Quick introduction

Below is a quick analysis of some Upon Further Review (UFR) Plus/Minus data from MGoBlog. Brian from MGoBlog breaks down the offensive and defensive plays from the previous week’s game and assigns “pluses” to positive plays from players and “minuses” to mistakes. It’s not unlikely he gets things wrong here and there, but overall it seems to be a reasonable and in-depth assessment of a players effectiveness. For more detail I highly encourage you to head to the site and read his work, if you are somehow reading this and not already deeply familiar with it.

I’ve acquired data for every game he’s charted, though I have only broken down the +/- data so far, which does not take a QB’s passing quality into account, nor details like route running for receivers. I will get to some of that data in the future, but keep in mind that the main assessment of offensive skill players is missing from this data currently.

Also, keep in mind that there is only data for games that Brian wrote up a UFR, so there may be some bias comparing years due to games left out because he was too sad to look back at them.

With that, I’ll get right to some brief exploration of the data set.

(As a note: I also plan to post an interactive visualization for others to use in the near future. I will hopefully deploy it in the next few days.)

A first look at improvement in offensive playcalling

One of the numbers Brian keeps is the RPS (Rock-Paper-Scissors) count for the offense and defense. The gist of RPS is that the coaching staff gets between +1 and +3 when the play call gives their team decided advantage without any of their players needing to execute their role “above and beyond” the play call. They are given a -1 to -3 if it’s the other direction: if their play call meant that was doomed, even if perfectly executed.

An excellent coaching staff, one would expect, would “make things easy” on their players by given them a schematic advantage. Keep in mind, the coaching staff affects the team in many other ways and things that may look to Brian and others as a bad call one play, might be part of a larger scheme that is less apparent to a viewer.

The figure above shows the game-by-game RPS+ and RPS- scores, as well as the total score, RPS+ + RPS-, from 2011 to the most recent game. The colors separate the years for easy comparison. The x-axis is a bit cluttered, but we can see a general downward trend in the RPS total until 2015, driven both by declining RPS+ (especially in 2014) and a peak in RPS- in 2013.

It is noteworthy that 2015 is the first year Michigan has been consistently above 0 or around 0 for RPS. Also, as Brian has pointed out, the offensive play-calling has been conservative late in recent games, which limits the upside for RPS.

We can see even more clearly the improved RPS thus far this season by looking at the mean RPS for each season:

Keep in mind, 3 very tough defenses lie ahead - Minnesota, PSU, and OSU - however, early results are very encouraging.

Looking at position group improvements

The UFR charts break down offenses into Offensive Lineman, Backfield, and Receivers, while the defense is split into the Defensive Line, Linebackers and Defensive Backs. Before looking at any particular players, we can look to see how position groups have been improving or worsening over time. Some players may be listed in different places - looking at you Funchess. I left them listed in the group that Brian had them in for any given game, assuming that he had them in the place that fit their role that game best.

First, we can take a look at the…

Offensive Line:

Brian started charting offensive +/- in 2010. We can see the line’s output really cratering in 2013 and 2014, with the 2013 Penn State game being the clear bottom of the bottomless pit. The RPS-s are surprisingly stable, though the scale of variations may be somewhat obscured by that “27 for 27” game.

Like the play-calling, this is another area where we look substantially better and more consistent in 2015. Looking at the year-by-year averages:

We see that the O-line has seemingly rebounded quite nicely. Caveats about remaining, noteworthy, definitely respected opposition aside, the line seems to have strongly pulled out of the rut. Again, the line did seem to drop off late in the BIG season in 2010, 2012, 2013 and 2014 also, so we are not out of the woods for some regression here. Kevin Tolbert, do your thing buddy!

Some quick summary information about the position group:

## Mean total for  OL  is:  2.032828 
## Mean number of + for  OL  is:  4.729798 
## Mean number of - for  OL  is:  2.69697 
## OL  with the most single game + is:  Omameh  with  21  in  2010  against  notre-dame 
## OL  with the most single game - is:  Omameh  with  -16  in  2010  against  uconn 
## OL  with the best single game total is:  Omameh  with  19  in  2010  against  notre-dame 
## OL  with the worst single game total is:  Kalis  with  -11  in  2013  against  penn-state 
## Player with the best single season total is:  Molk  with  92  in  2011 
## Player with the worst single season total is:  Funchess  with  -11  in  2013 
## Player with the best career total is:  Lewan  with  219 
## Player with the worst career total is:  Funchess  with  -14.5

It’s probably not fair to consider Funchess a lineman for this, as he was moved out to WR fully only a couple games into that season, so let’s summarized excluding him:

## Mean total for  OL  is:  2.114187 
## Mean number of + for  OL  is:  4.831315 
## Mean number of - for  OL  is:  2.717128 
## OL  with the most single game + is:  Omameh  with  21  in  2010  against  notre-dame 
## OL  with the most single game - is:  Omameh  with  -16  in  2010  against  uconn 
## OL  with the best single game total is:  Omameh  with  19  in  2010  against  notre-dame 
## OL  with the worst single game total is:  Kalis  with  -11  in  2013  against  penn-state 
## Player with the best single season total is:  Molk  with  92  in  2011 
## Player with the worst single season total is:  Bryant  with  -10  in  2013 
## Player with the best career total is:  Lewan  with  219 
## Player with the worst career total is:  Bryant  with  -10
## The career top 5 OL are: 
## 1) Lewan with 219
## 2) Molk with 173
## 3) Omameh with 147.5
## 4) Schofield with 120.5
## 5) Glasgow with 94
## The season top 5 OL are: 
## 1) Molk with 92 in 2011
## 2) Molk with 81 in 2010
## 3) Omameh with 78.5 in 2010
## 4) Lewan with 64 in 2013
## 5) Lewan with 63 in 2011
## The game top 5 OL are: 
## 1) Omameh with 19 in 2010 against notre-dame
## 2) Lewan with 17 in 2013 against indiana
## 3) Schofield with 16 in 2013 against indiana
## 4) Lewan with 15 in 2013 against cmu
## 5) Molk with 14.5 in 2011 against sdsu

Now onto the…

Backfield (RB) and Receivers (WR):

I lump them together because I feel the +/- communicates the least about these groups. For the backfield because the changing nature from mobile QB’s in Denard and Devin to a pro-style O obscures any other trends. Many +/- came from QB decisions on reads/pitches and things of the sort. For receivers, +/- mainly just reflects their blocking skills, which is handy to check, but doesn’t communicate much of their impact. Still, it doesn’t hurt to show the plots:

I can’t really tell if the big drop in production from the backfield is all about the change from mobile QB’s or if it also has to do with RB’s needing to make plays.

Only partial summary info here:

## The career top 5 RB are: 
## 1) Robinson with 161.5
## 2) Toussaint with 130.5
## 3) V. Smith with 77.5
## 4) Gardner with 48
## 5) Kerridge with 43
## The season top 5 RB are: 
## 1) Toussaint with 78 in 2011
## 2) Robinson with 75 in 2010
## 3) Robinson with 45.5 in 2011
## 4) Robinson with 41 in 2012
## 5) V. Smith with 40 in 2011
## The game top 5 RB are: 
## 1) Toussaint with 18 in 2011 against nebraska
## 2) Robinson with 14 in 2012 against purdue
## 3) Toussaint with 12 in 2011 against illinois
## 4) D. Smith with 11.5 in 2015 against oregon-state
## 5) Robinson with 11 in 2010 against uconn
## The career top 5 WR are: 
## 1) Gallon with 41.5
## 2) Roundtree with 21
## 3) Odoms with 13
## 4) Stonum with 13
## 5) Norfleet with 12.5
## The season top 5 WR are: 
## 1) Gallon with 15.5 in 2011
## 2) Roundtree with 14 in 2010
## 3) Stonum with 13 in 2010
## 4) Gallon with 12.5 in 2012
## 5) Odoms with 11 in 2010
## The game top 5 WR are: 
## 1) Manningham with 9 in 2006 against notre-dame
## 2) Breaston with 5 in 2006 against notre-dame
## 3) Stonum with 5 in 2010 against umass
## 4) Gallon with 5 in 2012 against purdue
## 5) Chesson with 5 in 2013 against cmu

Next onto the defense… here or back to home