Indiana - By The Numbers Recap

Submitted by The Mathlete on
A couple of quick notes, the numbers look better than they seem like they should.  Opponent adjustments are based 100% on this year's results.  As the year progresses and Indiana faces stiffer competition, the numbers from the game will likely come down. 

Run Offense vs Indiana

Win for Indiana here as Michigan scored a +2 (two points better than the average team against Indiana) while Indiana scored a +8 (eight points better than the average team would allow to Michigan)

RBs:
Carlos Brown +9 (includes receptions) #2 RB performance of the week
Brandon Minor +1
Michael Shaw -1

Pass Offense vs Indiana

This one was a win for M. Michigan posted a +7 while Indiana posted a middle of the road -1.

QBs:
Tate Forcier +7 #22 QB performance of the week
Denard Robinson +2

WR:
Martavious Odoms +6

Run Defense vs Indiana

Another win for Indiana here, obviously.  Despite Michigan poor job in previous games against the run, the Hoosiers still "beat the spread" going +2 against the Wolverine rush D while the D was 3 points worse against Indiana than the average team.

RBs
Darius Willis +4
Trea Burgess -3

Pass Defense vs Indiana

Indiana was actually below average passing against Michigan, with a -1 while Michigan was +4 vs the Indiana passing game.

QB
Ben Chappell -1

WRs
Tandon Doss: +11 #6 WR performance of the week
T Turner +2

Field Position

Based solely on drive starts, Michigan should have lost the game 30-27, indicating that the offense overachieved by 9 points and the defense underachieved by 3.

Special Teams

Brian pointed this out right away but the ST were huge for Michigan this weekend.  My punting stats are based on a net average with adjustments made for shorter punts that are downed deep in opponent territory.  Zoltan's day was worth 2.5 points, a huge number for a punter and the 5th best tally for any punter in an FBS game this year.  Overall, Michigan special teams had the 22nd best performance of the year, which is tough to do without returning a kick for a touchdown.

Penalties

When Michigan was on offense, the effect of the penalties netted out to zero, despite Michigan getting hit with more yards.

Michigan picked up a single point of benefit when Indiana had the ball.
football
Indiana
By The Numbers

Comments

OSUMC Wolverine

September 28th, 2009 at 4:02 PM ^

Now brew me up a big mathematical picture as how badly we will mangle MSU. Not that I need the reassurance, but it will make me feel all warm and fuzzy inside. I don't think we should be so quick to write off Indiana as a fluke winning record, I think .500 in the Big Ten is possible for them, or at least 3-5, and that a Bowl season should make.

Joined: 09/06/2008

MGoPoints: 4403

Nothsa

September 28th, 2009 at 5:05 PM ^

This might be their best team in many years (I've followed IU since the 80's), but the schedule does Indiana no favors: they miss MSU and Minnesota, and the structure of the schedule is poor. They host OSU this weekend, which they will lose. Then, after a game at Virginia, comes the easiest part of the schedule: Illinois, @NU. I think they split those, but there's a chance the Hoosiers get both. Then @Iowa (loss), Wisco (probably a loss), @Penn State (almost surely a loss), and they finish at home in the Bucket game against Purdue, which appears to be a tossup. In short, there are three tossups: Illinois, NU, and Purdue. They will be favored in one or two of those. There are three really longshot possibilities: OSU, @Iowa, and @Penn State. Wisconsin might be a game they can steal. 4-4 is possible for IU in the same way 8-0 is possible for Michigan - it could technically happen, but no. 3-5 is the best-case realistic scenario, and 2-6 would not represent a collapse. But if they go 4-0 out of conference (A real possibility, as their remaining opponent, Virginia, looks awful) that bowl season is definitely a possibility. I wish they were playing MSU in place of OSU or PSU though.

Joined: 10/07/2008

MGoPoints: 1868

colin

September 28th, 2009 at 4:25 PM ^

but with charts. Lots of charts. Would like to see all Big Ten teams. Also: "Based solely on drive starts, Michigan should have lost the game 30-27, indicating that the offense overachieved by 9 points and the defense underachieved by 3." Could you flesh out your drive start scoring expectance doohickey? I've been using a simple linear equation ( Pts = .035YdLine + .728) that I derived from a Big Ten 2008 season play database that I put together. It gave 27-27 based on the drive starts, so it looks like we're close at least.

Joined: 07/01/2008

MGoPoints: 1472

The Mathlete

September 28th, 2009 at 5:35 PM ^

I don't use a straight formula, I use smoothed data from all 2008 games of FBS vs FBS opponents. I have a wealth of data and still trying to find the best way to present it and present the value, very open to any ideas. Here is what I used to get to 30-27. Drive Offense Scrimmage Start Expected Points 1 Indi 20 1.59 2 Mich 39 2.21 3 Indi 14 1.31 4 Mich 29 1.8 5 Indi 60 3.18 6 Mich 7 1.18 7 Indi 33 1.98 8 Mich 38 2.18 9 Indi 24 1.69 10 Mich 19 1.55 11 Indi 41 2.28 12 Mich 30 1.82 13 Indi 71 3.88 14 Mich 30 1.82 15 Indi 42 2.32 16 Mich 19 1.4 17 Indi 69 3.07 19 Mich 49 2.62 20 Indi 2 0.97 21 Mich 40 2.25 22 Indi 13 1.42 23 Mich 29 1.8 24 Indi 24 1.69 25 Mich 25 1.71 26 Indi 15 1.46 27 Mich 26 1.73 28 Indi 24 1.69 29 Mich 48 2.58 30 Indi 26 1.73

Joined: 10/01/2008

MGoPoints: 2905

octal9

September 28th, 2009 at 7:23 PM ^

Drive # Team Scrimmage Start Expected Points
1 Indi 20 1.59
2 Mich 39 2.21
3 Indi 14 1.31
4 Mich 29 1.8
5 Indi 60 3.18
6 Mich 7 1.18
7 Indi 33 1.98
8 Mich 38 2.18
9 Indi 24 1.69
10 Mich 19 1.55
11 Indi 41 2.28
12 Mich 30 1.82
13 Indi 71 3.88
14 Mich 30 1.82
15 Indi 42 2.32
16 Mich 19 1.4
17 Indi 69 3.07
19 Mich 49 2.62
20 Indi 2 0.97
21 Mich 40 2.25
22 Indi 13 1.42
23 Mich 29 1.8
24 Indi 24 1.69
25 Mich 25 1.71
26 Indi 15 1.46
27 Mich 26 1.73
28 Indi 24 1.69
29 Mich 48 2.58
30 Indi 26 1.73

Joined: 11/11/2008

MGoPoints: 5777

colin

September 28th, 2009 at 8:19 PM ^

the linear trend on that is almost exactly the same as mine, but it breaks down a bit past the 60. you could definitely use the exponential regression as a very good approximation and the linear for short hand. using the exponential, i have Michigan's offense at ~46 points above expected. that in the ballpark? as for presenting it, Michigan's points scored against expected compared to some rivals could be charted. that'd be fun to look at. more generally, show how it's useful. e.g. that yards and points and wins are connected through these expectations.

Joined: 07/01/2008

MGoPoints: 1472

B Ready

September 29th, 2009 at 12:39 AM ^

Great, great stuff. I am curious, though, on 2 things... There has been new information coming out at Football Outsiders that suggests that QB's should take some blame for sacks. Why is that not part of your ranking for QB's? And, catch rate is pretty constant to different receivers and the guys at Football Outsiders believe that targeted receivers should receive some of the blame for incomplete passes thrown their way. Why is that not part of your ranking for WR's? I'm just curious more than anything. It sounds like you come from a sabermetric background and that goes against new saber research on football.

Joined: 01/13/2009

MGoPoints: 113

The Mathlete

September 29th, 2009 at 11:09 AM ^

I have gone back and forth on some of these issues myself. I can easily include anything that's currently excluded (primarily sacks for QBs, fumbles lost and non-receptions for receivers). I do track the target vs catch for receivers and am toying with a metric for evaluation, something like gap percent which shows all targets/all catches to determine how effectively utilized a receiver is. I will add this into future updates. The other difficulty on this is that all I have is NCAA play by play data and there seems to be some discrepancy from team to team on how targets are recorded, drops and such should always be counted but if the receiver is noted on a throwaway, that shouldn't count against him. Ultimately, a UFR style video review would be great, but with 120 teams, its just not possible. For QBs, I agree that some value is due to the QB for sacks, but my view is that in general, the majority falls into the offense in total. Therefore, sacks count in team passing stats, but not individuals.

Joined: 10/01/2008

MGoPoints: 2905