OR "Statistics That Will Help Bring You Down from the Post-Bowl-Game-Clinching-Win High"
Boiled down to its basics, playing offense successfully in the sport of football is pretty simple:
- Gain 10 yards in 3 tries.
- Repeat until you reach the the end zone.
. . . At the hypothetical extreme, a team that could gain exactly 3.5 yards on every play could achieve that goal. Alternately, two rushing plays of 5 yards each or a single pass completion in three attempts for 10 yards does the trick.
The goal of any football defense, then, has to be to disrupt the opposing offense's rhythm, with the goal of creating a situation where picking up 3-5 yards per play isn't enough to keep a drive going.
In an attempt to quantify the success (or lack thereof) of the Michigan State defense in disrupting opposing offenses, we here at TOC have developed a new-fangled metric we call "disruption percentage." The percentage is calculated by summing the five defensive statistics below (presented in ascending order of disruptiveness) and dividing the total by the number of offensive plays run by the opposition:
- Quarterback hurries: Increasing the odds the offense will throw an incomplete pass and lose an opportunity to gain yardage toward a first down
- Pass break-ups: Ending an attempted pass play and reducing the number of downs the opposition has to pick up a first down
- Tackles for loss: Increasing the yardage the offense has to gain to achieve a first down by sacking the quarterback or stopping a running play for a loss
- Forced fumbles: Creating the opportunity to recover a fumble and end the opponent's offensive possession
- Interceptions: Definitively ending the opponent's offensive possession
(Notes: (a) If specific stats on penalties by the opposing offense--holding, etc.--were readily available, we'd include those, as well. (b) We define "forced fumble" liberally, counting any fumble by the offense, even if it's not directly attributable to a defensive player. (c) Initially, we didn't include pass break-ups, but we've added them. That helps eliminate a bias toward blitzing; if you can rush 3 and drop 8 effectively, that should show up in pass break-ups. (d) There's some double-counting going on, as a QB hurry can lead to an interception, etc.)
Here are the updated game-by-game numbers for MSU:
| Opponent | QBH | PBR | TFL | FF | INT | TOTAL | Plays | Disrupt% |
|---|---|---|---|---|---|---|---|---|
| Montana St | 5 | 2 | 4 | 0 | 0 | 11 | 54 | 20.4 |
| Central Mich | 0 | 4 | 6 | 0 | 1 | 11 | 76 | 14.5 |
| Notre Dame | 1 | 3 | 4 | 1 | 1 | 10 | 71 | 14.1 |
| Wisconsin | 4 | 2 | 3 | 1 | 0 | 10 | 81 | 12.3 |
| Michigan | 1 | 3 | 10 | 3 | 1 | 18 | 60 | 30.0 |
| Illinois | 6 | 3 | 8 | 0 | 1 | 18 | 60 | 30.0 |
| Northwestern | 2 | 2 | 6 | 3 | 0 | 13 | 76 | 17.1 |
| Iowa | 0 | 4 | 5 | 0 | 0 | 9 | 66 | 13.6 |
| Minnesota | 0 | 1 | 5 | 1 | 1 | 8 | 72 | 11.1 |
| Western | 5 | 4 | 4 | 1 | 0 | 14 | 55 | 25.5 |
| Purdue | 0 | 3 | 5 | 2 | 0 | 10 | 92 | 10.9 |
| TOTAL | 24 | 31 | 60 | 12 | 5 | 132 | 763 | 17.3 |
The first time we went through this exercise was after the Northwestern game. At that point, we were hopeful that the back-to-back 30% performances, followed by the decent 17.1% against Northwestern's short-pass-based offense portended good things for the defense. It looked like maybe the new starters were gelling in the base 4-3 defense and our remaining opponents would suffer because of it.
Then we played two more games. Neither of which turned out very well.
LVS, during the week after the Minnesota game:
It probably comes as very little surprise to MSU fans that two of the three least disruptive defensive performances have come in the past two weeks. Iowa was somewhat predictable, as that's a team that will, ideally, play a conservative offensive game because their defense is so good. Of course, that's not exactly how things have worked out this season for them: one or two of the third quarter STANZIBALLS against Indiana would have been, uh, enormously useful in our game. Our complete failure to hurry the quarterback against Iowa probably cost us our chance at one or two easy interceptions. And the Minnesota game . . . well, we all knew that Saturday night was our worst defensive performance of the season, and these statistics certainly buttress that feeling.
Now we've played two more games. The defense's performance against Western was certainly acceptable (particularly given that the game had basically been decided before halftime).
The performance against Purdue? Not so much. While the disruption percentage was fractionally better than the season-low showing against Minnesota, I'm going to say the defensive performance against the Boilermakers was the worst of the season. Two reasons:
- The Gophers made several spectacular passes/receptions against us on their way to 42 points. Purdue, on the other hand, just seemed to take the easy yards available to them all afternoon, without having to doing anything of particular offensive brilliance. (In the words of Mark Dantonio: "Today was a day of big plays for us. We made a play at the beginning of the game on defense, one at the end of the game on defense, and somewhere in there, there were 90 other plays.")
- Four of the five tackles for loss MSU posted against Purdue were on running plays. So in terms of the pass defense, which has been the major concern all year long, the defense made only 4 disruptive plays on 55 Purdue passing attempts.
(We'll be able to say more about the exact nature of the miserable defensive performance against the Boilermakers when DrDetroit finishes his MSUFR later this week.)
The good news is that MSU got its first win of the season with a sub-15 disruption percentage. The bad news is that, going into the two remaining games against Penn State and a yet-to-be-determined bowl opponent that will probably have a decent offense, the Spartan defense appears to be getting less-and-less disruptive. The performances against Michigan and Illinois--two teams now at the bottom of the conference standings--clearly have to be classified as outliers. And the idea that sticking with your base defense in 90% of situations so that the players become a well-oiled machine as the season progresses doesn't carry much water any more.
Instead, predictable defensive schemes have lead to quotes like the ones below from opposing players/coaches.
We were just hoping that they were lined up in that same defense, and sure enough, first play walking up to the line, they’re in the defense we want them to be.
We had a great game plan. They have a very vanilla defensive formation. We knew what were up against and what we had to do. We just had to execute and I felt like we really did.
Just to make sure that I'm not mixcontextualizing (word?) the MSU disruption percentage data, I went ahead and ran the numbers for the other ten Big Ten teams.
| Team | QBH | PBR | TFL | FF | INT | TOTAL | Plays | Disrupt% | Yds/G | Yds/G Rk |
|---|---|---|---|---|---|---|---|---|---|---|
| Iowa | 18 | 56 | 61 | 17 | 19 | 171 | 716 | 23.9 | 294.5 | 3 |
| Wisconsin | 18 | 29 | 73 | 13 | 13 | 146 | 612 | 23.9 | 307.4 | 4 |
| Penn State | 6 | 43 | 91 | 9 | 11 | 160 | 704 | 22.7 | 272.0 | 2 |
| Ohio State | 10 | 30 | 65 | 20 | 19 | 144 | 701 | 20.5 | 258.3 | 1 |
| Illinois | 15 | 40 | 56 | 16 | 4 | 131 | 685 | 19.1 | 400.3 | 10 |
| Northwestern | 19 | 22 | 56 | 22 | 14 | 133 | 705 | 18.9 | 345.2 | 5 |
| Minnesota | 4 | 49 | 61 | 23 | 9 | 146 | 784 | 18.6 | 381.8 | 8 |
| Michigan | 7 | 31 | 77 | 13 | 10 | 138 | 768 | 18.0 | 400.2 | 9 |
| Purdue | 3 | 32 | 73 | 20 | 10 | 138 | 769 | 17.9 | 368.8 | 7 |
| Michigan St | 24 | 31 | 60 | 12 | 5 | 132 | 763 | 17.3 | 350.9 | 6 |
| Indiana | 6 | 25 | 63 | 22 | 16 | 132 | 774 | 17.1 | 407.5 | 11 |
| TOTAL | 130 | 388 | 736 | 187 | 130 | 1,571 | 7,981 | 19.7 |
These are full-season numbers. Ideally, we'd work with conference-only numbers to get a flatter playing field, but those aren't readily available.
I've ranked the teams from best to worst in terms of disruption percentage and included total defense (yards/game) figures and ranks for comparison. My working hypothesis is that disruption percentage may have some predictive value, as it helps separate out the degree to which a defense is forcing bad plays versus the degree to which an offense is just executing really well.
(That's not a perfect distinction. Good pass blocking, for example, is obviously a function of good execution. But, as a pro-predictive example, disruption percentage gives a defense credit for a good pass rush that results in a quarterback hurry, even if the offense nevertheless throws a really good pass to a well-covered receiver that goes for a long gain. And it doesn't give credit for a pass an opposing quarterback just flat underthrows even though he's under no particular durress.)
The disruption percentage and total defense ranks show a high degree of correlation. Illinois and MSU are the exceptions. Based on my theory, that would mean Illinois has been somewhat unlucky and MSU has been somewhat lucky in terms of how their fundamental defensive performances have translated into yardage gained by their opponents. It could also mean Illinois plays an aggressive defense that gets burnt by long plays quite a bit--and vice versa for MSU.
Regardless, the numbers demonstrate that the Michigan State defense has been among the least disruptive in the conference this season--just two-tenths of a percentage point ahead of last-place Indiana. The best defense in the league come close to doing something disruptive on 1 out of every 4 plays. MSU, meanwhile, is doing it on only 1 of 6 plays. In their worst performances, that ratio drops to 1 out of every 9 plays.
Most BCS-conference-level offenses are going to find ways to pick up first downs when the defense doesn't do anything to force them out of their rhythm on 8 out of 9 plays. Fixing that big-picture problem is the key factor if this football team is to have any success in the two games that remain this season and, perhaps more importantly, in the new season that begins less than 10 months from now.
Bonus conference-wide observations, since we're here:
- The league-leading number of quarterback hurries (24) for MSU is somewhat puzzling, given that 5 teams in the conference only register a QB hurry figure in the single digits. I initially thought this might be a function of inconsistent statistical scoring. But the number of hurries per game for MSU isn't any different at home than it is on the road; I'm assuming the same person scores the game for both sides in each game. Perhaps that's not the case for individual defensive stats, though. If that's the case and MSU QB-hurry numbers are inflated, their disruption numbers would become downright abysmal compared to the rest of the league. Alternately, the high number of QB hurries could be a function of blitzes that are only semi-effective; pass rushers are getting to the quarterback with a high frequency, but not in time to register a sack. (Although MSU does still rank 2nd in the league in sacks--so this whole thing has me puzzled.)
- The Iowa secondary is as good as advertised. Totaling pass break-ups and interceptions, you get 75 disruptive plays in pass coverage for the Hawkeyes. That's 17 more than the second best subtotal in the league (Minnesota's 58).
- Given Michigan's disappointing season, Brandon Graham won't win the Big Ten defensive player of the year award. But he may well be the single most effective defensive player in the league. On a team with only the 8th highest disruption percentage in the league, he's registered 26 individual disruptive plays (21.0 TFLs, 2 FFs, 2 PBRs, 1 QBH), which represents nearly 1 out 5 (18.8%) of his team's total disruptive plays.
- Those numbers do trail O'Brien Schofield's (29.5 disruptive plays; 19.5 TFLs, 7 QBHs, 2 FFs, 1 PBR; 20.2% of his team's total), but Schofield is presumably getting more help from his comrades on the Wisconsin defense so that opposing offenses can't focus on double-teaming him quite as much.
- Other top performers: Adrian Clayborn (Iowa) has 27.5 disruptive plays. Ryan Kerrigan (Purdue) has 23.5. Jamie Kirlew (Indiana) also has 23.5. Somewhat frighteningly, Penn State has three players with at least 16: Jack Crawford (17.5), Navorro Bowman (17.0), and Sean Lee (16.5).
- The individual numbers are obviously skewed toward defensive linemen, since tackles for losses make up almost half of disruptive plays.
- Greg Jones has 20.0 disruptive plays (11.0 TFLs, 7 QBHs,1 FF, 1 PBR). That's a very good total for a linebacker and equal to 15.2% of MSU's total.
- Returning to the frightening-Penn-State-stat category, the Nittany Lions' 91 tackles for loss are 14 better than the #2 team in the league (Michigan). Kirk Cousins is going to need to get the ball out to his receivers quickly on Saturday and, even more than usual, shouldn't count on any help from the running game.
- Ohio State leads the league in the percentage of plays on which a potential turnover (forced fumbles plus interceptions) is created: 5.6%. In the other three categories, they're actually pretty unremarkable.
- Michigan State, meanwhile, ranks dead last in that category at just 2.3%.
- Let's go ahead and end there, for empahsis.