Support MGoBlog: buy stuff at Amazon
This is my first diary, so I apologize in advance if my images do not embed properly, I'm still figuring this out......
Setup
There has been a lot of discussion regarding Michigan’s strength and conditioning program, with many drawing the conclusion that the superior conditioning of Michigan’s football team will result in wearing down teams through three quarters, and dominating a tired opponent in the fourth for a win. I did a little searching in an attempt to find out if there are any historical statistics from teams coached by Rodriguez and conditioned by Barwis to support this belief. Mike Barwis has been the head strength and conditioning coach for Rich Rod since 2003 at West Virginia, so I did the analysis on all of their games since 2003. Since there is no statistic to measure overall conditioning I used points scored or allowed as an indicator, since what I really want to know is whether conditioning or lack thereof translates to points (and wins). I know there are multiple problems with this, e.g. last year’s inability to execute the offense doesn’t necessarily have anything to do with effort or conditioning, but I hope that by looking over several years of data, I will be able to draw some solid conclusions.
Data
I looked at the quarter by quarter scoring for all of the games from 2003 to present. First I’ll give you the average quarter by quarter scoring for offense and points allowed for defense (including WVU’s 2007 bowl game which wasn’t actually coached by Rodriguez).
|
Q1 |
Q2 |
Q3 |
Q4* |
|
|
Offense |
8.551282 |
8.205128 |
7.371795 |
7.910256 |
|
Defense |
3.987179 |
6.628205 |
4.320513 |
6.589744 |
Not sure what the proper test for determining statistical significance is here, but the eyeball test says you probably shouldn’t count on the offense to score more points in the 4th versus quarters 1-3. As for the defense, you can probably expect them to give up slightly more points in the 4th quarter, at least as compared to quarters one and three (caveat: defensive data compiled under tenure of multiple defensive coordinators while assuming Rich Rod was in control of the offense in each year examined). This obviously doesn’t give the full picture, as play calling, personnel, etc. will be dictated by the circumstances of each game. So, to crunch the numbers further, I looked at the 46 games in which the teams were within 16 points of each other after 3 periods. I realize this cutoff is somewhat arbitrary, but it seems reasonable to me that if a team is up or down more than two touchdowns and two 2 point conversions with 15 minutes to go, play calling and strategy may be start to change, which would confound the results of this analysis. Nonetheless, in these situations Rich Rod is 26-20 and on average outscores the opponent by ~1.2 points in the final quarter. I know this crowd is big on charts, but a graph is a bit more useful for this particular analysis.
<a target='_blank' title='ImageShack - Image And Video Hosting' href='http://img193.imageshack.us/i/q3scorevsfinalscore.tif/'><img src='http://img193.imageshack.us/img193/8277/q3scorevsfinalscore.tif' border='0'/></a>
Here is a link in case that didn't work: http://yfrog.com/5dq3scorevsfinalscoret
Again, this is only games that were within sixteen points after 3 quarters. The abscissa is the point differential after 3 quarters ((WVU or M) – opponent), thus negative numbers mean Rodriguez’s team was trailing and positive numbers mean they were winning. The ordinate is point differential of the final score ((WVU or M) – opponent), so a negative value means a loss and a positive number means a win. Based on this, it looks like the third quarter score is a pretty good predictor of whether the team will win or lose the game. I am surprised by how well this holds up even for small 3rd quarter point differentials (between -7 and 7). I know it is a relatively small sample size and all that, but still. One encouraging point of note is that when leading, even by a small margin (<7), Rodriguez’s team usually comes out on top. Look at all the data points with a 3rd quarter lead of 1-5 points. He lost one game. That makes me smile inside. On a side note, if I include all of the games, not just the ones within 16 points after 3 quarters, the results are basically identical. However there was one pretty big comeback in the WVU days against Louisville. Down 17 points after 3 quarters, WVU rallied to win by 2 in overtime.
*Overtime points were included as 4th quarter points.
Conclusions
The idea that Michigan will completely own teams in the 4th quarter is probably not true. If the historical data is a valid indicator of what we can expect, in terms of points scored and allowed Michigan will perform in the final quarter pretty much the same as through the other three. Actually, the biggest differential between points scored and allowed is in the first quarter. I am not sure why there is a noticeable jump in points allowed in quarters 2 and 4. My first guess was that it was due to increased risky play calling during the 2 minute drill near the ends of each half. However, if this were true, I would also expect to see an increase in offensive points scored.
The score at the end of the 3rd quarter is a pretty solid indicator of game outcome. If Michigan is up at the end of 3, even by a little, we will likely come away with a win. Unfortunately the reverse also holds true, with Rodriguez only having two comeback wins (one of which was the 2008 Wisconsin game).
Wrap-Up
I think Michigan has a bright future under Coach Rodriguez. I think we can expect some exciting football and a lot of wins. I’m just not convinced said wins will come courtesy of the superior conditioning theory. In the end, as long as Michigan is piling up the W’s I don’t think any of us care in which quarter they score the points.
Bonus Tidbit
While doing research I came across this: When Morgan Trent was asked "What's the craziest thing about Wolverines strength coach Mike Barwis?," Trent replied "That he's a former MMA fighter and can kill any one of us if he wanted to with one punch."
Which got me to thinking, who would win if Mike Barwis and Chuck Norris got in a fight? I thought about it for a while and my head started to hurt. So I decided that some sort of black hole, kung-fu vortex would probably open up and suck us all in. We would never know who actually won; therefore it is an unanswerable question.
thanks for the research... conditioning does not matter if Nick Sheridan and Threet are at QB...otherwise, conditioning does have an impact for sure.
3rd quarters are the most important in the ball game, IMO... for a lot of reasons. 4th quarter stats can be skewed, especially if the game is easily at hand or if there are lots of late drives with a flurry of offense.
There have been quite enough criticisms of Steven Threet and Nick Sheridan.
I can't see why people still retreat to that card time and time again.
Two guys who found themselves thrust into a position in a system they were not ready for, and fucking assholes still rip them a year later when things are going much better.
Neg Bang for anyone who criticizes Threet and/or Sheridan for the rest of eternity.
First, how can you measure conditioning? You make this point yourself. As a result, your entire analysis assumes RR/Barwis teams are better conditioned than all of their opponents. Second, in many games, the team that is ahead is more interested in killing the clock than scoring points. It seems almost certain that better conditioned teams perform better late in games, but I do not know how to prove it.
"As a result, your entire analysis assumes RR/Barwis teams are better conditioned than all of their opponents."
Yes. That would be what he is testing. Here the null hypothesis would be the assumption that RR/Barwis teams are better conditioned, which should be shown by a point differential in the fourth quarter.
Of course, there's no way you can isolate point differential by quarter just on conditioning. In order to do so you would need some control case in the data set at the very least so you could see if the conditioning made any difference.
As people have said, this is just about impossible to get at statistically (much as I love regressions). HOWEVAH, a thorough enough analysis of game tapes with proper charting should provide a data set on which you could see a statistically significant difference. Even charting something like "fatigue incidents" could help if accurate. Charting these for both teams could test our team against our opponents (because it doesn't matter if, for instance, the Mountain West is better conditioned than Michigan). So the idea would be that if we have less "fatigue incidents" than our opposition, then we are better conditioned.
BUT WAIT you say, what if they have more depth and rotate more. Then just divide by snaps the player has had prior to the fatigue incident. Whatever. It's probably possible and almost certainly not worth the time for the average fan. Let's make Brian do it.
"Yes. That would be what he is testing."
Technically, what the OP's analysis is testing is whether RR/Barwis teams perform better in the fourth quarter, NOT whether better conditioned teams perform better in the fourth quarter. To interpret the OP's results to mean the latter, you must first assume that RR/Barwis teams are always better conditioned than their opponents. It is nice assumption, but it is impossible to back up with facts.
It must be noted that RR did this with players who frankly aren't as good as the players to which he now has access. The flipside of tomhagan's observation that it doesn't matter if Threet-a-Death are your QB's is that great talent CAN own the fourth quarter with superior conditioning.
Knowing you're a bad*&% always helps in a physical contest. Belief is half the battle. Conversely, if the team is half gassed and in a struggle, desperation and forboding sets in, and that makes conditions favorable for the shoe to fall off the other foot.
is yes. If two teams are equally talented, but one team is fresh, and the other is beaten down and tired, the well-conditioned team is going to play better than the tired, beaten down team. If you're tired, you cannot execute. If I can beat you to a spot, I'll block you or run over you.
Statistical analysis is great, but it doesn't beat common sense. Think to the last drive against Notre Dame - those defenders were tired. Our guys were fresh. Look who was getting open - Laterryal Savoy, Greg Mathews, guys not known for their speed. But, we do know they're in great shape.
This is good research, but this is probably too complex of a problem to figure out.
Factors that can skew the results (you point out some above):
-score of the game (if ahead try to run out clock, if behind try to score quickly)
-conditioning of opponent (might be close to Barwis team)
-injuries (if stud QB is out, you'll score less, if stud LB or DB is out, they'll score more)
-fluke plays can score regardless of conditioning
-mental errors
I think Chuck Noris would win, simply because his spinning wheel kick is so fast that it breaks the space time continuum, making time travel possible, in which case he could kill Brawis's dad before conceiving him.
Neither Barwis nor Chuck Norris were concieved. They have always existed. In fact the Big Bang is a result of their simultaneous punches colliding. They have come to our world only to bide their time before the forces of this universe collapse upon itself, and their battle can rage anew.
I mean Hulk has survived in the comic books through nuclear annihilation, so surely if he establishes himself as a mutant immortal, then he could survive a Chuck Norris roundhouse kick. I mean Barwis has already started experimenting with injecting players with his radioactive blood to create an entire subspecies of genetically superior trained killing machines (see: Mike Martin)
Plus you'd have to subscribe to the "I can kill your father" theory of time travel (aka the "Doc Brown theory"), as opposed to my belief in what i'll call the "Sliders" theorem which proposes that each key decision point in history spawns multiple threads of existence such that we are now living in but one of nth dimensions.
Wow, i just fit Chuck Norris, Hulk, Back to the Future and a Jerry Oconnell reference in the same post. Heavy mental lifting. I need a beer.
Blowout games typically have much lower fourth quarter scoring for the winning team, because they will put their subs in for the final quarter of play to give them some playing time and reduce the threat of injury. This generally doesn't apply to the losing team until later in the game, because the chance exists that the losing team will still work their way into contention if they get a couple quick scores.
I would suggest two modifications to the data.
1) Throw out the entire 2008 season since Barwis and RR hadn't had time to completely make their mark on this M team in terms of conditioning.
2) Record results only for games that RR's team was less than two TDs ahead going into the fourth. This, of course, will include those in which his team was behind.
Let's see what that does to our results.
In the nd game, I thought we had the game won going into the 4th quarter with an 11 point lead. In fact, I was yelling "It's Barwis Time" as the 4th began. There was no way nd would score a point a minute in the 4th.
I was stunned when nd scored 2 TDs (yes, one was after an INT). Guess we will just have to wait and see what happens the rest of the season.
has a statistical link to strength and conditioning that I've seen here. The INT was due to Mathews' bad read. If he was tired there, he made up for it 5 minutes later.
I was often thinking that Michigan was the team on the "being worn down" side of things. I think that the addition of Barwis and team, the conditioning of the football team is not on equal or better than other teams. Before I was beginning to think Michigan had issues and was usually not in as good shape as the opposing teams. Very hard to measure this, other than by the 7 P rule: proper prior preparation prevents Pi$$ poor performance.
Chuck Norris wins as he is like the Ace of your trump suit, Barwis while good is only the King of the trump suit. Sorry.
I think we should compare a conditioned team to one that was on the Michael Phelps diet without all the exercise. I think you need to compare the same period for Carr/Gittleson and show us the difference.
His teams were comprised of players that may not have been as athletic as the opponents they were facing. Strength and conditioning in those programs was a necessary equalizing force and probably a major factor in the success that he had against teams with greater aggregate talent.
I feel like this was debunked by the WLA, but I can't find the link.
West Virginia was not at a talent disadvantage compared to the rest of the Big East.
I would agree with previous comments that it is a pretty focused data set with so many factors not accounted for that it is difficult to accurately focus on conditioning as the sole influence for the point distribution. There is a lot of variance on a game-by-game, opponent-vs.-opponent basis.
It might be more interesting if there were something to relate thye data set to, i.e. compare these results to the same period for:
-Carr/Gittleson teams (which would be an interesting result from the play-calling, philosophy side of things also)
-the same numbers for college football teams as a whole to see how it deviates from a norm
-compare to other programs known/thought to have superior conditioning regimens, to see if it is a trend or an anomaly
Well-written diary. I had been wondering about this in recent days, and it seemed the team had superior energy in the 4th quarter, at least compared to their opponents. This dampens my enthusiasm on that front.
Your methodology has problems. First, there are too many other variables than strength and conditioning that go into fourth quarter scoring. For example, if WVU was winning by 28 pts, why would they put the same effort in the 4th quarter as they did in the 1st? Second, conditioning doesn't mean you score more in the 4th quarter than in the 1st, but that you score more in the 4th than you would have without the conditioning.
The best way to test this would be to compare WVU's 4th quarter scoring under RichRod and Barwis against WVU under its previous coach. Or, as an alternative, comparing WVU's 4th quarter scoring against conference opponents with similar schedules.
How many football players look like marathon runners? None. How many football players look like sprinters? Quite a few. Different type of conditioning dealing with different types of muscles.
The statistical analysis appears to show the complete opposite of what we've generally expected from a Mike Barwis conditioned team, the margin between points scored and points given up is greatest when the teams are the freshest (the 1st and 3rd quarters).
Perhaps the Barwis Regimen doesn't develop the stamina and endurance to outlast opponents but an explosive speed to strike hard and fast and first. Perhaps--despite our expectations to the contrary--this explosive speed advantage actually deteriorates as the game wears on. The Barwis/Rodriguez Advantage is greatest in the first quarter but appears to taper off in the second. It rebounds somewhat after halftime and then tapers off even more in the fourth quarter.
Personally, I've run with a few body builder/sprinter types and they generally die at about one mile. They can blow me away in a 100 yard dash, however. Our bodies are simply conditioned differently.
"to strike hard fast and first"
Agreed: I think maybe we're looking at this backwards. The Rodriguez offensive philosophy seems to be score quickly, early, and often. And as a football strategy, it probably makes more sense to put a lot of points on the board early than to be able to grind out a close game late: in the case of a shootout, at least you're not playing from (too far) behind, and in the case of a non-shootout, you put your opponent behind early and force his offense to play catchup. Playing catchup, in turn, makes the opponent more one-dimensional, which makes playing defense easier and gives you a better chance of maintaining your lead. While treating the game like a sprint rather than a marathon may be a disadvantage in terms of 4th quarter conditioning, this can be partially mitigated by rotating a lot of players (I recall hearing RRod wants to play 22 players each on O and D each game). And if the offense is designed to be explosive on a play-by-play basis, then even in a game that stays close late in the 4th quarter, the offense still has the potential to put up points quickly.
So in conclusion, your analysis does support an interesting interpretation, if not the one you initially sought to support.
While these numbers have issues as mentioned by smarter people than me, I'd be interested to know what OTHER teams numbers look like. What did Carr era teams scoring look like? Bo era? What about Tressel? Meyer?
I'm not sure that they have any significance in terms of conditioning, I simply find them to be interesting numbers.
Chuck Norris would win. He has the unfair advantage of being immortal and having a third fist (it's under his beard).
I wonder what pompous ass keeps neging anyone that comments with a bit of humor? Gonna give Fresh a point.
I think you are overanalyzing things. makes me think of the numerous times when the government puts together a task force to determine if doing "x" will result in something bad (i.e., is second hand smoke bad for your health).
In this example, common sense or logic is a better way to figure things out. When you are tired, you are more likely to make mistakes and not perform as well. So, if the M team is better conditioned, they will be less tired and peform better.
There are so many other variables in football that doing an analysis with historical numbers is basically impossible.
Most of this has already been covered in some way. The quarters shouldn't be compared to eachother. Each quarter is very different fundamentally from a coaching strategy standpoint, so conditioning should have very little to do with patterns of points between 1st and 4th quarter.
You would need to focus just on 4thQ and compare to other teams in same conf (as suggested above) or other successful teams/coaches. Even then, I'm not sure that would tell you much about conditioning.
I think for conditioning we just have to trust all the accolades we hear from everyone about Barwis, and understand that his job is also strength not just conditioning and further expect that no matter when it is during the game our players should never have an excuse of being tired. That said, we score so quickly (see time of possession disparity in last game) that our offesnive players really shouldn't be tired; our defense should have been tired. But they didn't seem to be, they seemed to have a consistent intensity level, even in the second half.
Observation: for what it's worth; your averages indicate that RR was up 25-15 on average going into the 4th quarter. The hypothesis you suggest you are testing is ". . . That many drawing the conclusion that the superior conditioning of Michigan’s football team will result in wearing down teams through three quarters, and dominating a tired opponent in the fourth for a win", but it appears that there is little need to dominate the 4th Q - rather just maintain the margin.
I concur that the conditioning is primarily for fast-twitch muscles, changing direction, speed bursts, quick power -- NOT endurance.
I agree with that...the difference is going to show up in speed and quickness more than endurance. Having said that they have seemed pretty strong at the end of these first few games. It may actually show as more of an advantage early in the game before the opposition adjusts to the speed.
Is there anywhere in particular you watch games that are not nationally televised (i.e. BTN)?
Let me know. I'm in the Lou as well and I'm sure my wife would love it if I didn't drag her along to the bars to watch games with me.
