Oh, I absolutely agree that the secondary is quite bad, and I grant that the UFR is inexact. However, Misopogon's two benchmarks for the star ratings were a UFR-equivalent (which I was speaking too), and qualitative adjectives like "all Big Ten". The player analogues are probably the most useful in terms of benchmarking a player's ability, though I must admit my familiarity with historical players isn't good enough to assess their validity.
I guess the issue I was grappling with is that we have not just one weak player, but several, and the UFR statistic seemed the best way to think about how all the players might fit together. We all know that the d-line has been pretty good, the linebackers have been up and down, and the defensive backs have been pretty terrible. My point was not to dispute that, or to trash Misopogon's ratings (which I quite enjoyed reading) - I just wanted to suggest that the defensive backs are bad but not "commit a huge blunder every three or four plays" bad.
I think this is an interesting exercise, and I want to thank Misopogon for putting in the time to think through this. However, it seems that his ratings are too pessimistic (especially in the secondary), given the UFR-equivalent definition he chose for each star and Brian's defensive UFRs.
As a rough approximation I assumed an 80-20 playing time split between starters and backups, and I counted Ezeh as the starter. Using a 70-30 or 60-40 split wouldn't change things much. Based on the definition of -2 for a one star to +2 for a five star every four plays, we should expect the following totals for each unit:
DL: +0.8 per four plays
LB: -1.6 per four plays
DB: -5.4 per four plays.
Total: -6.2 per four plays
If we add up Brian's UFR ratings, we get the following:
DL
LB
DB
Overall
U Conn
14
3
7
24
ND
13
21
-11
23
U Mass
29.5
-14.5
-8
7
BGSU
10
16
2
28
Indiana
32
3
-20.5
14.5
MSU
8
8.5
-9.5
7
Total
106.5
37
-40
103.5
Total 2 (Iowa = MSU)
114.5
45.5
-49.5
110.5
Total / 4 Plays
0.98
0.34
-0.37
0.95
Total 2 / 4 Plays
0.92
0.37
-0.40
0.89
Because Brian hasn't done the Iowa game yet, the first total ignores it, and the second total assumes it'll be the same as the MSU rating (imperfect, but the best I could do). Our defense has faced 496 plays (436 if you don't count Iowa).
As you can see, our DL has performed in line with your ratings, but our LBs and DBs have done much better. The DB rating is clearly an order of magnitude too negative. Averaging -5.4 per four plays means that in a 60 play game they should have a total UFR of -81.... four times worse than the Chappel-bombing.
Now, we could instead ask whether the stars are fine in a relative sense, but that the pluses and minuses associated are too extreme. I think that is part of it, but on the other hand according to Brian's UFRs the LBs have been a net positive, as has the team, and the DBs have been half as negative as the DLs have been positive. By contrast in Misopogon's ratings the LBs are rated as negatives, and the DBs are rated as almost 7 times more negative than the DLs are positive. Again, this suggests that the linebackers and (especially) the defensive backs aren't as bad as Misopogon's ratings would suggest.
I just listened to Brian's section of the podcast, and he makes an excellent point about the "Greg Jones as a spy" defense. On a running play, the MLB is always going to have a blocker assigned to him, and is unlikely to be the read key on an option - so it's not clear that having Jones "spy" will do much of anything. As Brian said "he'll just be playing run defense". I think spying the QB would matter more if Denard was scrambling out of the pocket on passing plays, or if we were running lots of true QB draws where we pass block for a few seconds at the start of the play. It also seems like Jones spying should open up the underneath coverage on play-action passes.
I do like FO's stats, I was just making the point that S&P+ is a derived statistic based on a model of what's important, not a raw statistic. I think both are important to look at.
Recently Wisconsin has tended to do worse than expected when initially ranked highly, and better than expected when initially ranked low. They'll probably end up in the middle of the pack in the Big 10 this year.
You have to move the sack yards from rush defense to pass defense for everyone before reranking all the teams. We move ahead two spots because two teams near us (Cal and Tulane) have more sack yards per game thank we do, so their rush defense gets worse by a larger amount than ours.
Specifically, we have 13.6 sack yards per game. Cal has 18.5 and Tulane has 18. That means for rushing defense we get...
Michigan: was 125.8/game (#38), now is 139.4/game (#36)
Tulane: was 119.75/game (#33), now is 142.25/game (#38)
I'm not sure why there is a slight difference from the NCAA.com stats you link to. It matters for Cal, since your link has them at 126 rush yards/game, moving them down to #38. If we take the NCAA.com stats then the raw numbers likely won't change much but we'll only move up one rank instead of two (since we'd start one rank higher).
The Football Outsiders stats are interesting, but do a lot different than just control for sacks and SoS.
If you just want to move sack yards from rush defense to pass defense I find that it makes no difference for our passing defense rank, and actually improves our rushing defense rank:
Pass Defense: was 307.8/game (#120), now is 294.2/game (#120)
Rush Defense: was 125.8/game (#38), now is 139.4/game (#36)
Brian Fremau talks about some of these same issues in his Football Outsiders column this week (http://footballoutsiders.com/fei-ratings/2010/fei-better-late-never). He handles things a little differently, since he strips out "garbage time" posessions and he ignores games with FCS schools. He only lists a couple of top 10 tables, but he lists Michigan as 9th in terms of points per possession on offense with 3.386. Top 10 defenses allow somwhere in the 0.5 to 1.3 points per posession range, so we're letting teams score twice as often as that. He also shows that top 10 defenses are able to make a stop for a drive that has already gone 10+ yards at least two-thirds of the time, where we're only doing it half the time (see below).
Overall he's got us ranked 16th with a rating of 0.142 (I don't have a good feel for the units on his metric, but the #1 team is 0.297 and the #25 team is 0.122. projects us to win on average 4.4 of our remaining games (so end with a total of 9.4 wins). For comparison Michigan State is ranked 31st with a rating of 0.95. He doesn't have a prediction yet (will probably come out later today), but for reference last week he had MSU at 0.73 and Wisconsin at 0.69 and gave MSU a 60% chance to win. He had Iowa at 0.148 and Penn State at 0.102 and gave Iowa a 71% chance to win. So I'm guessing we'll be like 65-70% favorites in his system.
I definitely think it would be interesting to break drives down by number of plays rather than yardage - I used yardage because that was the information bigmc6000 had in original table. I do think the yardage breakdown is useful if you think about three groups - immediate stops, short drives and long drives. About a third of the drives against our defense stop in 10 yards or less (i.e. one set of downs), about a third go for a while and then stop, and a third drive for 50+ yards and mostly score. Field position certainly plays into this, but again wasn't in the data yet.
I think the other thing that you can tell from yards is that when a drive has gone a long way, we basically need a turnover to stop it. We're not very good at just forcing a stop when the other team has already gotten deep in our half of the field. You see that clearly in the second table, which looks cumulatively rather than by total drive yardage (so avoids the problem of categories with few observations). The longer a drive has already gone, the more we need a turnover to stop it.
I would definitely be in favor of someplace where basic scraped data can be centrally located. bigmc6000 posting the individual drives in addition to his analysis made it easy for me to add my two cents. It would be very useful, for example, to have the basic play-by-play numbers from the UFRs in a spreadsheet. On the other hand people like Mathlete who do a lot to transform the raw data into something else like PAN shouldn't feel like they should have to share that (though if he wants to that'd be great!).
Oops, I left in the EOH drives for the total drives, but not for stops. Here are the corrected tables. I also switched the Indiana EOH touchdown to be a scoring drive.
Sure, go ahead. The other thing I was hoping to pull from the UFRs was whether on defensive stops we got other kinds of "big plays" in the last set of downs - sacks or TFLs.
Thanks for doing this - I wanted to go through the UFR's to look at what our defensive stops looked like, but I can use your data to do most of what I had in mind.
I took a look at all the drives that we got stops on (other than EOH drives), breaking them down by how long the drive was. I think this will show how much we're bending, and if we bend do we break.
Drive Length
Drives
% of Drives
Stops
% of Stops
Stops/Drives
Stops w/ TO
<= 10 yards
20
31%
16
42%
80%
19%
11 to 20 yards
3
5%
3
8%
100%
33%
21 to 30 yards
12
19%
10
26%
83%
30%
31 to 40 yards
3
5%
3
8%
100%
33%
41 to 50 yards
5
8%
3
8%
60%
0%
51+ yards
21
33%
3
8%
14%
100%
So, we actually get a fair number of quick stops. However, the longer the drive is, the more dependent we are on getting a stop via a turnover (interception, fumble or missed field goal) rather than just stiffening.
Another way to look at that is to ask, given that a drive has already gone X yards, what is the probability of a stop and/or a turnover?
Drive of at least
Prob(Stop)
Prob(TO)
Stops w/ TO
10 yards
50%
18%
36%
20 yards
46%
17%
37%
30 yards
31%
14%
44%
40 yards
23%
12%
50%
50 yards
14%
14%
100%
Not surprisingly, the longer the drive goes, the less likely we are to get a stop, and the more likely we are to need a turnover to get a stop. With Brian now including the number of rushers in UFR it would be interesting to see whether we're more likely to blitz the longer a drive has gone in order to get a TO.
So things actually look a bit better when I redo the analysis only for AQ conference teams.
Year
End 70th or better
End +15
Avg Incr
2004
12.5%
12.5%
4.6
2005
37.5%
37.5%
26.6
2006
20.0%
30.0%
6.9
2007
22.2%
44.4%
11.2
2008
40.0%
40.0%
24.4
2009
33.3%
44.4%
18.0
So, just based on historical averages, we have something like a 30% chance to improve enough to end up 70th or better in total defense, and something like a 40% chance to improve by at least 15 ranks.
I left the spreadsheet that had all the data at work, so I won't be able to check until tomorrow. I do have a list I was working on of all the teams that increased by at least 40 ranks. Of the 16 teams that did so, 8 were from AQ conferences.
In 2009, Syracuse went from 96 to 37, and Colorado went from 101 to 57.
In 2008, UCLA went from 102 to 47, and Purdue went from 109 to 61.
In 2006, Boston College went from 92 to 34.
In 2005, Oregon went from 90 to 44, LSU went from 102 to 3, and Arkansas went from 103 to 34.
I took a look at how often teams with terrible defenses early in the season improve to be somewhat better than terrible. It was easiest to compare Aug/Sept rankings to final rankings, so I looked at how many teams ranked 90th or below in total defense at the end of Sept improved to be at least 70th or above by the end of the year. Michigan was 92nd at the end of Sept (and is currently 102).
2004: 14%
2005: 23%
2006: 20%
2007: 17%
2008: 13%
2009: 16%
If instead you look at how many of the teams 90th and below after September improve at least 15 ranks by the end of the year, you get the following:
2004: 34%
2005: 44%
2006: 23%
2007: 30%
2008: 26%
2009: 32%
So, there's a small chance we can get "slightly below average" or better, and a reasonable chance we can improve somewhat.
I posted a more detailed breakdown of RR's teams below, but Denard has already gotten in 5 games as good or better than what 2005 Pat White did for the whole season.
Denard has 1008 passing yards on 96 attempts (10.50 YPA) compared to Pat White's 828 yards on 114 attempts (7.26 YPA).
Denard has 905 rushing yards on 98 attempts (9.23 YPC) compared to Pat White's 952 yards on 131 attempts (7.27 YPC).
For total offense Denard has 1913 yards on 194 plays (9.86 YPP) compared to Pat White's 1780 yards on 245 plays (7.27 YPP).
I just registered to post this (after lurking for a while), so I'm adding it as a comment rather than starting a new thread/diary. I apologize for some screwy formatting in the tables - I couldn't figure out how to remove it.
I wanted to look at the breakdown of Rich Rodriguez's previous offenses, and in particular the main QB's run-pass balance and the fraction of runs by the QB. I'm only looking at RR in Div 1A (so Tulane OC, Clemson OC, WVU and Michigan), and I'm skipping the mess that was the 2008 offense. Data comes from the year-end statbooks for each team.
Here is the overall production chart. QB is the main QB (from what I could tell) - in 1999 Brandon Streeter got a lot of playing time (mostly passing), and in 2001 Rasheed Marshall got a decent amout of playing time.
Year
Team
QB
Pass Plays
Pass Yards
Rush Plays
Rush Yards
Total Plays
Total Offense
QB Pass
QB Pass Yardsards
QB Rushess
QB Rush Yards
Total QB Offense
1997
Tulane
Shaun King
366
2608
425
2012
791
4620
363
2577
124
511
3088
1998
Tulane
Shaun King
375
3615
520
2483
895
6098
364
3495
156
633
4128
1999
Clemson
Woody Dantzler
423
3019
497
1812
920
4831
201
1506
146
588
2094
2000
Clemson
Woody Dantzler
296
2311
557
2600
853
4911
212
1691
172
947
2638
2001
West Virginia
Brad Lewis
357
1811
475
1992
832
3803
237
1339
54
41
1380
2002
West Virginia
Rasheed Marshall
279
1753
714
3687
993
5440
259
1616
173
666
2282
2003
West Virginia
Rasheed Marshall
252
2034
600
2762
852
4796
215
1729
101
303
2032
2004
West Virginia
Rasheed Marshall
259
1993
590
3034
849
5027
171
1426
130
684
2110
2005
West Virginia
Pat White
193
1398
625
3269
818
4667
114
828
131
952
1780
2006
West Virginia
Pat White
233
2059
590
3939
823
5998
179
1655
165
1219
2874
2007
West Virginia
Pat White
265
2067
628
3864
893
5931
216
1724
197
1335
3059
2009
Michigan
Tate Forcier
329
2380
494
2234
823
4614
281
2050
118
240
2290
2010
Michigan
Denard Robinson
119
1203
228
1622
347
2825
96
1008
98
905
1913
2010 Proj
Michigan
Denard Robinson
285.6
2887.2
547.2
3892.8
832.8
6780
230.4
2419.2
235.2
2172
4591.2
Denard has already had more passing yards and almost as many rushing yards as 2005-era Pat White. If he averages just over 100 yards passing per game for the rest of the season he'll have more passing yards than any of RR's QBs other than Shaun King. If he kept on his current pace (unlikely), he'd end up with almost as many yards as 1997-era Shaun King. If he averages just over 60 yards rushing per game for the rest of the season he'll have more rushing yards than 2007 era Pat White. For total offense he would need to average just over 160 yards per game to best Pat White's best season, and just over 315 to match Shaun King. At this point it looks like Denard is the best all-around QB Rodriguez has had to date: almost as good a passer as King and as good/better a runner as Pat White.
Next I want to look at the breakdown of plays and yards between run and pass, and in particular the QB's share of production.
Year
Team
QB
% Rush Playsays
% Rush Yardsrds
% Runs QBB
% Rush Yards QBrds qB
% QB Total Playsl Plays
% QB Total Yardsl Off
QB Rush % of TotalTot %
QB Rush Yards % of Total
1997
Tulane
Shaun King
54%
44%
29%
25%
62%
67%
25%
17%
1998
Tulane
Shaun King
58%
41%
30%
25%
58%
68%
30%
15%
1999
Clemson
Woody Dantzler
54%
38%
29%
32%
38%
43%
42%
28%
2000
Clemson
Woody Dantzler
65%
53%
31%
36%
45%
54%
45%
36%
2001
West Virginia
Brad Lewis
57%
52%
11%
2%
35%
36%
19%
3%
2002
West Virginia
Rasheed Marshall
72%
68%
24%
18%
44%
42%
40%
29%
2003
West Virginia
Rasheed Marshall
70%
58%
17%
11%
37%
42%
32%
15%
2004
West Virginia
Rasheed Marshall
69%
60%
22%
23%
35%
42%
43%
32%
2005
West Virginia
Pat White
76%
70%
21%
29%
30%
38%
53%
53%
2006
West Virginia
Pat White
72%
66%
28%
31%
42%
48%
48%
42%
2007
West Virginia
Pat White
70%
65%
31%
35%
46%
52%
48%
44%
Low
54%
38%
11%
2%
30%
36%
19%
3%
Median
69%
58%
28%
25%
42%
43%
42%
29%
High
76%
70%
31%
36%
62%
68%
53%
53%
2009
Michigan
Tate Forcier
60%
48%
24%
11%
48%
50%
30%
10%
2010
Michigan
Denard Robinson
66%
57%
43%
56%
56%
68%
51%
47%
2010 Proj
Michigan
Denard Robinson
66%
57%
43%
56%
56%
68%
51%
47%
The first two data columns are the percent of all plays and all yards that come from all runs. The third and fourth are the percent of all runs and rush yards that come from the QB. The fifth and sixth are the percent of all plays and all yards that come from the QB. The seventh and eight are the percent of the QB's total plays and yards that come from his runs.
RR has historically varied a fair amount in how much of his offense comes from running the ball - this year we're about average for what he's done in the past, and less run-oriented that for example 2005 West Virginia. However, our rush offense is by far the most QB-based of any previous offense, far outstripping the one-man show of 2000 Woody Dantzler, and 2007 Pat White. If we look at total offense, this year's team is more QB-focused than any of the Clemson or WVU teams, but actually on par with the Tulane teams. Looking at Denard's run-pass balance he's actually right around Pat White's typical split, though he is certainly more run-focused in his production than any of RR's other quarterbacks.
This is just a high-level overview. I can't break down the kinds of running or passing plays RR is using from this data. The offense certainly feels very different than the Pat White-era WVU teams in formation and play style, and the YouTube highlights of Woody Dantzler I've seen have the QB iso type feel that we're seeing a lot from this year's team. I think the main message is that even within his system RR will adapt his style, both at a high level and at the formation/play level, to match his talent - which is what he should do.
Recent Comments
Oh, I absolutely agree that the secondary is quite bad, and I grant that the UFR is inexact. However, Misopogon's two benchmarks for the star ratings were a UFR-equivalent (which I was speaking too), and qualitative adjectives like "all Big Ten". The player analogues are probably the most useful in terms of benchmarking a player's ability, though I must admit my familiarity with historical players isn't good enough to assess their validity.
I guess the issue I was grappling with is that we have not just one weak player, but several, and the UFR statistic seemed the best way to think about how all the players might fit together. We all know that the d-line has been pretty good, the linebackers have been up and down, and the defensive backs have been pretty terrible. My point was not to dispute that, or to trash Misopogon's ratings (which I quite enjoyed reading) - I just wanted to suggest that the defensive backs are bad but not "commit a huge blunder every three or four plays" bad.
I think this is an interesting exercise, and I want to thank Misopogon for putting in the time to think through this. However, it seems that his ratings are too pessimistic (especially in the secondary), given the UFR-equivalent definition he chose for each star and Brian's defensive UFRs.
As a rough approximation I assumed an 80-20 playing time split between starters and backups, and I counted Ezeh as the starter. Using a 70-30 or 60-40 split wouldn't change things much. Based on the definition of -2 for a one star to +2 for a five star every four plays, we should expect the following totals for each unit:
DL: +0.8 per four plays
LB: -1.6 per four plays
DB: -5.4 per four plays.
Total: -6.2 per four plays
If we add up Brian's UFR ratings, we get the following:
Because Brian hasn't done the Iowa game yet, the first total ignores it, and the second total assumes it'll be the same as the MSU rating (imperfect, but the best I could do). Our defense has faced 496 plays (436 if you don't count Iowa).
As you can see, our DL has performed in line with your ratings, but our LBs and DBs have done much better. The DB rating is clearly an order of magnitude too negative. Averaging -5.4 per four plays means that in a 60 play game they should have a total UFR of -81.... four times worse than the Chappel-bombing.
Now, we could instead ask whether the stars are fine in a relative sense, but that the pluses and minuses associated are too extreme. I think that is part of it, but on the other hand according to Brian's UFRs the LBs have been a net positive, as has the team, and the DBs have been half as negative as the DLs have been positive. By contrast in Misopogon's ratings the LBs are rated as negatives, and the DBs are rated as almost 7 times more negative than the DLs are positive. Again, this suggests that the linebackers and (especially) the defensive backs aren't as bad as Misopogon's ratings would suggest.
Fremau predicts Michigan by 8:
71.9% PWE; MICHIGAN 30, Michigan State 22
http://bcftoys.blogspot.com/2010/10/fei-forecasts-week-6-oct-5-oct-9.ht…
I just listened to Brian's section of the podcast, and he makes an excellent point about the "Greg Jones as a spy" defense. On a running play, the MLB is always going to have a blocker assigned to him, and is unlikely to be the read key on an option - so it's not clear that having Jones "spy" will do much of anything. As Brian said "he'll just be playing run defense". I think spying the QB would matter more if Denard was scrambling out of the pocket on passing plays, or if we were running lots of true QB draws where we pass block for a few seconds at the start of the play. It also seems like Jones spying should open up the underneath coverage on play-action passes.
Really interesting, thanks!
I think it'll be closer to 250 passing - 150 running. I'm guessing that our run success will be more balanced between Denard and the RBs.
I do like FO's stats, I was just making the point that S&P+ is a derived statistic based on a model of what's important, not a raw statistic. I think both are important to look at.
BTW, FO breaks down the Michigan-Michigan State game here:http://footballoutsiders.com/7th-day-adventure/2010/sda-defending-home-…
Has the Ball...
Has the Ball...
(5-0)
(5-0)
Off
Def
Def
Off
(Std. Downs)
(41st)
(10th)
(Pass. Downs)
(46th)
(47th)
And based on the F/+ ratings they pick Michigan (against the spread).
Recently Wisconsin has tended to do worse than expected when initially ranked highly, and better than expected when initially ranked low. They'll probably end up in the middle of the pack in the Big 10 this year.
OK, it looks like the Cal stats really are 504 rush yards total - that's what Cal's official webpage has).
That means it should be
Michigan: was 125.8/game (#37), now is 139.4/game (#36)
Tulane: was 119.75/game (#33), now is 142.25/game (#38)
Cal: was 126/game (#38), now is 144.5/game (#41)
Therefore we improve by one rank instead of two.
You have to move the sack yards from rush defense to pass defense for everyone before reranking all the teams. We move ahead two spots because two teams near us (Cal and Tulane) have more sack yards per game thank we do, so their rush defense gets worse by a larger amount than ours.
Specifically, we have 13.6 sack yards per game. Cal has 18.5 and Tulane has 18. That means for rushing defense we get...
Michigan: was 125.8/game (#38), now is 139.4/game (#36)
Tulane: was 119.75/game (#33), now is 142.25/game (#38)
Cal: was 125.5/game (#37), now is 144/game (#40)
BTW, I'm using stats from here: http://www.cfbstats.com/2010/leader/national/team/defense/split01/categ…
I'm not sure why there is a slight difference from the NCAA.com stats you link to. It matters for Cal, since your link has them at 126 rush yards/game, moving them down to #38. If we take the NCAA.com stats then the raw numbers likely won't change much but we'll only move up one rank instead of two (since we'd start one rank higher).
The Football Outsiders stats are interesting, but do a lot different than just control for sacks and SoS.
If you just want to move sack yards from rush defense to pass defense I find that it makes no difference for our passing defense rank, and actually improves our rushing defense rank:
Pass Defense: was 307.8/game (#120), now is 294.2/game (#120)
Rush Defense: was 125.8/game (#38), now is 139.4/game (#36)
Brian Fremau talks about some of these same issues in his Football Outsiders column this week (http://footballoutsiders.com/fei-ratings/2010/fei-better-late-never). He handles things a little differently, since he strips out "garbage time" posessions and he ignores games with FCS schools. He only lists a couple of top 10 tables, but he lists Michigan as 9th in terms of points per possession on offense with 3.386. Top 10 defenses allow somwhere in the 0.5 to 1.3 points per posession range, so we're letting teams score twice as often as that. He also shows that top 10 defenses are able to make a stop for a drive that has already gone 10+ yards at least two-thirds of the time, where we're only doing it half the time (see below).
Overall he's got us ranked 16th with a rating of 0.142 (I don't have a good feel for the units on his metric, but the #1 team is 0.297 and the #25 team is 0.122. projects us to win on average 4.4 of our remaining games (so end with a total of 9.4 wins). For comparison Michigan State is ranked 31st with a rating of 0.95. He doesn't have a prediction yet (will probably come out later today), but for reference last week he had MSU at 0.73 and Wisconsin at 0.69 and gave MSU a 60% chance to win. He had Iowa at 0.148 and Penn State at 0.102 and gave Iowa a 71% chance to win. So I'm guessing we'll be like 65-70% favorites in his system.
I definitely think it would be interesting to break drives down by number of plays rather than yardage - I used yardage because that was the information bigmc6000 had in original table. I do think the yardage breakdown is useful if you think about three groups - immediate stops, short drives and long drives. About a third of the drives against our defense stop in 10 yards or less (i.e. one set of downs), about a third go for a while and then stop, and a third drive for 50+ yards and mostly score. Field position certainly plays into this, but again wasn't in the data yet.
I think the other thing that you can tell from yards is that when a drive has gone a long way, we basically need a turnover to stop it. We're not very good at just forcing a stop when the other team has already gotten deep in our half of the field. You see that clearly in the second table, which looks cumulatively rather than by total drive yardage (so avoids the problem of categories with few observations). The longer a drive has already gone, the more we need a turnover to stop it.
I would definitely be in favor of someplace where basic scraped data can be centrally located. bigmc6000 posting the individual drives in addition to his analysis made it easy for me to add my two cents. It would be very useful, for example, to have the basic play-by-play numbers from the UFRs in a spreadsheet. On the other hand people like Mathlete who do a lot to transform the raw data into something else like PAN shouldn't feel like they should have to share that (though if he wants to that'd be great!).
Oops, I left in the EOH drives for the total drives, but not for stops. Here are the corrected tables. I also switched the Indiana EOH touchdown to be a scoring drive.
Sure, go ahead. The other thing I was hoping to pull from the UFRs was whether on defensive stops we got other kinds of "big plays" in the last set of downs - sacks or TFLs.
Thanks for doing this - I wanted to go through the UFR's to look at what our defensive stops looked like, but I can use your data to do most of what I had in mind.
I took a look at all the drives that we got stops on (other than EOH drives), breaking them down by how long the drive was. I think this will show how much we're bending, and if we bend do we break.
So, we actually get a fair number of quick stops. However, the longer the drive is, the more dependent we are on getting a stop via a turnover (interception, fumble or missed field goal) rather than just stiffening.
Another way to look at that is to ask, given that a drive has already gone X yards, what is the probability of a stop and/or a turnover?
Not surprisingly, the longer the drive goes, the less likely we are to get a stop, and the more likely we are to need a turnover to get a stop. With Brian now including the number of rushers in UFR it would be interesting to see whether we're more likely to blitz the longer a drive has gone in order to get a TO.
Do your stats for tackles include special teams tackles? This list of tackling stats (http://www.cfbstats.com/2010/team/418/tackle/index.html) appears to, since John McColgan (the FB) is listed as the #16 tackler.
So things actually look a bit better when I redo the analysis only for AQ conference teams.
So, just based on historical averages, we have something like a 30% chance to improve enough to end up 70th or better in total defense, and something like a 40% chance to improve by at least 15 ranks.
I left the spreadsheet that had all the data at work, so I won't be able to check until tomorrow. I do have a list I was working on of all the teams that increased by at least 40 ranks. Of the 16 teams that did so, 8 were from AQ conferences.
In 2009, Syracuse went from 96 to 37, and Colorado went from 101 to 57.
In 2008, UCLA went from 102 to 47, and Purdue went from 109 to 61.
In 2006, Boston College went from 92 to 34.
In 2005, Oregon went from 90 to 44, LSU went from 102 to 3, and Arkansas went from 103 to 34.
I took a look at how often teams with terrible defenses early in the season improve to be somewhat better than terrible. It was easiest to compare Aug/Sept rankings to final rankings, so I looked at how many teams ranked 90th or below in total defense at the end of Sept improved to be at least 70th or above by the end of the year. Michigan was 92nd at the end of Sept (and is currently 102).
2004: 14%
2005: 23%
2006: 20%
2007: 17%
2008: 13%
2009: 16%
If instead you look at how many of the teams 90th and below after September improve at least 15 ranks by the end of the year, you get the following:
2004: 34%
2005: 44%
2006: 23%
2007: 30%
2008: 26%
2009: 32%
So, there's a small chance we can get "slightly below average" or better, and a reasonable chance we can improve somewhat.
I posted a more detailed breakdown of RR's teams below, but Denard has already gotten in 5 games as good or better than what 2005 Pat White did for the whole season.
Denard has 1008 passing yards on 96 attempts (10.50 YPA) compared to Pat White's 828 yards on 114 attempts (7.26 YPA).
Denard has 905 rushing yards on 98 attempts (9.23 YPC) compared to Pat White's 952 yards on 131 attempts (7.27 YPC).
For total offense Denard has 1913 yards on 194 plays (9.86 YPP) compared to Pat White's 1780 yards on 245 plays (7.27 YPP).
I just registered to post this (after lurking for a while), so I'm adding it as a comment rather than starting a new thread/diary. I apologize for some screwy formatting in the tables - I couldn't figure out how to remove it.
I wanted to look at the breakdown of Rich Rodriguez's previous offenses, and in particular the main QB's run-pass balance and the fraction of runs by the QB. I'm only looking at RR in Div 1A (so Tulane OC, Clemson OC, WVU and Michigan), and I'm skipping the mess that was the 2008 offense. Data comes from the year-end statbooks for each team.
Here is the overall production chart. QB is the main QB (from what I could tell) - in 1999 Brandon Streeter got a lot of playing time (mostly passing), and in 2001 Rasheed Marshall got a decent amout of playing time.
Denard has already had more passing yards and almost as many rushing yards as 2005-era Pat White. If he averages just over 100 yards passing per game for the rest of the season he'll have more passing yards than any of RR's QBs other than Shaun King. If he kept on his current pace (unlikely), he'd end up with almost as many yards as 1997-era Shaun King. If he averages just over 60 yards rushing per game for the rest of the season he'll have more rushing yards than 2007 era Pat White. For total offense he would need to average just over 160 yards per game to best Pat White's best season, and just over 315 to match Shaun King. At this point it looks like Denard is the best all-around QB Rodriguez has had to date: almost as good a passer as King and as good/better a runner as Pat White.
Next I want to look at the breakdown of plays and yards between run and pass, and in particular the QB's share of production.
The first two data columns are the percent of all plays and all yards that come from all runs. The third and fourth are the percent of all runs and rush yards that come from the QB. The fifth and sixth are the percent of all plays and all yards that come from the QB. The seventh and eight are the percent of the QB's total plays and yards that come from his runs.
RR has historically varied a fair amount in how much of his offense comes from running the ball - this year we're about average for what he's done in the past, and less run-oriented that for example 2005 West Virginia. However, our rush offense is by far the most QB-based of any previous offense, far outstripping the one-man show of 2000 Woody Dantzler, and 2007 Pat White. If we look at total offense, this year's team is more QB-focused than any of the Clemson or WVU teams, but actually on par with the Tulane teams. Looking at Denard's run-pass balance he's actually right around Pat White's typical split, though he is certainly more run-focused in his production than any of RR's other quarterbacks.
This is just a high-level overview. I can't break down the kinds of running or passing plays RR is using from this data. The offense certainly feels very different than the Pat White-era WVU teams in formation and play style, and the YouTube highlights of Woody Dantzler I've seen have the QB iso type feel that we're seeing a lot from this year's team. I think the main message is that even within his system RR will adapt his style, both at a high level and at the formation/play level, to match his talent - which is what he should do.