Moving the Stati-Sticks: Post-Indiana

Moving the Stati-Sticks: Post-Indiana

Submitted by Adam Schnepp on November 25th, 2016 at 11:00 AM


[Bryan Fuller]

The title of the post still says “Post-Indiana,” but I seriously considered bucking convention and naming it “Pre-Ohio State.” With the stakes of The Game as high as they’ve been in a decade, it only felt right to look at where the two teams’ advanced stats are similar and where they’re different.

Still, it’s worth discussing what happened against Indiana. The offense took a fairly large hit in overall efficiency, falling from 23rd to 41st in success rate. The rushing offense’s success rate saw a nearly identical drop in the national rankings, falling from 21st to 42nd. The passing offense did even worse, with the success rate falling from 45.3% to 42.5% and from 26th to 50th. On standard downs, Michigan’s offensive success rate only dropped from 51.5% to 49.5%, but their ranking tumbled from 22nd to 43rd. The offense fared worse on passing downs, with their success rate dropping from 34.4% (39th overall) to 31.5% (62nd overall). The offense had a difficult time keeping on track, and their inability to pick up the necessary yardage to stay in manageable down-and-distance situations led to the drops in success rate. Of note is that the offense struggled in that department across the board (passing, rushing, standard vs. passing down, etc.) but with little impact on their other numbers, which stayed fairly stable. A few long runs helped keep the offense’s explosive play-realted numbers afloat.

The defense, already at or near the top of most categories, saw little movement. One of the bigger changes was in the defense’s IsoPPP, the number Bill Connelly uses to track explosiveness; Michigan moved up from ninth to fourth. That’s pretty much it. The defense is good. The stats are good. They both remained so against Indiana.

[After THE JUMP: how Michigan stacks up against Ohio State according to S&P+ and FEI]

Moving the Stati-Sticks: Post-Iowa

Moving the Stati-Sticks: Post-Iowa

Submitted by Adam Schnepp on November 18th, 2016 at 10:01 AM

[Eric Upchurch]

We should have known from the start. Ace and I got dropped off at the northeast corner of Kinnick with a simple task: get the parking pass and credentials. We picked up the envelope, untangled the strings on our credentials, and found precisely zero things that were definitely or even slightly resembled a parking pass. We now needed the guys to somehow turn around and come pick us up without bringing traffic to a screeching halt, and from there we needed to go to the parking garage and hope they’d accept the email that said we were approved for parking.

Ace and I crossed the street to wait and ran into Kelly and Jacob, two of The Daily’s beat writers. They were looking for a missing co-worker. Hindsight’s 20/20, but it should have been obvious in that moment that there were four Michigan media members on a corner and not a single thing going right for any of them, and that ended up serving as an omen for the night. We then dove into the van as the guys drove up, pulled up to the parking garage, and were shut down by the attendant. And the one in the next lot. And the one in the lot after that.

We ended up parking behind the equivalent of a coney on the outskirts of campus, running out of time to go to Demorest’s tailgate, and one of us slipped and fell partially down a hill while making fun of someone’s taste in music. It was me. I fell down a hill.

I didn’t notice the blinking neon “Hey, I’m a metaphor and something’s not right here” sign, and a lot of that is because you can usually count on nine games worth of data to be somewhat predictive. According to Football Study Hall, Iowa went into the game with a 5% win probability. They then forced Michigan’s offense into far and away their worst performance of the season en route to a bizarre, special teams-fueled upset. With game and cumulative numbers in hand, we can see how big an impact one bad game has while also preliminarily judging how predictive the performance was.

[Hit THE JUMP to find out how off Michigan’s offense was according to S&P+ and FEI, and a quick look at Indiana’s advanced stats]

Moving the Stati-Sticks: Post-Maryland

Moving the Stati-Sticks: Post-Maryland

Submitted by Adam Schnepp on November 11th, 2016 at 10:14 AM



Things actually went pretty well for Maryland. They piled up 367 yards at a 5.6 yards-per-play clip. Their average start was on their own 25.1-yard line, which is a couple of yards behind opponents’ average starting spot against Michigan, but their success rate was a very respectable 35%. Maryland had four players who had 30+-yard receptions on the day, including one that went for 47 yards. They even picked up 19 first downs, a surprising 12 of which came via the pass.

All of that didn’t lead to much in the way of points, of course. Maryland finished with three points on three scoring opportunities; their doinked field goal and tunnel screen that got tackled at the one-yard line as time ran out in the first half could both have been converted and it still wouldn’t have mattered considering the way Michigan’s offense was operating on Saturday.

Michigan averaged 10 yards per play. They had 10 drives and posted 10 scoring opportunities. They averaged 5.9 points per scoring opportunity. Their success rate was 65%, they picked up 31 first downs (including a 14-14 split passing and rushing), and they finished +2 in turnover margin. They were, in effect, unstoppable.

Maryland had a pretty good game relative to what opponents usually do against Michigan’s defense; it once again came mostly via a couple big busts and didn’t make a big impact on the defense’s overall stats. The story of Michigan’s season to this point isn’t how dominant the defense has been, but how far the offense has come. Last season the defense was statistically great enough to prop up (until that one game) the middling offense.

This year, Michigan’s defense has spent most of the season atop the S&P+ rankings and in the top 10 of defensive FEI (they’re currently ranked sixth). The offense’s success rate and explosiveness (IsoPPP) have steadily climbed, their average starting field position has been the best in the nation for a while, and the offense as a whole is now ranked eighth per S&P+. According to FEI, the offense is the best in the nation; they were ranked third before Maryland, and they climbed multiple spots after they game in every category offensive FEI tracks. This is not 1997 and it is not 2015. Michigan finally has an offense that should worry opposing coaches in its own right. They happen to couple it with the best defense in the country. It’s going to be an interesting two months.

[After THE JUMP: more on Maryland, Connelly’s Five Factors, combing through FEI, and looking at Iowa’s stren—looking at Iowa]

Moving the Stati-Sticks: Post-MSU

Moving the Stati-Sticks: Post-MSU

Submitted by Adam Schnepp on November 4th, 2016 at 1:02 PM



Three years ago, Devin Gardner walked off the field defeated in every conceivable way. Really, he “walked” only under the loosest possible definition of the word. “Staggered” is far more accurate. I was reminded of that this week when it somehow came up in conversation and I saw a gif BiSB posted from the 2013 game of Gardner, mud-stained to the point where it’s hard to make out the numbers on his shoulders, hobbling off. [Ed. (A)- BiSB conveniently led Opp Watch with the gif, so scroll down a bit to see what I’m referencing.] That year’s gif post was six gifs. Six. Ace mentioned in our MGoSlack chat that he clipped 75 things this season.

As it stands today, ESPN’s FPI gives Michigan a 39.6% chance to win out. Bill Connelly’s numbers at Football Study Hall are far more generous; he gives Michigan a 61% chance to win out. According to Connelly, Michigan has a 96.4% probability of finishing 11-1 or better. Against Ohio State, Michigan’s lone remaining game whose win probability isn’t 93% or higher, his numbers say Michigan has a 68% probability to win. Yes, Michigan busted a few times against Michigan State. They got scored on late in the game, scores that lifted the game out of garbage time. That Michigan went on the road in a rivalry game and met Connelly’s criteria for garbage time is a feat in itself, and you only have to click on that gif post above to really appreciate what this team has been able to do. That may seem like an overly emotional response, but Michigan’s aggregate numbers bear out their dominance, particularly on defense.

[After THE JUMP: the numbers that bear out Michigan’s dominance, particularly on defense. oh, some Heisman stuff, too.]

Moving the Stati-Sticks: Post-Rutgers

Moving the Stati-Sticks: Post-Rutgers

Submitted by Adam Schnepp on October 14th, 2016 at 10:07 AM



Bill Connelly used a photo from the Michigan-Rutgers game to head his Week Six Five Factors box scores, and it’s not hard to see why. Scrolling across the stats will lead to either your jaw dropping or some guttural chuckle, or maybe both: Rutgers averaged 0.28 yards per play, had one scoring opportunity (read: had the ball inside Michigan’s 40), got zero points off that opportunity, and had a Success Rate of 11%. Michigan’s offense averaged 8.11 yards per play, had 11 scoring opportunities, averaged 7.00 points per opportunity, and had a Success Rate of 55%. Rutgers had 17 drives to Michigan’s 18. Sometimes things really are as lopsided as the final score indicates.

ESPN’s Football Power Index took note, as Michigan now holds the top overall spot in FPI. ESPN says that FPI “represents how many points above or below average a team is.” Michigan’s 1.8 points better than the next nearest team (Alabama), and they’re 2.5 points better than Ohio State. One dominant game over Rutgers really boosted Michigan’s stock in the eyes of ESPN, as their chance at winning out rocketed from 16.6% to 34.2%, and their chance of winning the conference jumped from 30.8% to 49.5%. ESPN also predicts Michigan will win 11.8 games.

The season outlook wouldn’t be nearly as shiny if the offense wasn’t above average, but it’s no secret that the defense is performing at a level we haven’t seen in quite a while and driving the numbers up. There is literally no section of Connelly’s advanced stats profile where Michigan’s defense isn’t ranked first nationally in some category. The defense is even ranked first in two of Connely’s Five Factors. I decided to take a closer look at passing downs defense this week because it has been exceptional, but at this point the defense is so good that I don’t have to dig through specific categories for something to discuss so much as rotate through them. This week, the visual representation of the defense’s dominance is as sharp as it’s been in the last two seasons; before you read anything below, you can scroll through the graphs and see exactly what I mean.

[After THE JUMP: Connelly’s Five Factors and a closer look at passing downs defense]

Moving the Stati-Sticks: Post-Wisconsin

Moving the Stati-Sticks: Post-Wisconsin

Submitted by Adam Schnepp on October 7th, 2016 at 11:00 AM



Michigan’s win over Wisconsin wasn’t put away until Jourdan Lewis executed one of the more dramatic and insanely athletic interceptions most of us will ever see, but the stats show that it should have been over sooner. That won’t come as a shock—if you watched you’re probably thinking of the missed field goals right now—but it does reinforce how good Michigan’s defense is.

Even the most basic stats hint heavily at the defense’s dominance. Wisconsin only ran 53 plays for a paltry 154 yards, or 2.91 yards per play. Michigan’s offense fared far better, running 80 plays for 339 yards, or 4.24 per play. As Jedd Fisch noted this week, no team has crested 330 yards of offense against Wisconsin since the 2015 Alabama game. Not a bad yardage total against a defense that’s still ranked fifth in S&P+.

It’s not all sunshine and roses (or lipstick-shaped trophies), though. Michigan had six scoring opportunities to Wisconsin’s three, but both walked away with 2.33 points per scoring opportunity. Michigan was averaging 6.3 points per trip inside the 40; drives bogging down and missed field goals knocked the points per scoring opportunity number down to below a field goal for the game.

We don’t yet know whether the missed field goals were an aberration or the new, hand-over-the-eyes, college-kickery normal, but it’s relatively clear that Michigan’s offense on the whole did well against one of the best defenses in the nation. Michigan’s offensive Success Rate was 40%, which must have been like a walk in the park for Speight and co. compared to Wisconsin’s offense’s 21%. Wisconsin’s defense is superb; Michigan’s defense is a black hole.

[After THE JUMP: how a low-scoring day impacted the fancy stats]

Moving the Stati-Sticks: Post-Penn State

Moving the Stati-Sticks: Post-Penn State

Submitted by Adam Schnepp on September 30th, 2016 at 10:00 AM



Another game, another batch of data lost to garbage time. That’s certainly not a complaint, but it does explain why there wasn’t a ton of movement in the advanced stats. Michigan hit Bill Connelly’s criteria for garbage time with 46 seconds left in the first half, as Karan Higdon’s two-yard touchdown run put Michigan up 28. From there, the lead never shrunk to the point where there was any real threat of exiting garbage time.

The traditional stats paint a picture that matches what we observed: Michigan dominated in almost every area. You can run down the stat sheet and blindly point and likely find a stat Michigan handily won. First downs? Michigan more than doubled Penn State, 25 to 12. Rushing first downs? Michigan had a 14-3 edge. Yards per play? Michigan fell just short of doubling up Penn State with 6.1 to their 3.5. Rushing yards? Michigan’s 326-70 advantage was one of their biggest statistical wins of the day. That stat obviously needs to be contextualized a bit; Michigan still blew Penn State away when looking at average yards per rush, of which Michigan averaged 6.7 to Penn State’s 2.5. That Penn State stayed fairly balanced (28 rushes and 27 passes) and walked away with the aforementioned average yards per rush and 4.5 yards per pass is quite the accomplishment for the defense.

The stats in the preceding paragraph got me thinking about Michigan’s rush defense, and after looking through the numbers it deserves the spotlight this week. The traditional stats help set the table: Michigan has allowed 64 and 70 rushing yards over the last two games. Looking at things a little closer, Michigan has held opponents back in a variety of ways and situations; Bill Connelly’s advanced stats profile reflects that, and some of the non-Bill Connelly total win predictions and win-out percentages seem to have taken notice as well. 

[After THE JUMP: Five Factors, counting the ways Michigan’s rush defense has been exceptional, and looking ahead to Wisconsin]

Moving the Stati-Sticks: Post-Colorado

Moving the Stati-Sticks: Post-Colorado

Submitted by Adam Schnepp on September 23rd, 2016 at 4:00 PM

29718243306_443f97a95e_z (1)


After two games that took barely over a quarter each to hit the criteria for garbage time, Michigan got their first real test of the season last Saturday. The players all said it was good for the team in the postgame presser, and it was also good for the numbers, as there’s now a substantially larger set of data from which to drawn way-too-early conclusions.

By and large, Michigan’s numbers are still pretty shiny. Wilton Speight, whose elbow was injured early in the game, is completing 63.8% of his passes for 7.7 yards per attempt. The top four receivers are all averaging over nine yards per target. De’Veon Smith, so long thought of as a three-yards-and-a-cloud-of-dust back, is averaging 5.6 yards per carry, getting more than five yards past the line of scrimmage on 37% of his carries, and averaging 6.6 extra yards per carry beyond those first five yards on the 37% of runs that get there. There aren’t the same types of individual stats available for players on defense, but that side of the ball has been excellent in all areas save one: explosive plays allowed.

On that note, there’s been a ton of discussion over the last week about the defense in various corners of the internet and on the airwaves filled by sports talk radio commentators and callers. Is there real cause for concern after a couple of bad quarters? If there is, what exactly do the numbers say are the areas that are most distressing? Can stats tell us whether these look to be persistent issues, or are they one-off miscues?


Moving the Stati-Sticks: Post-UCF

Moving the Stati-Sticks: Post-UCF

Submitted by Adam Schnepp on September 16th, 2016 at 11:02 AM



Welcome to the first 2016 edition of Moving the Stati-Sticks, our weekly deep-dive into the advanced stats that have become increasingly popular in college football and the weekly in-season post that you most likely forgot existed. I’m hoping to change that this season. Instead of a bunch of tables or numbers in text, I’ll be creating graphs (and they’re mostly intelligible!) for items of note. This won’t be an exhaustive look at every single line item in the advanced stat profiles that are out there, but rather items of interest spun out of last week’s game, followed by a quick look ahead to Saturday’s opponent.

We’ll start with a thousand-foot view of Michigan, and the view’s pretty nice, actually. ESPN’s FPI predicts Michigan’s record will be 10.8-1.7 with an 11.9% chance they win out and a 39.9% chance they win the conference. Bill Connelly’s advanced stats profile at Football Study Hall, from which most of the rest of this post is derived, gives Michigan a 9% chance to win out and 35.2% chance to finish 11-1 or better. Brian Fremeau’s FEI doesn’t have opponent-adjusted stats available yet, but Michigan’s first in the country in unadjusted game efficiency. They’ve also played teams that FEI ranks as either below average (ranked 71-90) or awful (ranked lower than 116), but, uh, wooo we’re number one. More detailed stats are obscured by a lot of garbage time, but on a macro level available projections are very favorable to Michigan.

[After THE JUMP: the run game, quantified]

Moving the (Stati)Sticks: Week 13

Moving the (Stati)Sticks: Week 13

Submitted by Adam Schnepp on December 4th, 2015 at 11:01 AM

adv stats header

adv stats osu win %

Looking at a 0% win expectancy might not cause you to slam your laptop shut in frustration if you’re thinking of it purely in terms of wins and losses, where a win would be 100% and a loss would be 0%. Unfortunately for all of us, that’s not how this stat works. You might want to pick your coffee cup up off your desk before you read Bill Connelly’s definition:

It is intended to say "Given your success rates, big plays, field position components, turnovers, etc., you could have expected to win this game X% of the time."

Before you put your mug back down, Michigan only had a win expectancy under 50% once, and that came against Minnesota. Also, their predicted win expectancy heading into The Game was 61%. The only silver lining is that this could be the game that killed the Dumb and Dumber “So you’re telling me there’s a chance” meme. No, Lloyd, there was not. There was literally no chance in a game that S&P+ projected Michigan to win by almost five points.

[After THE JUMP: Mathlete’s Four Factors, some depressing numbers, and some colorful charts to distract from said depressing numbers]