6 points per game worse than they do in a normal week.
Other Notable Coaches
Openers on the x-axis and post bye week on the y-axis
Charlie Weis has seen his career reflect his seasons at Notre Dame. A first season/game that was significantly better than what happens afterwards. I guess his decided schematic advantage expires after one week. Barry Alvarez is apparently the king of the bye week as his teams turned 3 bye weeks under him into a +21 advantage, even after accounting for opponent and team strength. Joe Paterno was the opposite case. His Penn St teams played over two touchdowns worse after a bye week. Mack Brown and Jim Tressel also had teams that have found bye weeks to be counter-productive. The only entry several points worse on both standards was GERG Robinson during his tenure at Syracuse. LLoyd Carr’s openers were never great, even when The Horror is excluded but his teams where some of the best coming off of a bye week.
As always, let me know about any request for off-season material you would like to see.
How much is Hoke's opening game number impacted by the Alabama game? Not that I think it should be excluded, just curious.
it is at least a third of his score for michigan and overall i think he has coached 8 years...
so I would guess that 12.5% of the number for the opening game would be made up of the Alabama massacre. lol....
2003 W vs. Indiana State (FCS)
2004 L vs. BC
2005 L @ Iowa
2006 W vs. EMU
2007 L vs. Miami (OH)
2008 W vs. Northeastern (FCS)
2009 L @ UCLA
2010 W vs. Nicholls State (FCS)
2011 W vs. WMU
2012 L @ Alabama (Neutral)
Just taking into account Division 1A / FBS teams, Hoke went 2-5 in his openers. Clearly, his teams were overmatched against power conference teams (BC, Iowa, UCLA) when he was at Ball State and SDSU.
trying to figure out what the hell team "Mar" was...hope they're worth the money
I appreciate your work Mathelete...
I think the problem inherent to opening week and after the bye week is that college football teams vary so much from one to another. Especially compared to something like the NFL; we all know not to go against Andy Reid the week after a bye but I dont' think anyone actually thinks Alvarez is worth 21 points the week after a bye either...
jdon
I agree. Small sample sizes and extreme variability of opponent quality are probably creating the vast majority of the differences observed.
to conclude that "Alvarez is worth 21 points the week after a bye." That is one possible conclusion, but not the only one, right? It could be that Alvarez is a good coach with the bye and a bad coach without the bye, since what the Mathlete is measuring is the difference between bye and non-bye games. That is, the fact that Alvarez's teams did better after a bye compared to the expected performance doesn't mean that Alvarez is a great coach after bye weeks, just that he (or his team) is better (or perhaps less bad) after a bye week.
well, yeah. of course...
data just is. there isn't a good, there isn't a bad. I wouldn't even call this trend data because of the variability in team strength at the college level...
These graphs conclusively prove that Rich Rodriguez is a better coach than Lloyd Carr or Brady Hoke. Don't let wins and losses deceive you -- this is math, people!
Last graph shows that mediocre coaches prepare better for opneing games because they need to get every possible win?
Then good coaches tend to overlook opening games (what me worry?) that leads to things like the horror?
well played brother, well played...
If only EVERY game were an opener or after a bye week!
Nick Saban is frightening with a few weeks of extra preparation. We all saw that last year. This year, he gets a bye before he goes to Texas A&M. Johnny Manziel better WATCH OUT because Saban is likely to take their offense apart, piece by piece. It's going to be ugly.
but the graph sure doesn't reflect that -- he's sub-par both in openers and after byes. Maybe because his "par" starts out higher than most?
Yeah, given his reputation, the data on Saban was really surprising to me. Then again, Alabama does generally schedule pretty good openers and the bye is more likely to be against an SEC openent than a non-conference foe. My guess is that the drop off comes from the beatdowns Alabama puts on its other, shitty nonconference foes, and the shitty ones in the SEC (Tennessee, Auburn, Kentucky, etc). The post-bye weeks and season openers can't catch up to the pace set in the games against bad teams.
that he's also coached at MSU and LSU. Neither of which were great programs when he took over at those schools. Also, when Saban took over Alabama the Crimson Tide wasn't that good either. They went 7-5 in his first year.
left out for readability's sake, but do you have any error analysis for these stats?
^This guy*
*Hoping this was accidental
I was thinking the exact same thing. I'm guessing the thought process went something like this......."Hmmm.....the absolutely free analysis that this MATHLETE charachter has provided is not to my liking. I prefer my free analysis done on others free time at no cost to me to be prepared better and demonstrate statistical consistency more in line with my preferences. Next time, perhaps, this Mathlete will take a little more time and prepare his free work in a format I like better."
not paying for something doesn't equate an inability to criticise... I think this is a point lost on many in the mgoblogsphere
love,
jdon
I'm sorry but yes it does. Human decency dictates that when somebody does a LOT of work as the Mathlete obviously has you do NOT criticize his work unless you've got something better to offer or you think his conclusions could potentially harm someone.
In real life nobody would ever walk up to a co-worker, friend or family member who had expended a ton of energy and time on a project and say "that sucks" and walk away. That's the problem with the blogoshpere in general - it's considered ok to be an asshole and its really not. Either here or in reality.
Love,
MgrowOld
Wait, not paying someone for something robs you of the right to critique it? Am I not allowed to critique David Brooks' column in the New York Times?
and downvoting something, which is purely negative. I think that was mGrowOld's point.
So it's not OK to use the voting system that the website put on the very post?
Downvote away. I thought it was weird that the post was downvoted but I don't think it's a big deal either.
You can do anything you want. My point is a downvote, absent of anything else, is the electronic equivilent of yelling "you suck" to someone and I cannot possibly see how or why that would apply here. And again, IMO, in real life if the Mathlete or anyone else showed someone the results of a lot of work they did pro-bono I simply dont believe anyone but a complete asshole would look at it and say "you suck" and leave.
But hey, maybe I'm in the minority here being old. I just fail to see the value in being mean to other people with little or no purpose behind it other than being mean. You obviously see it differently.
1st off, let me make sure the record is straight and I think the mathelete does good work...
That said, if your point is that you can't criticise something that is free then that makes no sense at all. It isn't about common decency. When did it ever happen that criticism equals a lack of appreciation?
What you are saying is so illogical I don't even know how to respond to it except to repeat what I said which basically is that we all have the right to question or otherwise criticise (respectfully, no cuss words, not blaming, etc.) any analytical data that others provide. Hell, we should look at it skeptically if we really want to converse on an intelligent level...
Stop being so emotionally attached to the name on the top of the post and start looking at the content dude.
love,
jdon
Did you remove instances when the opponent is also coming off of a bye week?
How fitting that his post-bye numbers are the worst after negotiating all those byes before joining the Big Ten.
"Teams have the most deviation from their season average in week one (11.9 points) but the low point has a deviation within 1 point (11.0) that occurs during week one."
the low point of what? and i just have no idea what "that occurs during week one" is modifying. it's been a lot of years since waltzing through stat 350.
Comments