Week #3 Statistics and FEI Prediction for Notre Dame
Synopsis: Michigan is ranked #49 in the FEI. M is ranked #23 nationally in scoring offense and #59 in scoring defense.
The FEI Forecast for this Saturday is: Notre Dame 37 Michigan 14 with a 93% Probable Win Expectation for ND (yikes!). Notre Dame is ranked #6 in the FEI.
Fremeau Efficiency Index: The FEI is a drive based analysis considering each of the nearly 20,000 drives each year in FBS college football. The data is filtered to eliminate garbage time (at the half or end of game) and is adjusted for opponent. A team is rewarded for playing well against good teams (win or lose) and is punished more severely for playing poorly against bad teams than it is rewarded for playing well against bad teams. (This is why M was ranked relatively high after that loss to Alabama and actually fell 13 places after the blowout of UMass.)
National Rankings: The rankings for offense and defense are based on scoring (yardage statistics are inherently flawed). These are simply raw numbers without any adjustments for opponent, garbage time, or anything else. The data is from TeamRankings and includes only games between two FBS teams.
FEI Details: Here are the FEI numbers for Michigan ( Football Outsiders FEI ). FEI is predicting a 6-6 season for the Wolverines (FBS Mean Wins = 5.8). Like most predictive tools, the FEI is less reliable at the beginning of the year because there is so little data. Last year after 3 games (WMU, ND, EMU), FEI had M ranked #19 and was predicting a 9 win season.
Points Per Possession: The 2 charts show the raw data for offense and defense. The charts both indicate significant improvement in each subsequent game. Of course, this may be due in great measure to the fact that the opponent has been weaker in each subsequent game.
Using Scoring Offense and Scoring Defense National Rankings for the past 5 years (FBS AQ teams only), this table shows the percentage of teams that finish the season with a +WLM and a +5 WLM. For example, teams that finished in the Top 40 in both offense and defense had a 100% chance to be +WLM and an 82% chance to be +5 WLM (9-4 or better).
September 18th, 2012 at 4:34 PM ^
Interesting, but this same thing produced a prediction of 25-23 Spartans last week.
If computers could predict college football outcomes Vegas would be bankrupt.
September 18th, 2012 at 4:44 PM ^
Eh, FEI like any other predictions early in the season are based largely off of nothing.
September 18th, 2012 at 5:32 PM ^
September 19th, 2012 at 3:52 AM ^
Michgan definately won two more games than the statistics would suggest...not sure I would count Notre Dame and Virginia Tech as repeatable outcomes given the teams overall performance in those games.
September 18th, 2012 at 5:50 PM ^
predicts a 14pt win by ND (11 points from the predictor ratings plus 3 for home field). However, the Sagarin also lacks validity at the beginning of the season (before all teams become interconnected through intermediary games).
So, if I were to bet on the game, I wouldn't be too eager to use this rating method either.
September 18th, 2012 at 6:13 PM ^
Seriously...the only thing that makes sense in there is the 3 pts for homefield advantage. Otherwise, as others noted, it's just too early.
We'll know something worth knowing after ND. Not before.
September 19th, 2012 at 12:12 PM ^
Even if we beat ND, we're probably going to be underrated by the computers all season because of our margin of defeat to Alabama. The computers can't understand how good Alabama is, and that even a legit top 10-15 team could get trashed by them.
September 18th, 2012 at 10:25 PM ^
UConn at home and will lose to PSU on the road. I not buying this.
September 19th, 2012 at 10:35 AM ^
beating a gimmicky AF team is worse then getting crushed by Alabama and then beating up on a crappy UMass team is worse than both?
FEI doesn't make sense. It may be useful at the end of a year, but it is not at all usefull now.
September 19th, 2012 at 11:04 AM ^
Data: we has none
Nor does the FEI
Such things will be a lot more predictive towards the end of the year.
September 19th, 2012 at 12:57 PM ^
September 19th, 2012 at 2:08 PM ^
NEED MOAR DATA!
Looking at opponents, hard to know exactly what's been going on with ND. Purdue is looking legit and MSU's D has been legit 3 years running now so we know ND's offense is good because they moved the ball against both teams. They are going to score points.
How is their D? MSU shit the bed but their QB is terrible so hard to say how much was self inflicted vs. just being terrible. If our Oline can give Denard some time, we should be ok.
I think it is a repeat of our last 3 games where whoever has the ball last, wins. We will both put some points on the board, that's for sure.
September 21st, 2012 at 1:06 AM ^
This formula has Michigan 12th nationally overall - 3rd in offense, 65th in defense.
FEI is a drive based metric, while the data for S&P+ is derived from each individual play. These distinctions lead to some inherent differences.
This makes FEI more results-based - was your offenseive drive successful in scoring / was your defensive drive succesful in preventing them from scoring. S&P+ is more process based. Taking each play and incorportaing both success rate and PPP (points per play), an explosiveness measure, means to answer whether or not you're doing the things that will lead to success long term. The key components in success rate (staying ahead of the sticks by getting to 2nd and medium and 3rd and short and converting 3rd downs), and PPP (getting chunk yards so you don't have to be perfect in so many consecutive events) are both factored in.
Their combined F/+ ratings, a blend of the two rankings, have Michigan 24th.
Im not sure which data set is more trustworthy with a limited sample, but at this point in the season I wouldn't take a particular ranking as gospel, rather using it as a confirmation tool - i.e. the rankings agreeing that Notre Dame's defense looks legit and Michigan's doesn't.
Comments