Moving the (Stati)Sticks: Week Ten
[Barron/MGoBlog]
Football’s what you make of it. If you want to watch for the big hit or the big pass, you can. If you want to watch because you’re hoping to get a glimpse of how pattern matching works on switch routes, you can. It’s a game of nuance that can be enjoyed without; you can guess which direction I lean considering this is a column predicated on the usual stats not providing enough information. Rarely, though, you see the same thing no matter what you intended to look for; sometimes watching a team get punched in the face is also a statistical drubbing regardless of the set of stats you use.
The Mathlete’s Four Factors:
Once again, a quick reminder of what the factors mean:
Conversion rate = [1st Downs gained]/[1st Down plays (including first play of drive)]. A three and out is 0/1. A one play touchdown is 1/1. Two first downs and then a stop is 2/3, etc.
Bonus Yards = [Yards gained beyond the first down line]/[Total plays from scrimmage]
This is an adjustment to how I have previously calculated, to account for the plays a team runs.
Field Position = Expected team points based on starting field position. This accounts for all elements of field position: turnovers, special teams, drive penetration etc.
Red Zone: Points per red zone trip (TD’s counted as 7 regardless of PAT)
Offense:
| Field Pos. | Conv. Rate | Bonus YPP | Red Zone | |
| Week 1 | 21.0 | 73 | 1.52 | 5.7 |
| Rank | 60 | 30 | 59 | 27 |
| Week 2 | 25.0 | 68 | 1.63 | 5.8 |
| Rank | 77 | 70 | 95 | 30 |
| Week 3 | 25.3 | 70 | 2.56 | 5.8 |
| Rank (B1G Rk) | 84 (12) | 54 (6) | 43 (6) | 30 (5) |
| Week 4 | 24.5 | 72 | 2.92 | 5.8 |
| Rank (B1G Rk) | 91 (14) | 30 (4) | 31 (3) | 27 (4) |
| Week 5 | 26.9 | 67 | 2.78 | 5.8 |
| Rank (B1G Rk) | 59 (10) | 54 (3) | 39 (3) | 20 (2) |
| Week 6 | 28.2 | 68 | 2.93 | 5.8 |
| Rank (B1G Rk) | 41 (4) | 39 (1) | 25 (2) | 21 (2) |
| Week 7 | 28.4 | 65 | 2.75 | 5.5 |
| Rank (B1G Rk) | 36 (4) | 61 (5) | 36 (3) | 31 (2) |
| Week 9 | 27.6 | 65 | 2.59 | 6.1 |
| Rank (B1G Rk) | 45 (6) | 51 (3) | 44 (6) | 3 (1) |
| Week 10 | 27.6 | 65 | 2.64 | 6.0 |
| Rank (B1G Rk) | 46 (6) | 35 (2) | 43 (5) | 4 (1) |
Defense:
| Field Pos. | Conv. Rate | Bonus YPP | Red Zone | |
| Week 1 | 27.9 | 73 | 1.64 | 5.7 |
| Rank | 47 | 44 | 20 | 30 |
| Week 2 | 25.1 | 67 | 1.60 | 6.1 |
| Rank | 51 | 58 | 23 | 88 |
| Week 3 | 24.0 | 63 | 1.28 | 6.1 |
| Rank (B1G Rk) | 35 (4) | 38 (6) | 9 (3) | 100 (13) |
| Week 4 | 23.1 | 59 | 1.23 | 6.1 |
| Rank (B1G Rk) | 29 (5) | 17 (4) | 4 (1) | 110 (13) |
| Week 5 | 24.5 | 55 | 1.10 | 6.1 |
| Rank (B1G Rk) | 32 (4) | 7 (2) | 1 (1) | 115 (13) |
| Week 6 | 23.6 | 54 | 1.01 | 6.1 |
| Rank (B1G Rk) | 21 (4) | 6 (1) | 1 (1) | 115 (12) |
| Week 7 | 23.4 | 57 | 1.24 | 6.4 |
| Rank (B1G Rk) | 20 (3) | 6 (1) | 1 (1) | 124 (13) |
| Week 9 | 23.4 | 60 | 1.54 | 4.6 |
| Rank (B1G Rk) | 17 (3) | 7 (2) | 6 (2) | 42 (6) |
| Week 10 | 24.1 | 59 | 1.61 | 4.1 |
| Rank (B1G Rk) | 22 (4) | 6 (1) | 5 (2) | 13 (2) |
[After the JUMP: tables, attempted explanations, and scout-by-number]
The offensive stats again pass the eye test. Michigan’s Field Position and Conversion Rate haven’t changed much in the last few weeks; having a similar Conversion Rate over multiple games is hinting at Michigan’s drives being of similar length, and if that continues again this week then I’ll probably dig into it a bit more in next week’s post. Either way, a 65% Conversion Rate puts them close to the top 20% of offenses in the conference. Bonus YPP also took a little jump, and the first thing that comes to mind is the fullback wheel route; imagine if they’d hit on the play called for intent to deceive. Still, being in the top third of the conference with an offense that hasn’t looked explosive (and that FEI says isn’t) is either impressive or really, really BIG TENNNNNNNN, depending on your perspective.
Defensively, the difference in Field Pos. and average points given up is really impressive. Michigan’s expected to give up 24.1 points per game based on the expected point values of opposing offense’s drives, but they’re actually giving up a paltry 11.9 points per game, which ranks first nationally. Red Zone scoring has made a drastic jump since Michigan picked up its new habit of allowing people inside the 20 and then either stopping them at the goal line or holding them to a field goal after giving up a 54-yard run and teasing them by letting them run six more plays and get to the five yard line. The aforementioned long run might also have something to do with the small uptick in Bonus YPP.
Advanced Box Score
I’ve used this space the last few weeks to point out how lopsided things have been in favor of Michigan’s opponents and imply that Michigan’s been a bit fortunate to have a win from those two weeks, and though the stats from this game are also skewed to one side they’re authoritatively in Michigan’s favor.
Michigan had one more drive than Rutgers (13 to 12), and with it they managed to run eight more plays than Rutgers (73 to 65). Those are the only two categories that were close.
Michigan averaged 3.07 yards per play more than Rutgers (7.16 to 4.09). Seriously, read that again. Per play! That’s mind boggling, and the mind boggliness continues when we look at Michigan’s nine scoring opportunities and compare them to Rutgers five and consider that Michigan averaged 5.33 pts/opp to Rutgers’ 1.8 pts/opp. Rutgers was fortunate to have that many scoring opportunities considering some of the down-and-distance situations they were in; Michigan’s Success Rate was 42% while Rutgers sputtered along at 23%. Oh, and Michigan’s average starting field position was the 36.8 while Rutgers was, on average, starting from their 28.1.
Five Factors
Despite dominating Rutgers, Michigan’s defensive Five Factors barely budged. The big run they gave up certainly didn’t help; they dropped to 65th in Explosiveness, down one spot from the week prior. Michigan’s Success Rate is still excellent at 29.9%, which is ranked third overall. Allowing 3.24 points per trip inside the 40 is the fifth lowest in the country; helping that a bit is opponent’s average starting field position of the 26.2 yard line, which is 10th overall.
There was a lot more movement on the other side of the ball, as the offense I described last week as “middling and not great” improved in almost every S&P+ and FEI category. They’re ranked 48th in Explosiveness (a 24 spot jump), 38th in Efficiency (a nine spot jump), and 46th in Finishing Drives (a seven spot jump, while their average points per trip inside the 40 rose 0.07 to 5.05).
FEI also showed offensive improvement. 76.6% of drives contain at least one first down; that’s a 1.6% increase from a week ago, which resulted in a 12-spot jump in the national rankings. One of Michigan’s bigger gains came in available yardage, which jumped 3.4% to 49.8% (36th overall). A game against Rutgers is fodder for rankings jumps (even when things are opponent adjusted)3, but things weren’t exactly perfect. Michigan’s 82nd in explosive drives (to be fair, this is a 14 spot jump from the week before), 102nd in methodical drives, and 49th in value drives. The methodical drives (drives of more than 10 plays) ranking looks bad at first brush, but keep in mind Michigan’s excellent average starting field position and recent semi-explosiveness.
Opponent’s Run Game: Overall rushing S&P+ took a one-spot dip to third overall, due in part to ceding 128 yards on the ground. Fifty-four of those yards came on a play where it looked like Bolden was held; the run went through his lane, so if he wasn’t ripped out of the hole you can guess what would’ve happened. Rushing Success Rate ranks a still-impressive third, with the real problem, as it was last week, Rushing IsoPPP. The aforementioned third-ranked Success Rate is 27.3%, so teams aren’t able to run well consistently. (Contrast that with the national average, which is 42.1%.) Michigan’s also again ranked fifth in Opportunity Rate after Rutgers; Opportunity Rate is a measure of how often the offensive line produces at least five yards for the runner, and that’s only happening on 28.3% of carries. Michigan gives up a big run fairly regularly, it dings the advanced stats, and opponents otherwise have trouble moving the ball via the run.
Michigan’s Run Game: Things really aren’t much different than they’ve been for the last few weeks. Overall rushing S&P+ and Success Rate are pretty good (36th and 43rd nationally). IsoPPP, Opportunity Rate, and Adjusted Line Yards aren’t very good (77th, 104th, and 69th, respectively). The offensive line isn’t getting a ton of push, and the backs don’t create much for themselves (hence the low IsoPPP). That’s not an indicator that the line is bad; their Stuff Rate (plays where the runner is tackled in the backfield) is only 17.6%, which ranks 43rd.
Opponent’s Pass Game: Michigan sank to 30th in passing S&P+ after Minnesota, but the ol’ Rutgers Rebound bounced them up to 13th. Thanks, Delany! Michigan’s 10th in passing Success Rate and 25th in IsoPPP; the latter number also rebounded by about 20 spots because Rutgers wasn’t able to hit anything over the top.
Michigan’s Pass Game: Putting a secondary of all freshmen out against a fifth-year senior who’s getting comfortable with the playbook should end in carnage, and it did. Michigan moved up in all the passing categories Bill Connelly tracks. The most impressive jump was in IsoPPP (that’s the explosiveness measure), where Michigan went from 72nd to 38th. These stats are all opponent-adjusted, so it’s fair to say that Michigan performed really well even if Mortal Kombat was your baseline expectation.
Looking Forward, Looking Back:
| Opponent | Off. S&P+ | Def. S&P+ | Overall S&P+ |
| @ Utah | 42 (+1) | 19 (+2) | 24 (nc) |
| Oregon State | 109 (+1) | 100 (-8) | 106 (-2) |
| UNLV | 100 (+4) | 98 (-5) | 99 (-2) |
| BYU | 32 (-3) | 44 (+4) | 33 (+5) |
| @Maryland | 96 (+4) | 40 (+18) | 72 (+8) |
| Northwestern | 107 (+4) | 6 (+1) | 47 (+2) |
| Michigan State | 28 (-1) | 45 (-9) | 29 (-6) |
| @ Minnesota | 89 (-2) | 26 (-2) | 58 (-2) |
| Rutgers | 88 (-8) | 117 (-1) | 105 (nc) |
| @ Indiana | 21 (-1) | 110 (nc) | 67 (+1) |
| @ Penn State | 62 (-3) | 16 (nc) | 31 (-2) |
| Ohio State | 17 (-3) | 9 (-3) | 4 (nc) |
| Michigan | 43 (+2) | 1 (nc) | 3 (nc) |
| Opponent | Off. FEI | Def. FEI | Overall FEI (includes Special Teams FEI) |
| @ Utah | 47 (-8) | 10 (+3) | 11 (+3) |
| Oregon State | 109 (nc) | 84 (-3) | 113 (-5) |
| UNLV | 87 (+12) | 108 (+4) | 97 (+2) |
| BYU | 32 (-3) | 53 (+8) | 34 (-4) |
| @Maryland | 84 (+6) | 71 (+7) | 81 (+6) |
| Northwestern | 88 (nc) | 9 (+3) | 36 (nc) |
| Michigan State | 13 (-1) | 42 (-11) | 7 (nc) |
| @ Minnesota | 93 (+2) | 26 (+1) | 76 (-4) |
| Rutgers | 67 (+3) | 114 (-8) | 107 (-9) |
| @ Indiana | 29 (+4) | 99 (-8) | 60 (nc) |
| @ Penn State | 77 (+4) | 13 (-4) | 45 (+5) |
| Ohio State | 28 (-3) | 12 (+3) | 8 (nc) |
| Michigan | 41 (+17) | 5 (-3) | 12 (-2) |
What about Saturday?
Indiana’s 21s in offensive S&P+ and 110th defensively, so balance isn’t exactly their strong suit. Their run game is okay but not great; they aren’t likely to break a big play (82nd in IsoPPP), and they’re just a smidge above average with a 42.8% Success Rate (62nd). Continuing the theme of poor balance, Indiana’s passing game is actually really good. They’re 16th in passing S&P+, and the lowest value listed is passing IsoPPP, which is still 54th nationally.
Defensively, they’re not great versus the run or pass, but they’re ranked higher (72nd) against the pass. Despite that, they’re 110th in passing IsoPPP, so explosive plays should be there. Those explosive plays should also be there against the 95th-ranked rushing IsoPPP team. Indiana’s ranked 104th in Power Success Rate (% of runs on third or fourth down with two or less to go that result in a first down or TD), so Michigan should be able to move the ball in both short-yardage situations and in big chunks.
November 13th, 2015 at 10:13 AM ^
So, Indiana is like BYU offensively from a strength standpoint (good because we handled them), and overall they are a better Minnesota (bad because we derped them).
This game makes me nervous. If the defense derps again, we get into the shootout mode. Our offense isn't designed for this kind of game.
Just jump on Indiana fast and keep your foot on their throats... please?
November 13th, 2015 at 10:41 AM ^
Indiana generates their offense from their QB-RB tandem in a classic spread; their receivers are meh. In terms of style they're anything but BYU's glory-or-death. I have no idea how they wind up looking like BYU on the stat sheet but you certainly won't defend them the same way.
Also, a close game actually favors Michigan. Harbaugh has shown he can RPS for TDs when the offense really needs it, and Indiana has trouble finishing. I don't think it's adjustments because their defense -- as bad as it is -- has almost completely shut down opponents in the 3rd quarter. Adjustments are fine. But they've been vastly outscored in the 4th quarter this year, fading in a number of close games, so I suspect conditioning. They're a tempo team that lacks the endurance for tempo.
The concern for Michigan is what kind of day they'll have. A very gross oversimplification of Michigan's season is that they either control the game or don't. Meaning, there aren't big momentum swings, like blowing big leads or rallying from big deficits. They either dominate all game or struggle all game. This is not a bad thing overall -- Michigan simply hasn't been overmatched by an opponent all season -- but if they do struggle, even with IU's conditioning issues they could turn this into a toss-up. In "struggle" days, Michigan is 1-2 with each game decided in the final minute.
November 13th, 2015 at 11:18 AM ^
IU is better than BYU offensively. Their running game is definitely better, and their QB is much more experienced. IU's statistics are weighed down by that stretch where Sudfeld was injured, and their tailback has also battled injuries (both were out for their game at PSU). We're going to get their best shot, with both reasonably healthy tomorrow.
November 13th, 2015 at 10:29 AM ^
offensive rankings and then recalling how they moved the ball on us makes me want to blow all the chunks. That said, as long as Sudfeld doesn't go all SUPERMAN on us, we'll be fine.
November 13th, 2015 at 11:06 AM ^
I think it would be helpful to break out IU's data from when they had Sudfeld/Howard healthy vs. not.
November 13th, 2015 at 11:13 AM ^
Hoosier? Kill it before it multiplies.
Go Blue!
November 13th, 2015 at 12:57 PM ^
I've been looking at this post every week throughout the season, trying to figure it out, and I still haven't (completely). IMO this data representation needs some tweaking. I'm not trying to be overly critical. The Mathlete's work is among my favorite on this board, but I'm just trying to make sense of what's being presented:
Taking the Offense Four Factors table as a basis for these questions -
1 - (Minor format quibble) Why are the Four Factors table columns out of order with the column descriptions?
2 - Ok, so from the column description, "Conversion Rate" is a fraction. Then why isn't it a fraction in the table? Why are they large whole numbers? Don't understand. (OMG - And now, week 10, I finally see from the data description commentary under the table that these are intended to be %'s. Face palm.)
3 - There is no description of what "rank" means. Does low mean good or does high mean good? Similar question for the BIG rank. I assume low means good, but I have no idea how to confirm this from the data. (And now that I've looked at the "Red Zone" column I'm 99% sure low is good, but it took reading through 4 columns before figuring it out for sure.)
4 - Is all the column data and ranks cumulative from week to week or is it based on that week's data only?
5 - I think I get "Bonus Yards". To confirm, does the data mean that we were "most explosive" against Northwestern (week 6) and "least explosrive" against Utah (week 1)?
6 - I think I get "Field Position". To confirm, does the data mean we were expected to get the most points against Michigan State (week 7)?
OVERALL SUGGESTIONS:
- Put columns in order of column descriptions
- Label the columns with units (that would've saved me 10 weeks of not comprehending the Conv. rate numbers)
- Shade the data blue for "good/better than avg" data and red for "bad/worse than avg" data - sort of like Ecky Pting's awesome "Adv. Stats Schedule Rundown" diary.
- Clarify that the data is individual week data only (if this is indeed the case)
Thanks.
November 13th, 2015 at 11:46 PM ^
Thanks for the feedback, I appreciate it. I think you've brought up some very valid points; expect to see some tweaks next week. The biggest thing I need to emphasize from your list is that The Mathlete's four factors are cumulative. I include every week in the table because I thought it'd be interesting to see how things change as the season moves along.
I have some ideas for how to revamp this post in the offseason; finding a good way to visualize the data with charts and tables is my priority. I'll try and make some charts for The Mathlete's data in the next few weeks, but a lot of what I want to do with the data from Connelly's statistical profiles is lost because it updates every week, so unless I wrote about a certain stat the week before I can't necessarily track it down. I guess what I'm saying is thanks for reading, look for some clarification, and expect some cool stuff further down the road.
November 13th, 2015 at 3:09 PM ^
There is a ton of work in each week's update. It takes a day or two just to digest, and one of the many great things bout MGoBlog. Mucho appreciado.
Comments