Hokepoints: The Defense is Still The Wall

Stuffing. Upchurch

Eleven months ago I used this space to discuss Michigan's crazy success in defensive short situations. That was brought on by a staggering performance against Illinois, at which point Michigan had stopped 15 of 27 3rd- or 4th-and-one situations, and 13 of 19 against real competition. This was up from stopping less than a quarter of such plays the previous two years, and almost as far above the going rate for all defenses.

This was huge. Getting one yard for any offense is far easier that stopping it for any defense—one good block can usually do it. Forcing a 4th down situation from 3rd and 1 or a turnover on downs on 4th and 1 is worth half a turnover or more. Jamie Mac addressed this further in his HTTV article, showing that the stoppage situation was affecting the happy margin between our yards-ceded defense and scoring defense as much as having a ridiculous year in turnover luck.

Michigan last year was really good at stopping the short stuff, but folks chalked it up to Martin and Van Bergen playing to their strengths and figured it was a blip. Except it wasn't just those guys. Here's last year's chart for short situations, through OSU:

Two thirds of Michigan's short-down production from last year returned (as did bad refs). Demens, Roh, Ryan, Kovacs, and Campbell were all key role players in that ridiculous shutdown rate, and if the UFR can be trusted, they weren't getting it just because of things the Team 132 seniors were doing.

This doesn't even count things like stopping Ohio State on 3rd and goal from the 2. Actually it doesn't count goal line situations at all, though 1st and goal from the 1 is as hard to stop as 3rd and 1 from the 40. So I revisited when updating the UFR database. Get ready to be happy (through MSU):

It's still happening. It's happening more. We replaced Martin and RVB with Washington and Campbell, and if anything got better! And like last year Michigan's short defense seems to be getting tougher as the season goes on. Since Big Ten play started, the non-stops have read thusly: Purdue converting with 16 seconds left in the half while down 18, Illinois benefiting from a terrible spot, two plays where Bell was forced to cut back into the pile and just managed to squeak through, and one bust.

[After the jump, what's causing it, and the plays vs. State]

Culprits via UFR through Illinois (play-by-play details at bottom of the article) as compared to last year:

I guess mentally double the 2012 numbers to project the final numbers. Nobody has slid backwards, and the linebackers are MAKING PLAYS with greater frequency.

The defensive UFR for Michigan State should be up by tomorrow but I pulled the relevant plays. Michigan stopped just one of four against Le'Veon Bell and MSU's patchwork offensive line. Still, there were some good things to see:

Play the first (No Gain):

State motions to the playside, M slants away from it. Washington was beat by a double but Roh made his dude completely whiff to cut off any cutbacks, and Ryan has shot inside the right tackle (who jumps on Ryan' s back--no flag) at the LOS. Bell did the Bell stutter-step thing he always does, then tried to dive outside, but Demens and Gordon have come up quickly and stop it short. Morgan had shed his block and was there in case Bell tried to power it. This brings up a few bars of Temptation and a Spartan punt.

Play the second (They get it, barely):

Fullback motions from H-back on the playside to H-back on the other side then back to an offset I on the playside again. Michigan is like 'whatever' and stays put. Roh and Ryan blew this up. Roh fought past a double to plug the middle (negating the left tackle and McDonald) and Clark got enough penetration to cut off the backside, so it's cut to the frontside. There, Ryan stuck the fullback a yard behind the LOS, fought past him, and grabs Bell's leg. However Demens was a bit slow to impact the pulling guard so Bell's cut and leap is just enough to fall past the 1st down marker. Washington had flowed down to help. So crose.

Play the third (They get it, RPS'ed?):

MSU overloads the boundary side (two TEs plus a fullback in the I) and Michigan walks Kovacs up to be tiny LB. Again with the sike! motion, this time from the TE, who starts at playside H-back, goes past the midline then winds up an offset fullback. Kovacs is moving with him and that momentum is still carrying him toward the boundary when the ball is snapped. Campbell is blown off the line, Washington is smacked into and turned over by the center, and Bell can pick his way for the first. Demens ran himself out of the play and into a block, giving up three more yards before pursuit from Clark and free hitter Floyd can get there.

Play the fourth (They get it)

No motion this time from an I-form 2TE formation, same play as the 2nd one above. M is slanting again but this time when Roh beats the guard blocking him the guard just pushes him (from the back, nooch), past the play. Slant gives Clark a two-for-one a la Ryan in play the second above, and so Bell has to cut back behind the guard shoving Roh. Ryan gets into Bell's knees with the OT hanging on his jersey and manages to momentarily stop the play short of the sticks if a LB can just arrive to hit Bell with authoritah. However Morgan used up his free hit generated by Clark to plug the pulling guard, while Demens tries to whack the pile and ends up just thumping Ryan. Bell keeps churning and squeezes inside that for the 1st. C'est la vie.

How Much Effect?

We can kinda figure this out using expected points from the various game states before and after these plays occurred. I tracked down the Mathlete to plug the performance so far into his database, and here's what he came up with:

Year	Opponent	Score diff	Time	Ln	Dn	Ds	Yards	Difference
2012	Alabama	-27	5 min 4th	M1	2	G	1	0.0%
2012	Air Force	4	3 min 1st Q	M3	3	1	-1	5.6%
2012	Air Force	4	9 min 3rd	M1	3	G	0	0.1%
2012	Air Force	4	9 min 3rd	M1	4	G	1	-4.9%
2012	Massachusetts	18	10 min 2nd Q	O35	3	1	3	-0.7%
2012	Notre Dame	-10	1 min 2nd Q	M1	2	G	0 (Pen -5)	1.6%
2012	Purdue	0	13 min 1st	O34	3	1	0	UPDATE: 3.1%
2012	Purdue	18	EOH	M16	3	1	3	-0.3%
2012	Illinois	0	10 min 1st Q	O48	3	1	0	3.8%
2012	Illinois	7	5 min 1st Q	O34	3	1	-5	3.5%
2012	Illinois	17	1 min 2nd Q	M41	3	1	0	0.2%
2012	Illinois	17	1 min 2nd Q	M41	4	1	-1	4.9%
2012	Illinois	38	EO3Q	O33	3	1	2	0.0%
2012	Illinois	45	4 min 4th Q	O37	3	1	-2	0.0%
2012	Michigan St	3	5 min 2nd Q	M21	3	1	0	5.1%
2012	Michigan St	6	<1m 2nd Q	O34	3	1	1	-1.3%
2012	Michigan St	6	9 min 3rd Q	M16	3	1	4	-2.0%
2012	Michigan St	2	7 min 4th Q	M30	3	1	2	-2.8%

That's about 12.8% 15% of a win added from this year's short situations, mostly from Illinois. Unfortunately he couldn't give me expected points instead of Win % since scores that were already out of hand made the results of a few of these moot.

Still, things: stopping the Illini on 3rd down from our 41 didn't make much difference because they were still likely to get it on 4th; when they didn't it was a big deal. Also note one stop against State was worth as much as losing the other three combined. In a tight game, stopping 1 of 4 is treading water; we're getting 3 of 5!

Takeaways: Even after graduating two NFL defensive tackles, Michigan remains exceptional at stopping short situations thanks to better than average play from our defensive tackles, better than good play from Roh and the linebackers, Mattison's ability to correctly guess which direction the offense wants to go and shut that down with his playcalling, and the existence of Jake Ryan, who is boss. With turnovers way way down from last year, Michigan's 'and-short' aptitude has become an even more important factor in maintaining an elite scoring defense.

------------------------------