A proposed explanation of the play-calling using a UFR review of the offense going into Improvement Week
After the Purdue game, I started to compile the UFRs from this season into a large chart to try and see what the offense has done well and what hasn't worked so far. This unfortunately but very coincidentally turned out to be a very useful tool to give some data on what Michigan may have been trying to build upon going into the Improvement Week and context for the play calling in the MSU game.
Disclaimers:
- This is an objective analysis (with some editorializing) to explain the playcalling, not to defend it.
- I did not collect any data on the types of passes thrown (i.e. long bombs vs. mesh vs. screens, etc.) so I cannot look at the decision against testing the MSU safeties more.
- I did not look at blocking schemes or individual blocking scores, as this is a higher level review. That does lead to some clear limitations here.
- There is no data to explain why Michigan did not account for the monsoon. But, my fan theory based off of this review is that Michigan spent the bye week (prior to any forecast for storm) working on plays that are weather-independent, and then were stuck between a rock and a hard place because the main things they had worked to improve would not work well in a monsoon. The rock and the hard place both split the victory, unfortunately.
- There is no data to explain why Michigan totally abandoned the running game after the first drive. In fact, the data makes it pretty clear that Michigan's passing game had its greatest success when there is a viable PA (which requires running plays to be called, regardless of their own success). But, I will attempt to show why Michigan chose to pass so much on first down. And when you fail to gain yards on first down, you're stuck between a rock and a hard place on 2nd or 3rd & long. We all know who won there...
Three basic observations from the game
- Michigan utilized a ton of Shotgun and Ace formations (albeit with some exotic twists and motions.)
- Michigan passed a ton on first down.
- Michigan utilized a ton of 4 & 5 receiver passing plays.
Explaining the three basic observations
1. Why did Michigan utilize the Shotgun and Ace formation so heavily?
First, let's look at this from the vantage of the Michigan running offense:
O Form | Big | Ace | Shotgun | Offset I | I-Form |
n | 43 | 41 | 29 | 17 | 6 |
% Form | 0.32 | 0.30 | 0.21 | 0.13 | 0.04 |
Yards | 133 | 276 | 151 | 87 | 1 |
YPP | 3.09 | 6.73 | 5.21 | 5.12 | 0.17 |
Success rate | 0.47 | 0.44 | 0.55 | 0.29 | 0.17 |
Off-schedule | 0.09 | 0.27 | 0.45 | 0.24 | 0 |
RPS sum | -13 | 0 | 0 | 5 | -1 |
Box | 8.45 | 7.66 | 6.47 | 7.41 | 7.42 |
Blocking Avg. | -0.70 | -0.14 | 1.00 | 3 | 0.20 |
RB sum | 1.5 | 13 | 10.5 | 1 | -2 |
Running out of the Ace formation led to the highest yards per play of any formation, with the rate of successful runs being 3rd out of the 5 formation bases utilized.
Running out of the Shotgun formation led to the highest rate of succesful runs, and second highest yards per play. On average, Michigan had the best blocking execution out of the Shotgun than of any other formation by a wide margin.
For whatever reason, the RBs also had the most success making positive plays out of Ace and Shotgun formations.
The caveats with Shotgun is that it had the highest proportion of Off-schedule use (i.e. runs called on 2nd or 3rd & long), as well as the lowest average defenders in the box (which may be explained by the former). However, this can be seen as an advantage for passing plays, as it gives credibility to play fakes on seemingly obvious passing downs.
Now, let's look at the successful plays (at least 4 yards on 1st down, at least 60% of the yards to gain on 2nd down, at least 100% of the yards to gain on 3rd or 4th down), and see how it breaks down from these formations when it comes to Michigan executing its bread-and-butter running plays:
Successful Plays | |||||
O Form | Ace | Shotgun | |||
Play | Zone | Counter | Crack sweep | Zone | Power O |
% Form/success | 30.43% | 100% | 50% | 34.78% | 50% |
% Form/play | 23.94% | 80% | 46.67% | 23.94% | 26.92% |
Relative success | 1.27 | 1.25 | 1.07 | 1.45 | 1.86 |
To explain the rows, on the Zone plays out of Ace formation, 30.43% of successful Zone runs were out of the Ace formation, compared to only 23.94% of all Zone runs being called from the Ace formation. This is a relative proportion (success rate/play call rate) of 1.27.
Overall, this suggests that Michigan believed it had a relatively strong running foundation to build off in the Ace and Shotgun formations.
----------
Second, lets have a similar look at the Michigan passing offense:*
O Form | Shotgun | Offset I | Shotgun Empty | Ace | I-Form | Big |
n | 58 | 17 | 16 | 15 | 12 | 9 |
% Form | 44.96% | 13.18% | 12.40% | 11.63% | 9.30% | 6.98% |
Yards | 222 | 142 | 203 | 145 | 65 | 50.00 |
YPP | 3.83 | 8.35 | 12.69 | 9.67 | 5.42 | 5.56 |
Success rate | 0.29 | 0.47 | 0.69 | 0.8 | 0.25 | 0.22 |
Off-schedule | 0.36 | 0.65 | 0.00 | 0.73 | 0.75 | 0.89 |
RPS sum | 0 | 5 | 6 | 8 | 4 | -1.00 |
Blocking Avg. | 0.30 | 1.06 | 1.73 | 0.91 | 0.55 | 1.00 |
QB DSR | 60.98% | 64.29% | 85.71% | 100% | 80.00% | 28.57% |
Route/YAC sum | 5.00 | -1 | 3 | 0 | -4 | -1.00 |
*Notes:
- I have the plays separated between O'Korn vs. Speight in my chart, as well as red zone vs. between the 20s, but I have combined them all for this review. So again, some obvious limitations apply. See my previous post for a comparison of the performances between the two going into Improvement Week
- My yards and yards per play includes sacks AND penalty yardage. Same goes for the success rate. This is a results-based analysis, i.e. drawing a 15 yard PI call is a success in my book.
This again shows that Michigan found a relatively high proportion of its success out of the Ace formation. The Shotgun formation, however, is where a lot of the problems with the OL rear their ugly heads. The caveat here is that the shotgun is the go-to formation on standard passing downs, as indicated by the huge discrepancy in off-schedule rate from the Shotgun compared to the other formations. But when you look at the Shotgun empty, there is a great amount of success.
2. Why did Michigan pass so much on first down?
Again refer to the passing chart above. There is a stark difference in success rate out of formations that have a higher proportion of plays on non-standard passing downs (Off-schedule).
Overall, Michigan had a 41.86% success rate on passing plays. That success rate jumps up only slightly to 43.33% on off-schedule downs, but the highest jump is in the Shotgun formation, from 29% as above to 38%.
3. Why did Michigan use so many 4 and 5 route passing plays?
I already made a post alluding to this prior to the game: UFR review shows pass protection in Shotgun Empty formation is stellar.
That breakdown, as shown again above, suggests that the OL has an easier time blocking when there is no RB and/or TE to confuse the assignments (or whiff their own block).
Let's look closer at the Shotgun formation, where much of the blocking woes become most visible:
n | 58 |
% Form | 44.96% |
Routes* | 3.778846154 |
Pass rushers | 4.784313725 |
Blocker discrepency | 1.39 |
Blocking Avg. | 0.3 |
*Note: I counted a route where the TE or RB blocked prior to the route as 0.5
Michigan had a number's advantage in pass blockers:pass rushers on every play, yet the rate of adequate pass protection:failed pass pro was barely above zero.
However, a break-down by # of routes suggests that that the relative lack of routes and QB is also to blame:
On Shotgun passes with 3 or 3.5 routes:
Shotgun Formation | ||||
Routes | Blocker Discrepency | Blocking Avg. | Success Rate | DSR |
3-3.5 | 1.64 | 1.35 | 0.18 | 43.75% |
4-5 | 1.20 | -0.29 | 0.36 | 65% |
The success rate jumps going from 3/3.5 to 4-5 routes, but the blocking averages plummet. The blocking dip could partially be due to the lack of extra blockers (as the blocker discrepency does go down a touch), but, from recall, most of the stunts that killed Michigan came during 4 WR shotgun sets, so extra blockers would not necessarily have helped there.
It's hard to fully explain the jump in success rate, but the jump in DSR could possibly suggest that receivers were getting open more when there were more routes.
Regardless, the data overall shows that Michigan had relatively more success on 4-5 receiver routes.
--------------
Discussion
My take-away from this information is that Michigan had a clear idea of what had previously worked best going into the Improvement Week. But, it felt like a dad joke gone wrong: Drevno & Co. told a funny joke last week, and then they kept repeating it and repeating until they beat the dead horse so silly that Little Brother snapped and cut the bolt on the gun locker and shot everyone dead.
Again, none of this touches on the offensive staff's failure to account for the rainstorm, but I've already stated above that my guess is they had a gameplan (or at least several packages) in place prior to the weather forecast and failed to adjust.
And there's only so much you can do when the OL is a seive, the QB throws an INT turkey, and your players fumble the ball twice before it even starts raining.
But, at the end of the day, the play-caling had a very Borgesian feel in that they took what worked previously totally out of context and created a game-plan that was doomed from the start. The passing plays on off-schedule downs out of standard running formations only succeeded in the pass because of downs dedicated to running the ball made play-action a conceivable threat. It was frustrating to see Michigan abandon the run after falling behind quickly, as it totally threw off the basic premise of the gameplan's vision.
October 9th, 2017 at 11:47 PM ^
Great analysis! Extremely interesting. Thanks for sharing!
October 10th, 2017 at 1:58 AM ^
October 10th, 2017 at 11:27 AM ^
If we could have flipped field position and gained 15-20 yards, it would have opened up more options for the play calling.
October 10th, 2017 at 1:13 PM ^
" I think the coaches ... probably thought that the early success on the ground was not going to last once MSU made some adjustments."
My hypothesis is that Harbaugh & co continue to make NFL-style coaching adjustments against college coaches (& college players). That is, they quit while they're ahead, assuming their opponents will compensate and that a successful play will only be successful once, rather than going back to the well again and again until it goes dry. Against good coaching staffs, you have to do this -- witness our own defensive adjustments. And it's certainly highly preferable to the sort of banging your head against a rock until it breaks that we saw under Hoke. But it's also frustrating to watch a team go away from what's working when it's still working, and wonder, "wouldn't that have worked again?"
October 10th, 2017 at 1:28 PM ^
MSU was, in fact, adjusting. There were gripes about the lack of screens, but recall that (for example) a swing pass to Grant Perry that was poorly thrown incomplete in the second half also wasn't going to work anyway because there was a guy in Perry's face--MSU was ready for it.
October 10th, 2017 at 10:13 PM ^
we had so many freaking chances to make up a very small deficit. We easily could have won the damned game and be griping but not disconsolate. Yes MSU played tough as always, but they fell into the win.
It's the "40 formations in the first half" that continues to haunt me. I think there's still a chance that the offense gels. Or that the game plan is stuck to whether it gels or not, and that it all becomes an investment in next year. If so, Peters should be getting some time at the helm.
October 10th, 2017 at 10:52 AM ^
October 10th, 2017 at 11:28 AM ^
October 10th, 2017 at 3:51 PM ^
Thanks for putting in the time and sharing.
October 11th, 2017 at 8:10 AM ^
Seriously, 2002 style. Go ultra conservative, play field position, and our defense and special teams will win the games. Go back to different formations to run and pass out of, but drill it into the offense's head: do not fumble, no need to fight for the extra yard. If the throw's not blitheringly wide open in 2.5 seconds, run if you have space or chuck it out of bounds. Either way, our special teams will net us field position and our D will get us the ball back in, oh, about 3 plays. No need to be heroic. Like chinese water torture, eventually we will get enough points on the board for the W.
October 12th, 2017 at 8:01 AM ^
October 12th, 2017 at 9:32 PM ^
Is there a solution??
Comments