Hokepoints Goes to Plaid Comment Count

Seth October 22nd, 2013 at 11:02 AM

Going up early 'cause we're going on WTKA this morning to yap about HTTV Hoops/Hockey from 9-10 with Sam Webb and several of our authors.


What have I done? My brains are going into my feet!

Brady Hoke said his team was prepared for Indiana's ludicrous speed offense, in other words: "Buckle this." Following the science fiction movie at Michigan Stadium last Saturday, the old hypothesis is again making its rounds: teams that don't play up-tempo tend to not be as prepared for teams that do, leading to an uncharacteristically negative defensive performance.

Ludicrous Speed

Fortunately there are data here (thank you once again cfbstats.com). They say Indiana is indeed the fastest ship in galaxy.

To get a tempo stat I just divided time of possession (in seconds) by total offensive plays. Games against FCS opponents are removed entirely. The Big Ten by Tempo (all FBS in Google Doc):

Team T.o.P/Play Nat Rk %ile YPP
Indiana 18.5 1st 100% 6.56
Penn State 22.3 26th 71% 5.52
Northwestern 23.4 38th 63% 5.86
Nebraska 24.8 54th 52% 5.84
Purdue 25.8 73rd 45% 4.37
Ohio State 26.0 77th 43% 6.40
Illinois 26.1 78th 42% 6.22
Iowa 26.2 80th 42% 5.52
Michigan State 27.4 95th 32% 4.55
Michigan 28.3 105th 26% 6.35
Wisconsin 29.6 118th 16% 7.45
Minnesota 30.5 122th 9% 5.24

It's not perfect since you can't pull out the seconds actually spent in a play, or the actual seconds during clock stoppages because of out of bounds or incomplete passes or first downs, but in the aggregate I think it does the job.* Michigan, as you supposed, is pretty low: 105th, and in the 26th percentile at 28.3 seconds per play. Nothing before or since on Michigan's schedule is like Indiana; for objects in the mirror: CMU is 97th, Notre Dame is 85th, Akron is 58th, UConn 55th.


* Anomaly: thanks to all the stoppages Penn State's offense vs. Michigan charted as fast as Indiana's. That's why I didn't use game-by-game stats, since those sorts of things average out and betray the offense's truer intentions.


High tempo does not equate or really even correlate that strongly with Yards per Play. Observe chart:


Click embiggens (updated)

Cal's offense functions at warp speed but its output isn't any better than Florida's ambles through the swamp. Wisconsin and Alabama both manage to move even slower than we do, and FSU is hardly faster, yet those are elite scoring machines. The linear tilt might be tempo teams winning a few more plays here and there, or it could simply mean the spread guys who run many of the great offenses today are just accidental carriers of up-tempo alleles (like how blue eyes followed the path of Vikings, but didn't necessarily provide any advantage).

The question, however, is not what tempo can do for you, but whether teams on the right side of the chart are more susceptible to those on the left. [Jump to see]

Does a Fast Offense Teach Fast Defense?

Methodology (boring part): Pretty simplistic: I started with an expectation of yards per play that each team would cede to its opponent. I simply took the mean of the average yards per play ceded by the defense and the average yards per play of the opponent's offense. So if Michigan's average defensive day against their competition is 5.05 yards per play, and Indiana's YPP has been 6.56 on average this season, the expectation is Indiana gains 5.81 yards per play against Michigan. Then I subtracted the expectation from the actual, so, again using our example, Indiana put up 7.53 yards per play against Michigan, versus 5.81 expected, is a value of 1.72 yards per play more than expected.

For the X axis I did another simple difference between the year-long tempo of the defensive team minus that of its opponent. So in our example Indiana has a TOP/play of 18.5 and Michigan's is 28.3, so there's a difference of –9.7 (the 7th largest gap in the study).

Results: Another scatter chart time:


Big enclickens.

The light blue stuff on the left are the extremes: defensive performances (vs expectation) of very low-tempo teams facing very high-tempo teams (the cutoff is Michigan vs. Penn State). The yellow circles are Michigan games. Of the bad (top half) from left to right it's Indiana, Akron, ND. Of the lower four (good) it's Penn State, UConn, CMU, and Minnesota.

Anything in the upper quadrant is your defense giving up lots of yards; the lower is your defense performing better than expected. The linear progression is slightly steeper for that sub-group (the line for football as a whole is so horizontal you can barely distinguish it from the X-axis), but still doesn't say much. There is nothing here to suggest high-tempo teams facing low-tempo ones have any advantage.

This doesn't prove anything, but suggests that the tempo complaint isn't related to Michigan's offensive pace. Some teams have been bombed out of the house by high-tempo teams, but the light-blue buckshot is pretty strongly centered on the midline. It could be other slow-ass teams like us know how to teach high-tempo to their scout teams. It could be the defense this week, despite warnings, just wasn't buckled in for Ludicrous Speed.



October 22nd, 2013 at 11:44 AM ^

YPP is a good way to look at it, but there's no way to prove/disprove whether or not this is an advantage for a particular team unless said team willingly (or unwillingly) changes pace.  Indiana won't do that, as they have basically designed their offense around their pace.  I doubt, and the above shows, that it's not an inherent advantage (if it was then everyone would do it) as perhaps making the most of a team's ability.  I noticed near the end of the game that Indiana was making more checks after lining up; this seemed to work against them as their last drive fizzled out.  They gave the defense too much time to get settled.

YPP probably won't go up from one strategy to the next, but tempo will certainly bring about an outcome faster.  Consider that Michigan in fact did go Indiana's pace  If we take the final 63-47 score and multiply it by the Michigan-Indiana tempo ratio (18.5/28.3), the adjusted score if Michigan dictated the tempo would be (arguably). . . 41-31.  As far as Michigan games go, against a good offense and a bad defense that's kind of expected.  It's certainly not CHAOSBALL.  However, this assumes both teams maintain the same productivity at the slower pace.  Michigan certainly has the capability (see: the ND game); when they nerf their own offense it's as much due to playcalling as it is the pace.  But Indiana's YPP might go down if they consistently played at a lower tempo.  Arguably, they managed to keep it as close as they did because of their pace.  Unfortunately for the academics, this is the one thing you won't be able to verify as they probably aren't going to try to slow the game down just so we can see what happens.

Blue in Seattle

October 22nd, 2013 at 1:05 PM ^

I'm sure that Michigan prepared for uptempo, but seeing it in the game was a different story.  They did get better as the game progressed though.  Especially the secondary getting themselves lined up.  And once that started happening faster, then Indiana could see that things were set, and then waited for the coaches to indicate whether or not to check to a different play.

For me, it is like every other scheme.  If your team can't execute it efficiently in the bread and butter plays, then you can never get to the counters.  And despite having scout teams, you'll never have a coach say, "yeah we practiced up temp, and it was clear that despite practicing it for 1 and a half days this week, we never really got our team to line up quickly, and of course our scout team has only practiced it for a couple of days, and Indiana has been doing it for a season and a half now."

This year many facets of the Michigan team are nice lumps of iron being shaped in a forge, they aren't pretty yet, or ready for tempering, but at least they are starting to look like a pointy weapon.

I think many negatives will be assigned to the LB corp for this game. Yes the secondary got burned early, but later on they were in position, and just didn't have the technique to make at least two more picks or PBUs 


October 22nd, 2013 at 11:45 AM ^

The chart that shows no distinct advantage for an offense that plays fast (ie, no huddle) brings up what I always think about when people on this blog exclaim, "getting to the line late means you can't adjust to defenses!!"  Which is, "well, getting to the line late means that it's harder for a defense to adjust to *you*."  It seems like getting to the line late would be at-worst a wash for an offense.  At best, since the offense is the side that's running a scripted play, giving the D as little time to adjust (because the odds that they happen to pick the right scheme to defend are definitely <100%) could be a good thing.  Am I way off here?

Space Coyote

October 22nd, 2013 at 11:53 AM ^

And I've tried to explain this before.

There are definately advantages to seeing a defense and adjusting what your play call is accordingly.

There are also advantages to the defense not being able to necessarily communicate everything immediately and as efficiently.

You can argue advantages for both, but I don't think one should be considered obsolete. It's similar to how not all huddle teams break the same. Some will break and make their way to the line and see how the defense aligns (but not necessarily get through all their proper communication), know their assignment, a snap the ball. Others will spring to the line and try to snap the ball before the defense is set, but then you have to adjust on the fly and it could result in a big play or a busted play. Teams do this with the speed of their motion as well.

Va Azul

October 22nd, 2013 at 12:11 PM ^

The premise "getting to the line late means that it's harder for a defense to adjust to *you*" cannot be true by itself.   Options always have positive value. (mathematical fact: options are bounded by a value of zero.  If exercising the option would have a negative value, you would simply choose not to exercise it.)  Getting to the line early gives you the option of hiking it as soon as the ball is set and it is legal to do so.  This length of time is the same regardless of how much time is on the play clock.  It's quite possible that running up-tempo would have a negative effect on expectation, but  it would have to come from some other mechanism, such as more time to make a better play call, poor communication, taking time away from technique work to practice executing the up-temp offense, etc. 



Space Coyote

October 22nd, 2013 at 11:49 AM ^

A similar, extremely slight to negligible uptick would be seen for defenses with up-tempo offenses facing teams with low-tempo offenses.

I think the slight uptick can be swallowed up by the idea "it's not what we do so there is at least a little unfamiliarity there" concept. So this probably cuts both ways.


October 22nd, 2013 at 11:51 AM ^

I have two thoughts.

First, Indiana seemed insanely fast. And statistically, they appear to be one of the faster teams in the country. If only one of RichRod's Michigan teams had ever been anything like that fast and efficient...

Second, while that speed clearly bothered Michigan on some huge plays, there were several other huge plays (that had a large effect on the averages) where the Michigan player was in perfect position and just didn't stop the play or tackle the reciever. So he ran for 235 YAC. Those two, the totally flumoxed by the speed and not so much but still unsuccessful, look the same to the numbers.

So, I think that there are some explanations why the Indiana data point might be an outlier--as it appears to be on the chart.

I should also note that I thought this analysis was quite well done.


October 22nd, 2013 at 12:04 PM ^

These two get mixed together often but you have to separate.  While no-huddle can be up-tempo it isn't always...

I have no idea how to do this but I am guessing by using my God given "I think I remember" powers that it seemed many of Indiana's big plays were pure up-temp affairs.  There were times where they got to the line and then checked out of plays and I could see that the defense was able to at least stop moving and communicate amongst each other.  I would be curious to see if these plays were on-par as successful as the up-temp ones.


October 22nd, 2013 at 12:27 PM ^

Brian showed IU scoring four times with drives taking under 2 minutes, so when they got the ball back with about 6 minutes left, I expected more of the same. However, if you re-watch that drive, they spent a ton of time checking the sideline and many times ran the play clock down to a few seconds. It's as if they didn't trust their own system. When push came to shove, they slowed it down to get the correct play call in. I was happy to watch them slow it down, and then we got a pick anyway to end that drive. I think they would have been better off up-tempo'ing, especially down two scores.


October 22nd, 2013 at 12:51 PM ^

But not the end-all-be-all. I don't understand why folks think the huddle is "archaic". Teams in the NFL still huddle. I don't think the huddle vs. no-huddle thing has anything to do with whether an offense can be successful. I personally think it just doesn't matter. 

That being said, I think that the ability to go uptempo is an asset and is necessary to success. Not just because of end-of-game situations that demand up-tempo play, but because of the ability to catch a defense off-guard at certain times in the game. Variety in offense doesn't just have to be the formations/pre-snap motions you use. Tempo is a valuable variation that should be used from time-to-time. And it really bothers me that Michigan has not gone up-tempo at all so far this year. Are they saving it? Are they waiting to fully install or utilize it? Surely Borges understands the value of up-tempo...most NFL teams use it at times, and Borges likes to emulate NFL-style offenses.

I think, in terms of both sides of the spectrum, neither side is fully optimal. You shouldn't be full-on dinosaur pace the whole game as much as you shouldn't go ludicrous speed the whole game. IU's blazing fast speed may get them a ton of points, but it leaves their already-terrible defense in very tough situations. You can't talk about the value of up-tempo style without talking about the ramifications of possibly tiring your defense out early in the game. 

One other thing I want to talk about...spread vs. pro-style. I don't think Borges "went to the spread" against IU. He spread the field out, for sure, but he was still calling plays primarily within the parameters of his pro-style playbook, save for a few read-option calls for Devin. If that makes sense. I mean, how many NFL teams run the shotgun? That doesn't make them spread offenses. I guess what I'm saying is it's not like Borges literally said "welp, my offense doesn't work. Time to get out RR's spread playbook". I think Al has plenty of shotgun 4-receiver formations. I think Hoke and Al specifically did not use them against PSU as much because they feared Devin's turnover problems. And the plan ended up backfiring due to their currently terrible Oline...I'm happy he opened the field up more and played to his offense's strengths again, turnovers be damned. Hopefully he continues with it.




October 22nd, 2013 at 1:29 PM ^

The question to ask is not simply whether there's any total yardage correlation between uptempo versus plodding offenses, but whether up-tempo offenses have a better chance of breaking a big play against the plodding offense. In other words, I wasn't surprised to see a lack of correlation strictly between uptempo versus plodding regarding total offense, or YPA. But I think you might see a correlation on plays over 25, 50, 75 yards.

I also, however, expect the same trend against any team, plodding or lightning fast. I mean, the primary utility of uptempo is the big play. Every once in a while, it'll get the defense misalined, or the secondary confused. It won't happen often, and against well-coached defenses with solid fundamentals (keeping the play in front and in the middle), it'll happen even less often. But you're essentially forced to play a 2-minute drill for an entire game. I think stats will show uptempo might be creating more big plays, or chances for big plays, than the traditional plodding offense.



Eye of the Tiger

October 22nd, 2013 at 1:20 PM ^

I think it's high time people stop evangelizing one way or the other about tempo--it is not "better" to control ToP and it is not "better" to race down the field at warp speed. Each conveys specific advantages and disadvantages. 

If you go up-tempo, you keep defenses from making adjustments and substitutions. But you also don't give your own defense a rest, will find it difficult to adjust to the defense on the fly and will find it close to impossible to kill the clock when that's what you need to do to win. 

If you go down-tempo, you can grind the opposing offense down, audible at the LoS, manage the clock and keep your own offense off the field, but may have trouble switching into high gear when needed and you give the opposing defense time to substitute and adjust to you. 

Both approaches can work really well--I mean, the current Oregon Ducks offense (up-tempo) and the Russell Wilson-led Wisconsin offense (down-tempo) are/were both brilliant at what they set out to do. 

I guess we can say IU is also quite impressive at playing up-tempo offense. But I wonder how they'll end up with a good enough defense to take it to the next level? Oregon does it by platooning good-not-great defenders, but they also have those incredible facilities and that amazing record over the past 5-7 years to attract talent in numbers to Eugene. 


October 22nd, 2013 at 1:38 PM ^

I was wondering this weekend about the effects a high-tempo offense has on that team's defense -- especially when it plays road-graders like Bama, Wiscy, etc.  It was thought in the RR years (not knocking them) that having a "D" constantly practicing against a high-tempo spread offense killed the D's ability withstand a punishing, plodding run offense.  I'm sure, though ,that had no effect on Wisconsin's 31 rushes in a row that one year recently against us.  (And all 28 plays in the second half.)

IU's defense isn't good, but, are they particularly bad because they aren't used to more plodding teams, e.g., MSU?  (Not certain if we're in that category.)

In other words -- a defense which has a low-tempo offense, apparently, isn't necessarily at a disadvantage facing a high-tempo offense.  But, is the defense with a high-tempo offense at a disadvantage to the low-tempo offense?

Seems this chart answers the question -- "No."  Weirdly, the far right of the scatter plot looks like the data is mirrored over the x axis.  No correlation at all.

CONCLUSION:  IU defense simply sucks, period.


October 22nd, 2013 at 3:25 PM ^

they have fast paced offense but also has elite defense.  Both of their offense and defense are in the top 10 in the country according to advance stats metric.  Of course, it helps to have NFL talent on the defensive side to curb the effect of seeing a lot snaps on the field.

Eye of the Tiger

October 22nd, 2013 at 4:40 PM ^

Since advanced statistical metrics sometimes overnormalize (i.e. with FEI).

But granted, Oregon's defense is at least "very good," and unusually so for a defense paired to such an up-tempo offense.

They do that by recruit so they can platoon on defense. Doesn't hurt to have the best facilities in FBS and the best aggregate record over the past 5 years either...


October 23rd, 2013 at 5:44 PM ^

That's actually a good addition to the analogy. Rich Rodriguez's early spread offenses went up-tempo because it was an advantage to screw further with people who never saw his offense before, or up-tempo before. The point is it wasn't necessarily part of the spread to run it at a breakneck pace, but the guys who adopted it were the kind to look for new advantages and try the cutting edge stuff.

Blue eyes are a thing mountainous people maybe developed over generations, but they were a separate thing from the cultural evolutionary traits that gave them an advantage in raping and pillaging.