Is Our O-Line Really That Young? Third Time is the Charm!!!
Just when I thought I was I out, they pull me back in!
I don't know why I am such glutton for punishment, but I am finding this topic interesting (and not just in football sense, but statistically as well). I want to contribute one last time.
Many people on the threads have pointed out that just counting the class experience (basically age) is not enough, you need to count the actual games started as well.
I agree, games started should be part of this analysis.
AmazinBlue pointed out that Phil Steele has published a convenient list of all the games started by the players on the roster before the season began (http://www.philsteele.com/Blogs/2013/JUN13/DBJune08.html). Since the data is so handy, I figured I would go ahead and combine both sets of data and make a handy dandy XY Scatter chart. X-axis is the total combined number of Class Experience (i.e. Frosh=1, rs Frosh = 1.5) and Y-axis is the total number of previous games started.
As you can see from above, Michigan is in a better place than at least four teams (Auburn, UCLA, LSU, and Texas Tech), and surprisingly not that far away from Alabama.
Statistically, Michigan is within one standard deviation from the mean on Total Games Previously Started and just .16 away from one standard deviation for Total Class Experience. That, by definition, says Michigan o-line is not an outlier.
Again, the data says Michigan o-line is young, but not "outlier" young. There are other teams in top 25 who are just as inexperienced and a few who are even in a worse position. Blaming all of our woes on o-line experience does not paint the entire picture.
November 7th, 2013 at 12:03 PM ^
There are two definitions of "skew".
It's a technical term, a measure of the asymmetry of a set of data. That's how I've been using it, and Space Coyote.
It's also a word used colloquially for a measure that's misleading, intentionally or unintentionally, because of an outlying data point. ("Lewan and Schofield are skewing the data.")
They're two different things. My comments on skew don't have anything to do with using an average across each o-line.
November 7th, 2013 at 12:55 PM ^
...at comparing Michigan to a set of peers that aren't selected based on this year's performance.
For now I'll just print the raw data and let you folks have a go with whatever metric you want to use (except for convenience, and because it's his diary after all, I'll calculate gameboy's metric for him). The starters are based on last week's participation reports, except that Ohio State weirdly lists the same player starting at both LG and RG so I plugged in the RG we all know is playing. (As an aside, Ohio State's website blows. They have the sparsest, weakest bios on their players I've seen yet.)
The names take up a lot of space so I'll just put the classes, It's LT, LG, C, RG, RT, gameboy's metric for the line as a whole, and then that team's ranking in offensive FEI (out of 125 FBS schools). Schools are presented in order of offensive FEI, too, to get the good offenses at the top and the tire fires at the bottom.
- Indiana: SO, JR, RS JR, RS FR, RS SO, 2.5, 6
- Ohio State: RS SR, SR, RS SR, RS SR, SO, 3.9, 7
- Wisconsin: RS SO, RS SR, RS JR, RS JR, RS JR, 3.5, 21
- Michigan: RS SR, FR, RS SO, RS FR, RS SR, 2.8, 43
- Iowa: RS JR, RS SR, RS SO, RS SO, RS SR, 3.5, 51
- Nebraska: RS SR, RS JR, RS SR, RS JR, SR, 4.0, 52
- Illinois: RS JR, RS JR, RS JR, RS SO, RS SR, 3.5, 54
- Michigan State: RS FR, RS SR, RS SO, RS SR, SR, 3.4, 59
- Minnesota: RS SR, RS JR, RS SO, RS JR, RS SO, 3.3, 62
- Northwestern: JR, RS SO, RS JR, RS SO, RS JR, 3.0, 67
- Penn State: RS SR, RS JR, RS SR, RS SR, RS JR, 4.1, 84
- Purdue: RS SR, RS FR, RS SO, RS FR, RS SR, 2.9, 118
November 7th, 2013 at 1:22 PM ^
If I find the time maybe I'll do another conference or two sometime. There's not enough here to really drive any conclusions but here are some thoughts, some of them not new:
1. Michigan's line is indeed young. That's 35 lines we've looked at and only 4 have been younger by gameboy's metric, and I don't think anyone is going to argue that his metric understates Michigan's experience.
2. Some offensive schemes emphasize line play more than others. IU doesn't ask it's line to do too much--that system is all about the QB and WRs (and the novelty of the extreme pace). It's possible Michigan would be better this year if they were doing something like that. But at some point you have to get some stability in your program and install the system you plan to run long-term. There may be short-term and long-term goals in conflict here.
3. Michigan's offense may look terrible but it isn't historically bad. Somehow it's managing to perform above the level of Carr's last offense, or Rodriguez's first two, and it's in the upper half of the conference.
4. If there's a correlation between gameboy's metric and offensive performance, it isn't very strong. Obviously there's a lot more that goes into an offense, but for me it's an open question: would another metric have a stronger correlation, or is the influence of raw line experience jsut too weak to show through? I'll be interested to see how reshp1's metric does here.
5. I'm heading this off at the pass--the "interior is what matters" crowd shouldn't cherry pick Purdue. But I'll bet it's tempting to do it....
November 7th, 2013 at 1:38 PM ^
"Ohio State's website blows. They have the sparsest, weakest bios on their players I've seen yet."
Maybe it's like my mother always said: if you don't have something nice to say about someone....
But I digress.
November 7th, 2013 at 1:22 PM ^
November 7th, 2013 at 2:20 PM ^
That I hadn't thought about. One thing that we are consistently stuck doing though is simply comparing to others, which doesn't really take into account how others are performing. Really, all we're figuring out, is how the experience is relative to others and trying to base an outcome on that, rather than taking an experience and determining an expected outcome based on that experience. I'm sure that could be done, but like you said, that'd take a lot of work.
November 7th, 2013 at 3:04 PM ^
And that's going to have to be it for now. Here's the PAC12, same format as before.
- Arizona State: RS SR, RS JR, RS SR, RS SO, RS JR, 3.7, 3
- Oregon: RS SO, RS SR, RS JR, RS JR, FR, 3.0, 10
- Washington: RS JR, RS JR, RS JR, RS JR, RS JR, 3.5, 11
- Utah: SO, RS SR, RS SR, RS FR, RS SO, 3.0, 15
- Arizona: RS JR, SO, JR, RS SR, RS JR, 3.3, 20
- USC: RS FR, RS SR, JR, JR, SO, 2.8, 29
- Oregon State: RS SR, RS SR, SO, RS SR, FR, 3.3, 31
- Stanford: SO, SR, RS SR, RS SR, RS JR, 3.7, 35
- UCLA: JR, FR, RS SO, FR, FR, 1.7, 44
- Washington State: RS SO, RS SO, RS SR, RS SR, SR, 3.6, 66
- Colorado: RS SR, RS JR, RS SR, RS JR, RS SO, 3.7, 87
- California: RS FR, FR, RS SO, RS JR, RS FR, 2.0, 93
Can we agree yet that Michigan's line is young? Of these 24 teams only 3 have younger lines by gameboy's metric.
That's the top 25, the B1G and the PAC so far. We're about a third of the way through FBS and about halfway through the major conferences, and here's a list of all teams with two or more freshmen (true or redshirt) starting:
California, Michigan, Purdue, UCLA
Michigan's been the best offense of the four. UCLA's been basically its equal, the other two have been horror shows.
November 7th, 2013 at 3:32 PM ^
I'm getting all bogged down in arguing stuff in the front of the thread, I'm missing on the good stuff in the back! (That's what she said...?)
Nice work, Yeoman.
November 7th, 2013 at 6:54 PM ^
I found Yeoman's info interesting as well.
November 7th, 2013 at 3:40 PM ^
This is all really good stuff, and it's one of the more interesting discussions I've seen on the blog in quite awhile (apart from 200+ threads calling for everyone to be fired, which were, of course, the best). I appreciate Gameboy taking the time to synthesize and visualize some of this data.
One of the biggest critiques so far seems to be "who cares about our awesome tackles, the NT is destroying our true freshman LG." Basically our interior line is why we suck. Statistically, would it make sense to try and determine whether interior line experience is more important than tackle experience in producing an effective offensive line? I mean, could you run a regression with interior line expereince against ypc and do the same with tackle experience against ypc to see whether one or the other is more important? And if there is no difference, would that imply that mean average isn't that bad of a measure?
Just curious, I find this all a pretty interesting. Good work, people.
November 7th, 2013 at 4:06 PM ^
Downvoted because you chose not to edit your original diary to post this when you easily could have. You're pushing other diaries off the front page with all the re-postings.
November 8th, 2013 at 10:10 AM ^
It would be interesting see where UofM's 2011 and 2012 OLs end up in the scatter plot.
I think it would also be good to add in UofM's 2008, 2009, and 2010 OLs. I believe the 2008 line featured one starter from 2007 and 4 others who had zero starts. The 2008-2010 teams also had 1st year starters at QB (Threet in 2008, Forcier in 2009, and Denard in 2010)
November 8th, 2013 at 2:18 PM ^
I've mentioned this before, but if we are assessing experience we need to look at snaps as well as starts. I remember last year before the Alabama game someone mentioned that the "new starters" on Alabama's oline had all played quite a bit in several games prior to starting. They alluded to this being the norm at Alabama. I did not bother to check on it so my memory may be flawed or that comment may have been in error.
In any event, my point is that there is a tremendous difference between a Sophomore with no starts who played say a full quarter in 8 games the previous season and a RS freshman or true freshman whose never seen the field. When a player is starting for his second year the importance of the snaps he saw before his first year as a starter diminishes. When he is starting his first year the importance of the experience gained by those prior snaps probably far more important.
Lacking information about the number of snaps creates a problem with the granularity of the data that washes out as the number of starts increases, but misses key experience among younger less experienced players.
I don't know if number of snaps by player is available if so it might be interesting to look. It also might be interesting to track starts by each player and compare where each starting position maps against each other. It would be interesting to see for example if coaches value experience more at one position than another.
Comments