landing spot. will be interesting to see how he does.
A quick guide to where my numbers come from and how they are calculated.
Where Does The Data Come From?
My sole source is the NCAA website, which hosts the play by play data for every year since 2003. 2004 and forward is nearly all there but 2003 is a bit hit and miss.
Thanks to MCaliber I can pull each week’s games down directly from the site into Excel where I translate the text into a variety of field and calculations that ultimately end up in an Access database. My tools are somewhat crude but they work and I can get what I need from them.
To data I have 992,624 plays in the database.
All games between two FBS teams. Any games against FCS teams don’t exist as far as I’m concerned.
Every play from these games are in the database but not all plays go into calculations. End of half drives are excluded as are any drives in the second half where one team leads by 16 points or more. Only plays under those circumstances are excluded, all other plays from those games are included.
Sacks are counted as pass plays and all fumbles are excluded due to their random nature.
What’s The Baseline?
Based on all of this historical data, each down, distance and line of scrimmage are given an expected value. For example:
1st and 10 from your own 20: 1.53 expected points
1st and goal from the 1: 6.48
Since each situation has a value, the value of any play is the change in value created. A 79-yard pass on 1st and 10 from the 20 to the other 1 is worth 4.95 points (6.48 points – 1.53 points). If the running back then punches it in from the 1, he is awarded .49 points (6.97 – 6.48). Touchdowns are worth 6.97 because they create the opportunity for the PAT which is successful 97% of the time. If the PAT is good, the values for the drive look like this:
QB/WR 3.95 points
RB: .49 points
K: .03 points
Thus the 7 points the offense generated are accounted for between the initial 1.53 from field position and the remaining 5.47 from play.
Even plays that gain yards can yield to negative expected point changes. A two-yard gain on 1st and 10 puts the offense in a worse spot than they began even though it was positive yardage. If a drive ends, all of the initial field position points are “left on the field.”
Let’s say a team hands the ball to their running back three times from the 20 and gains 3 yards each play. A punt on fourth and 1 means that the initial 1.53 expected points is now 0 so the running back now has three plays for –1.53 on the books. Third down plays are typically swing plays and can provide large deviations. Convert a lot of third downs and your value/play will be larger than your yards indicated. Fail on a lot of third downs and it quickly swings in the opposite direction.
What Adjustments Are Made?
We are finally getting to PAN, Points Against Normal. All previous calculations are done independent of opponent. Once several games are on the books in a season, we start to get a picture of who is good and who is not so we can make calibrations to performances.
The baseline as calculated above is adjusted based on the strength of opponents' rush/pass offense/defense. Last year Michigan allowed 0.19 points/rush, which [Ed-M: moment of shock coming] is really bad. So even if the opponent averaged 0.15 points per rush initially, their final tally was negative at –0.04 per play since they performed below what the average team did versus Michigan. A team would have to have an initial average of at least 0.20 to come out positive on the final scoring.
The final scoring is what I will refer to as PAN. It is a measure of actual scoreboard points above the average team you are. PAN can refer to a specific unit such as passing offense, total defense or kick returns, or for a team in total. It is also a good metric for comparing quarterbacks and running backs. It is only somewhat effective for wide receivers since they rarely yield negative plays.
What Does It All Mean?
Zero PAN means you are completely average. For a BCS conference team like Michigan this typically means bottom third of the league. A three-points swing in PAN typically equates to an additional win or loss over the course of a season.
+7 will put you around the Top 25 on the season
+14 is typically Top Ten and potential BCS game
+21 is best in class and probably playing for a national championship
The top rated team I have is Florida 2008. They finished +13 on offense, +7 on defense and +3 in special teams. The top Big Ten team is Ohio 2005 at +19 (7/9/3). The top Michigan team is 2006 at +14 (4/6/4). They come in at 50th overall in the last 8 seasons.
I will try and add relevant updates if more questions come up in the comments.
All numbers included in this preview are using my PAN metric, Points Above Normal. PAN is essentially how many points above an average FBS team was a team/unit/player worth. For reference, an average FBS team is approximately equal to Northwestern or a top team from the MAC.
All games against FCS teams are excluded, as well as any plays in the second half where one team leads by more than 2 touchdowns or any end of half run-out-the-clock situations.
Unless otherwise noted, numbers include adjustment for opponent, which are starting to be more accurate as most teams have played at least half their schedule vs in-conference or quality non-conference opponents.
Rush Offense vs Penn St
Michigan Off +7 PAN, 2nd nationally, 1st Big Ten
PSU Def +1 , 53rd, 6th
Michigan has been well above average in every game this year and four of the six outings have been very positive. Alabama’s unit has been largely average. It had three decent performances and two below average games.
With Denard missing significant time against Iowa, the Cam Newton express has pushed Auburn into the #1 rushing spot nationally but Michigan is still a strong second. Cam also pushed Denard out of the top spot in the overall QB ratings but like the team rushing, Denard is still holding strong at #2, with all of Michigan’s +7 coming from him.
Hopkins has been the most efficiently productive back, averaging +1 in the games he has gotten carries, while Vincent Smith as netted out to +0 and Shaw is at –1.
With all of the injuries, Penn St may be able to slow the Michigan ground game but it should still be a big advantage for Michigan.
Pass Offense vs Penn St
Michigan Off +5, 16th, 2nd
PSU Def +2, 37th, 4th
Although the value and ranks are higher for the Penn St defense you’ll notice the large negative performance posted by the Nittany Lions when they faced their only other mobile QB against Illinois. Illinois had by far its best passing day of the season, leaving Penn St with an ugly –10 for the day.
Denard currently ranks 7th among Big Ten QBs in passing value but the team rating is much higher on the fact that Michigan is currently #2 nationally, having only allowed about 1.5 points lost due to sacks for the full season.
Hemingway, Roundtree and Stonum are presently ranked 10, 14, and 19 respectively among Big Ten receivers. Indiana is currently the only other team to have three receivers ranked in the top 20 in the Big Ten.
Even without an Illinois-level meltdown for the Nittany Lion pass defense, Michigan should still expect to come out ahead in the matchup through the air.
Rush Defense vs Penn St
Michigan Def –2, 93rd, 9th
Penn St Off –2, 96th, 9th
Two groups don’t get as evenly matched as the Penn St ground game and the Michigan rush defense. Both have the same value, nearly identical national ranks and both are #9 in the Big Ten. They both even have one terrible game (Iowa for PSU and MSU for Michigan) to go with a couple decent showings and a bunch of below average games.
When Royster has gotten carries he has been successful, ranking second individually among Big Ten running backs. Penn St just hasn’t been able to get him consistent carries and there hasn’t been any support behind him.
This shouldn’t be a matchup that is capable of killing the struggling Michigan defense. As long as Michigan can keep them from getting Royster a ton of carries a draw would be a likely outcome and a welcome showing.
Pass Defense vs Penn St
Michigan Def –3, 99th, 9th
PSU Off –3, 90th, 10th
Weakness on weakness. Just like the rush defense, the pass defense is in a pretty equal matchup. That is assuming that Penn St can match the production of their concussed true freshman quarterback with an upperclassman walk-on.
Michigan is hitting the critical portion of their schedule where they need to pick up some wins against the Big Ten’s three lowest rated quarterbacks. Robert Bolden is likely out on Saturday but he was already ranked dead last in the Big Ten quarterbacks with –3 per game.
Penn St has only produced one above average passing game of the season and that was against Kent St in Week 3. In about a half’s worth of time last weekend against lowly Minnesota, presumed Michigan starter Matt McGloin was –2.5 (unadjusted) despite throwing a pair of touchdown passes. In mop-up duty throughout the season, Kevin Newsome was –1 (unadjusted) in total for both rushes and passes.
No matter who the quarterback ends up being, these are the matchups over the next three Saturdays that Michigan’s pass defense has to make headway on to finish the season out strong.
Special Teams vs Penn St
Michigan –1.5, 108th, 11th
PSU +3.8, 14th, 1st
After four nice looking matchups, we get to the ugly one. Penn St has been very solid on special teams this season. The big score against Illinois is from two fumbled Illini punt returns. The Nittany Lions aren’t much of a threat to break a long return, but their punt and kickoff units, along with their kicker, have been among the best in the nation this season.
Penn St has the number one ranked KO cover unit with only one return past the 26 allowed on the season, and even that only went to the 35. Michigan’s strong offensive advantage should help neutralize any field position gains from Penn St’s good kicking units.
Prediction Almost Certain to Cost You Money if Taken Seriously
Michigan 35 Penn St 24
My model is calling for something closer than this but of the four major units, there is one great (Michigan Offense), one average (Penn St Defense) and two bad (Michigan Defense and Penn St Offense). Also part of the closer score is the big special teams disparity. I don’t think that will come to fruition because Penn St’s strength has come in the form of defensive field position but Michigan has proven this year that field position is largely irrelevant for their high-powered offense.
Elsewhere in the Big Ten
Iowa 28 Michigan St 26 – Sparty’s dream season could certainly clear it’s last major hurdle but I do think this is where it ends.
Northwestern 35 Indiana 31 – Should be a fun one to watch even if there isn’t much on the line.
Ohio St 42 Minnesota 17 – Another easy one for the Buckeyes
Illinois 27 Purdue 13 – The [NAME REDACTED] redemption tour continues
Utah 35 Air Force 31
Georgia 31 Florida 28
Upset Special: USC 35 Oregon 34 – Despite the gaudy numbers, my database is still not sold on the Ducks due to their relatively weak schedule.
It is our nature to assign narratives to games. Against UConn it was brutal offensive efficiency. Against Notre Dame it was the first appearance of clutch Denard. UMass and Indiana it was praying our offense would have the ball last and Bowling Green was relief that we finally took care of business.
For me, Michigan St was a game of 7 plays. There were seven big plays on Saturday, and all of them went Michigan St’s way. There was plenty that didn’t go right in between, but those seven plays, but the seven plays masked what in some ways was a better performance than it felt, and in some ways worse. Denard’s three interceptions, two in the end zone (-10 points). Edwin Baker 61 yards, +5. Le’Veon Bell 41 yards, +4. Cousins to Dell for 41 yards +4. Cousins to Dell again, 44 yards, +3. Those seven plays, a 26 point swing. Only two plays for the offense +3 or higher, no running plays worth more than 1.4 points. All of the big plays went Sparty’s way.
What we hoped was not true now appears to be so: this team will not win Big Ten games without exceptional offensive performances. We are who we are at this point.
Rush Off: +5
Pass Off: 0
Rush Def: –15
Pass Def: –10
Field Position: Michigan +4
Denard was +3 on the ground and a +1 through the air.
Shaw and Smith were both +0 and Hopkins was +1 on two strong carries.
Cousins was +13 combined.
Baker was +7
Caper was +3
Bell was +7
I went 4-1 against the spread last weekend in my picks. I did not pick MSU, Purdue, Wisconsin or S Carolina to win like three of them did, but all four managed to cover as I predicted. Only Indiana let me down. The game went down as I predicted but the Indiana offense couldn’t quite put together enough garbage time points to cover the spread.
Michigan’s poor showing on Saturday, along with increases from Illinois, Ohio St, Purdue and Wisconsin, move the win projections down a game and a half or so. That is a huge drop for one week.
Win odds drop across the board, Illinois is probably over valued but after back to back strong showings they look to be better than expected this season. Wisconsin is starting to come closer to conventional feelings, but is still too high.
Team – National Rank – Win Odds
Iowa – 26 – 54%
@ Penn St – 57 – 59%
Illinois – 23 – 48%
@ Purdue – 77 – 73%
Wisconsin – 52 – 72%
@ Ohio St – 9 – 17%
Projected Big Ten Standings
1. Ohio St
2. Michigan St
9. Penn St
Just like last week, Illinois and Wisconsin feel like they are reversed, but to date, the computer likes what Illinois has done a lot and still isn’t sold on the Badgers.
It is a statement perpetuated on many outside of Wolverine-fandom in response to the 2010 start, and the deep dark fear inside the hearts of many Michigan fans: This year isn’t going to end up like last year, is it?
The argument for "Yes" usually boils down to the only teams Michigan has beat this year are the same ones they did last year (more or less) before the fallout, oh and because after five games we had a hyped young quarterback last year as well. The response from Michigan fans is subsequently, "Yeah, but Denard!!!"
Until we play a few more games and win one that we didn’t last year, we’re stuck answering the question in purely philosophical form. And who is better at throwing some numbers out there and seeing what sticks than The Mathlete?
Here is the normal disclaimer/overview of what I do for the uninformed:
All numbers included in this article are using my PAN metric: Points Above Normal. PAN is essentially how many points above an average FBS team was a team/unit/player worth. For reference, an average FBS team is approximately equal to Illinois or a top team from the MAC.
All games against FCS teams are excluded, as well as any plays in the second half where one team leads by more than 2 touchdowns or any end-of-half, run-out-the-clock drives.
For this particular exercise I will look at this year’s performance-to-date through two different lenses: 1) raw performance with no adjustment for opponent and 2) an opponent-adjusted view using how that opponent has performed this year to date. Normally I would forgo the unadjusted view to do a comparison but it is still early enough in the season that both views can provide perspective.
The Matchup: Offense
Let me just kill the suspense right now: this offense is better than last year’s. Shocking, I know. Through four FBS games this year, Michigan is averaging an unadjusted +23 PAN per game, +13 rushing and +10 passing. In the four FBS games Michigan won last year, Michigan was +12 overall, +8 rushing and +4 passing, and it’s pretty safe to say that UConn is a solid step up from Western Michigan and BG is probably a slight step up from Eastern.
To put more focus on the magnitude of this season's success, look at last week against Indiana, where the Michigan offense posted a +33 on only 44 plays. The 0.75 points per play is higher (by 10%!) than any other performance in my database, which stretches back to 2003. In fact, Indiana, Bowling Green and UConn are the three highest-rated offensive performances from Michigan I have on record. Western and Eastern Michigan were the only games last year that ranked higher than any game this year (Notre Dame is behind them).
Although impressive under any circumstances, those numbers were all without adjustment for the respective strengths of opponents' defenses. When you look at how Michigan’s performance compares with other offenses that ND, BG, Indiana and Uconn have faced this year, Michigan still comes out pretty well. All four games are at least +6 PAN and the average is +15, with +8 coming on the ground and +7 coming through the air. Last year in the comparable games, Michigan was +8 with +3 coming on the ground and +5 through the air.
Based on the sets of numbers, Michigan initially has been 7 to 11 points-per-game better than year’s offensive unit. This represents a very high level of play.
The Matchup: Defense
Unadjusted, Michigan has allowed +9 PAN per game this season. Almost all the damage has come through the air, and almost all of that was against Indiana. Excluding the Indiana game, the number was +6, with the damage split almost evenly between rush and pass defense. The Hoosiers' performance was +17 PAN with –5 on the ground and +22 through the air. This pushes the overall numbers to +9 with +8 coming through the air.
In the same games last year, Michigan’s defense was much more effective. Through four games, the defense held opponents to –7 PAN and was –6 against the pass. The defense moved to the middle through the rest of the season, finishing –2 PAN on the year, with –1 apiece coming on the ground and in the air.
How you evaluate this year really depends on good you think Indiana’s offense is going to be. If they continue to have success in Big 10 play, Michigan’s defensive prospects could be trending to on-par or slightly better than last year. If you think the Chappellbomb will be a dud against the rest of the Big 10, then last year’s performance is probably a best-case scenario.
One thing to consider about this defense is that its traditional stats are going to look bad no matter what. Based on the pace and success on the offensive side of the ball, Michigan is going to face more aggressive versions of their opponents, and they are going to face them on more drives, especially if the offense keeps scoring within the first minute of touching the ball. Everything you see from me will be adjusted to account for the pace. Remember: just because we gave up a ton of yards, it doesn’t mean that we had a bad day.
I am giving the defense an incomplete so far. Until we see how we fare against MSU and how Indiana does against Ohio State, the verdict is still out. If the defense can hold serve occasionally against Sparty, and Indiana can find some success against the Buckeyes, then the defense should at least be good enough to let us stretch a lead in a few games. If MSU torches us and Ohio St shuts down Chappell and Doss, we could be in for a full season of excruciatingly exciting games.
Our health, especially at key offensive positions, remains good.
The offense remains highly potent against the top tier Big 10 defenses.
The Indiana game was more of a reflection on Indiana’s great passing attack, and not our poor pass defense.
Although it doesn’t look like the defense has progressed like we had hoped (or maybe at all), the dilithium-powered offensive quantum leap has moved this team well beyond last year’s. There are still plenty of question marks out there, but it looks like until we face Ohio State’s defense to end the regular season, a Denard-led offense should be the best unit on the field. That fact alone should make a 2009 like swoon all but impossible. How much better is a question of defensive progress and Denard’s ability to shine as the defenses get better.
All numbers included in this preview are using my PAN metric, Points Above Normal. PAN is essentially how many points above an average FBS team was a team/unit/player worth. For reference, an average FBS team is approximately equal to Illinois or a top team from the MAC.
All games against FCS teams are excluded, as well as any plays in the second half where one team leads by more than 2 touchdowns or any end-of-half, run-out-the-clock situations.
Post Game Notes
The numbers predicted a one-possession win for Michigan and that’s largely how they played out. I audibled against the numbers and said the score might be a bit lower because Michigan might try a UConn game plan, slowing it down and limiting possessions. If that was the gameplan, no one told Denard.
Running the ball was Michigan’s obvious advantage coming in. I projected its worth to be in the range of 9 to 15 points for the game. We ended at the high end, with +16 PAN for the game on the ground. Denard put up his usual +12, while Vincent Smith put up all of his +4 on the long TD run. In four qualifying games Denard now has 4 of the top 14 rushing performances of the season in PAN. Bryce Beall from Houston is the only other player in the country to have two top-30 performances.
The big separation came in the passing game. Every time Michigan dropped back to pass it was basically worth a point. Michigan was +17 PAN on 17 attempts and Denard was +19. In terms of overall quarterback performance, Robinson’s three top-15 performances (the BG game still cracked the top 50!) is compared with only one other player with multiple top-15 appearances.
Ben Chappell's +27 performance was the second-best overall QB performance of the year, and he now has three top-40 games. Robinson and Chappell are the #1 and #2 rated QBs in opponent-adjusted PAN so far this year. Chappell’s number may very well come down as he faces defenses tougher than those of Towson, Western Kentucky, Akron, and Michigan, but there is no doubt that he is an exceptional player.
Indiana’s ground game was labeled by me as a pillow fight going in, and you could say that it ended like that. However, it was a pillow fight that Michigan won. The Hoosiers ended the game –5 on the ground as contrasted with a +22 through the air. Every time Indiana ran the ball it was a time they didn’t throw the ball, and a win for Michigan. At least for Indiana’s sake, the running game kept Michigan from pinning their ears back and rushing the passer…with three.
On predictions outside of Michigan, I almost called the OSU-Illinois score outright, but correctly had Illinois covering. I correctly had Iowa winning but did not have them covering, and I almost called the Michigan St.-Wisconsin score. Minnesota couldn’t put a last minute rally together to make me correct but I did correctly have them covering. I did not see Alabama’s domination over Florida coming and the Stanford over Oregon pick was looking great at halftime but looked terrible by the end. Pretty nice week for my Big 10 picks but I missed big on both of the national games.
Even with all of the struggles on defense, the season total projection keeps on rising. My team ranking has now eliminated the pre-season component and is made up entirely of in-season performance, with 50% of the opponent adjustment coming from 2009 performance and 50% coming from 2010 performance. For some teams there is a wide variation on which form of opponent adjustment is used. Michigan looks much better (ranked #12) based on in-season adjustment, but is still #23 when adjusting opponents based on their prior year success. The hybrid of the two has Michigan a full game ahead of where they were projected going into Indiana.
10 wins is now projected to be the most likely scenario and the odds of running the table have risen to 1 in 18.
Michigan State - 20, 60%
Iowa - 15, 56%
@ Penn State - 48, 66%
Illinois - 46, 83%
@ Purdue - 80, 90%
Wisconsin - 60, 92%
@ Ohio State - 14, 37%
Between the loss to Michigan St and the move to eliminate the preseason portion of ratings, Wisconsin has taken a massive beating in the numbers this week. The move is almost certainly too far but we will find out more as the Big 10 season progresses. Illinois and Michigan St were the big movers up, while Penn St, Wisconsin and Ohio St all dropped back after worse-than-expected showings on Saturday.
Projected Big 10 Standings
After last week things looked like three tiers with a mess in the middle. After this week, things are starting to separate a bit.
- Ohio St
- Michigan St
- Penn St
Switch Wisconsin and Illinois and this seems pretty reasonable, even after so few matchups against quality competition overall. Penn St could be higher but this doesn’t look like a season where defense alone is going to get you very many Big 10 wins.
|8||Oklahoma St||Big XII||15.87|
|12||Texas A&M||Big XII||13.85|
|14||Ohio St||Big Ten||13.51|
|17||Kansas St||Big XII||12.61|
|20||Michigan St||Big Ten||11.99|
On a purely data-driven model like mine there are going to be some oddballs, especially since this is more of a power poll than a rankings. Teams are rewarded strictly for how they performed relative the competition, without regard to win or loss.
Of the biggest head scratchers, Stanford #3 and ahead of Oregon, the Big XII seems highly overvalued and Notre Dame checking in at #25 seems a bit crazy at first, but they went to the wire with both Michigan schools who are obviously both undefeated and their other loss was to the highly rated Stanford squad.