FEI: Michigan #26
Similar caveats about this being early as my S&P posts. Link here.
For those who don't know, FEI is an advanced metric developed for Football Outsiders:
The Fremeau Efficiency Index (FEI) is a college football rating system based on opponent-adjusted drive efficiency. Approximately 20,000 possessions are contested annually in FBS vs. FBS games. First-half clock-kills and end-of-game garbage drives and scores are filtered out.
Basically it's drive-based, whereas S&P is play-based. I generally prefer S&P later in the season, but think FEI is less prone to zany results early on.
Big 10 Teams:
#2 Ohio State
#11 Michigan State
#25 Northwestern
#26 Michigan
#27 Wisconsin
#50 Iowa
September 28th, 2015 at 10:51 AM ^
Yeah, but are the injuries taken in account?
September 28th, 2015 at 10:53 AM ^
If only someone would start a thread and ask. Maybe I should make one to see if there's any news.
September 28th, 2015 at 10:55 AM ^
It only makes sense if you post it EIGHT times. Anything less would be candy-assed.
September 28th, 2015 at 10:59 AM ^
Attacking Injury updates with an enthusiasm unknown to mankind!
September 28th, 2015 at 10:56 AM ^
The 11am slot is open. Go for it.
September 28th, 2015 at 10:56 AM ^
I don't think you'll get enough attention. You may need to make 7 of them.
September 28th, 2015 at 10:53 AM ^
Could FEI also filter out an entire 2nd half clock killing???
September 28th, 2015 at 11:27 AM ^
It actually does. The entire second half of the Brigham Young game is thrown out by FEI.
Here's their definition of garbage time: "a game is not within 28 points in the first quarter, 24 points in the second quarter, 21 points in the third quarter, or 16 points in the fourth quarter."
So garbage time this week was the last 31:02 of the game.
September 28th, 2015 at 11:54 AM ^
They would be counting very few of of our drives vs BYU, therefore giving the game as a whole less weight than the others. I would think they should make an adjustment to the value of individual drives based on how much they were above or below the average number. Maybe they do, but I couldn't find any reference to it.
September 28th, 2015 at 12:16 PM ^
But we, as fans, think of games as the single unit to be analyzed. This system, not unreasonably, sees a drive as the single unit to be analyzed. Michigan scores a touchdown in the first quarter? That TD drive is a single event, not related to anything else.
Think of Michigan and BYU playing 9 "games" against each other on Saturday--4 ended in Michigan touchdowns, 1 ended in a Michigan punt and 4 ended in BYU punts. If Michigan plays 9 "games" against BYU and 22 against Utah, well, that's all we have. I guess they could count the drives against BYU double, but that's skewing the numbers in a different way.
(By the way, I said the last 31:02 was "garbage time" per FEI. I was wrong--it was the last 36:57, after Michigan went up 28-0).
September 28th, 2015 at 12:54 PM ^
inaccurate about any system that counts some games dramatically more than others, without being directly tied to the strength of the opponent. I understand that they do it for everyone, which mitgates the effect somewhat.
Another way to look at it is those extra drives against BYU did exist. They're choosing not to count them because of the situation, but if we did poorly on them or on special teams, they would have started counting our drives again. Rather than using the same calculations on those missing drives, they could factor in a multiplier for the drives they did count. If there were say 20 drives, but they only count 10, they could multiply those ten by 1.5. That still weights the game less than average, but not drastically. A number like that probably doen't skew the stats in the opposite way, while simply doubling them might.
Alternatively, they could a set a maximum and minimum that each game can count towards a team's total. Is there a reason not to add an adjusted measurement to the one the do now? They already tinker with it to make it more accurate.
September 28th, 2015 at 2:34 PM ^
They definitely should be weighted less. But they should have at least SOME weight.
I have a drive-built model that I am playing around with --- I mathematically handle things by weighing any drive where a team's win percentage is between 10% and 90% equally. For drives where a team's win percentage is less than 10% or greater than 90%, then you apply diminishing returns.
The important thing: no matter how many drives in a game, the weights for an individual game always sum up to 1. Each game, AS A WHOLE, gets treated equally.
FWIW, my rankings after 4 weeks. NO Bayesian priors at all in my model, so there are still some wonky results:
Top 10: LSU, Utah, Michigan State, Michigan, Notre Dame, Western Kentucky (?!?!), Iowa, WVU, Miami Florida, Baylor.
And --- the Ohio State University Buckeyes at #79!
September 28th, 2015 at 10:53 AM ^
I still don't know what to think about Northwestern...they beat Stanford and Duke, neither of which is a slouch, but are we really believing that Northwestern is a legitimately good team? I guess we'll find out first hand in two weeks.
September 28th, 2015 at 10:58 AM ^
You could say this about Michigan at times, but Northwestern seems to be a team that starts off hot sometimes and then falls on its face. I'm not sure if they just lack depth to make up for any injuries, or if they just don't adjust as the season goes along, but I'm taking their early season success with a grain of salt. I also think Stanford just really sucked that day.
September 28th, 2015 at 11:18 AM ^
However you slice it, looks now to be kind of a trap game the week before MSU. They showed a lot of mettle on the road coming back against Duke, who has a quality coach and a tough defense.
September 28th, 2015 at 11:03 AM ^
They are a confusing 4-0 team. Beating Stanford and Duke is impressive. But struggling against Ball State at home? Ball State gave up 36 points to Virginia Military.
The pessimist in me says be prepared for the Northwestern team who played Stanford.
Sent from MGoBlog HD for iPhone & iPad
September 28th, 2015 at 11:08 AM ^
Similar to us--potential to be very good but not clear they can consistently play their best.
Sent from MGoBlog HD for iPhone & iPad
September 28th, 2015 at 10:55 AM ^
I think there is some bias towards a pre-season ranking in FEI, and then week 7 that drops out.
September 28th, 2015 at 10:58 AM ^
That would make sense. Without that, I don't see how OSU can be #2.
September 28th, 2015 at 11:02 AM ^
Or maybe just the usual "we really don't know anything" that comes with every September.
September 28th, 2015 at 10:59 AM ^
Pat Fitzgerald benefitting from the Harbaugh proximity effect after speaking at Exposure U. Expect to see SEC coaches break ranks next year to experience the same results, with Bielema leading the charge.
September 28th, 2015 at 11:01 AM ^
They were crushed by both teams with a pulse that they played. Clemson has also not played anybody.
September 28th, 2015 at 11:15 AM ^
Problem with FEI right now is that I believe they still use a certain amount of data from last year until week 7 of this year. So UM is still burdened a bit from the Hoke era in FEI. I think S&P+ phased out last year's results this week. And yes, sample size does come into effect this early, but when you're play-based, you already have a pretty good sample size by week 4 as opposed to drive based in FEI.
September 28th, 2015 at 11:25 AM ^
favored in every game including OSU. We project out to a 9 win season which seems very doable at this point.
#TheHarbaughEffect pic.twitter.com/cL7oNrpFr8
— Drew Hallett (@DrewCHallett) September 28, 2015
I assume this is a S&P summary since it mentions S&P in the calculation.
September 28th, 2015 at 11:52 AM ^
If I remember how numbers work, that is a higher win probability vs. Sparty than either of @Minny or @PSU.
September 28th, 2015 at 11:17 AM ^
FYI- Utah is #15 to them.
Sent from MGoBlog HD for iPhone & iPad
September 28th, 2015 at 11:37 AM ^
If we beat Maryland handily I could see is moving into the top 10 in these rankings.
September 28th, 2015 at 11:42 AM ^
The estimate in the FBS Mean Wins column for Michigan moved over a half-point as well, which is pretty significant and now places us at a projected 8 wins - for the time being, in line with many of the expectations people had around here. That means, of course, that we would be 5-3 in the Big Ten in such a scenario, and compared to last year, a winning record in the Big Ten would be a success unto itself.
September 28th, 2015 at 12:01 PM ^
for this season, as Rudock is probably not going to throw that many picks again and the OL is very much improved with no signs of regression. If that BYU shellacking was our "ceiling," or floor-to-ceiling range puts us in a good position against everyone but OSU and MSU.
Northwestern at home and Minnesota and Penn State on the road are probably equal tests to BYU, if not more challenging. We let up on the gas the whole second half and still beat them 31-0. That predicts out to scores like 20-10, 24-14, 21-7 against those opponents. Offense may not be as dominant as they were against BYU but the defense isn't going to give up many points either. 9-3 is well within reach, even if our "ceiling" was the BYU game. But that ceiling could most definitely go higher as well.
September 28th, 2015 at 2:53 PM ^
They only use data from this season, including strength of schedule.
Michigan is currently #1.
September 28th, 2015 at 4:30 PM ^
Well obviously this is the right one then. Case closed.