25 Sep 2013
by Brian Fremeau
College football can be unpredictable at times, but this past weekend was not one of those times. Week 4 featured a handful of exciting games, but ultimately very few surprises. Favorites won 42 of the 44 FBS vs. FBS games this past weekend. (The fewest number of upsets in any weekend that featured at least 40 games last season was seven.)
FEI projections correctly picked only 37 winners; that was better than expected as well. I include a Projected Win Expectation (PWE) for each game, and that data provides a baseline expectation of success rate for FEI game projections. Based on PWE, FEI expected to pick 34 out of 44 games correctly in Week 4. On the season, FEI has been expected to pick 138 out of 173 (.797) FBS vs. FBS games correctly. FEI has actually picked 137 right (.792).
A weekend like this one might have suggested that the FEI ratings would remain relatively stable, but that wasn’t the case. Michigan, Notre Dame, Michigan State, and Oregon State all slipped out of the FEI top 25 even though three of those four won over the weekend. Georgia Tech, UCLA, Baylor, and Texas Tech all moved up into the FEI top 25. Seven teams moved up at least 20 ranking positions, and Missouri vaulted unexpectedly into the FEI top 10.
The reason for most of these changes had more to do with the continued reduction in weight of preseason data in the FEI formula than with the results of the weekend. And since some teams have only played two FBS opponents, the weight shift for preseason data can have a big impact. At this point in the season, preseason data accounts for around 25 percent of each team’s rating. In three weeks, we will have eliminated preseason data from the formula altogether.
There were a handful of ridiculously uncompetitive blowouts as well. Ohio State and Miami outscored FCS opponents Florida A&M and Savannah State by a combined score of 153-7. Louisville beat Florida International 72-0. Mississippi State beat Troy 62-7. Baylor dominated Louisiana Monroe 70-7 ... through three quarters. The Bears didn’t score over the final 22:51 of the game and still won by 63 points.
Baylor has only played two weak FBS opponents so far (ULM and Buffalo), but the Bears have been spectacularly efficient in those games on offense. On non-garbage possessions, Baylor has earned at least one first down on 100 percent of its drives. They have earned 84.2 percent of available yards. They have averaged at least 10 yards per play on 52.6 percent of possessions. They have averaged 5.5 points per drive. They have had 16 non-garbage possessions start on their own side of midfield; 13 of those crossed into opponent territory, and 12 of them reached the end zone.
I synthesize offensive efficiency as a measure of actual drive success versus expected drive success based on field position. The full data sets for offensive and defensive efficiency will be published in three weeks, but it should come as no surprise that Baylor is the early front-runner in this overall offensive efficiency metric. The challenge in publishing offensive efficiency (and many of our stats) is that the output (percentage better or worse than average) is not as tangible for fans as standard stats.
I received an e-mail inquiry last week from reader Chris Healey, who asked if I could look into a new approach for offensive and defensive success rates. Chris specifically wanted to know how successful each team is in moving the ball without its opponent forcing a stop.
For instance, a team that strings together consecutive 70-yard touchdown drives and then throws an interception on the first play of its third possession will have earned 140 yards before being stopped. If its opponent goes three-and-out on its first three possessions of the same game, it will have earned fewer than 10 yards per stop to that point.
Chris wanted to know what this would look like over the course of an entire season. He suggested that this "Yards per Stop" stat captures offensive efficiency in an effective way because it combines scoring success with field position and tempo adjustments, but delivers it in terms of football’s basic unit of measure: yards. I like this concept because it reminds me of the origins of FEI. When I first started working with the drive data sets that would eventually produce FEI, I was interested in measuring each team’s ability to string together consecutive successful possessions. Yards per Stop has similar potential.
I started with 2012 data since we have sample size issues for 2013 this early in the season. I eliminated garbage time possessions and scores just as I do with FEI data. The tables below identify the top-10 offenses and defenses in terms of Yards per Stop; each team’s rankings in the FEI data I usually supply is also provided for reference.
|2012 Yards Per Stop - Top Offenses|
|2012 Yards Per Stop - Top Defenses|
As would be expected, the correlation of Yards per Stop with these other measures is quite high. Offensive efficiency has a .975 correlation with Yards per Stop. The correlation of defensive efficiency with Yards per Stop is .974. These are nearly equivalent measures and the key difference between them has to do with the relative value of yards in different areas of the field.
Is Yards per Stop more palatable than offensive efficiency? It would certainly be easier to grasp for casual fans as an entry into some of the other analysis we’re doing at Football Outsiders. I think that was Chris’ primary motivation for suggesting it. I like the way it helps me visualize the way some teams move up and down the field with ease, as well as those that don’t.
And in that spirit, let’s take a look at the top offenses and defenses in 2013 according to Yards per Stop. All early-season sample size caveats apply here, but take a look at Baylor’s efficiency through this lens. You want to get the Bears offense off the field? So far in 2013, you have to give up nearly 300 yards first.
|2013 Yards Per Stop - Top Offenses||2013 Yards Per Stop - Top Defenses|
|8||Ohio State||98.8||8||Washington State||21.6|
I’m not sure if I’ll be publishing Yards per Stop data with regularity throughout this season, but I’m interested in more feedback on it and any measures I do regularly publish. Feel free to offer suggestions in the comments below or via email.
The Fremeau Efficiency Index (FEI) rewards playing well against good teams, win or lose, and punishes losing to poor teams more harshly than it rewards defeating poor teams. FEI is drive-based and it is specifically engineered to measure the college game. FEI is the opponent-adjusted value of Game Efficiency (GE), a measurement of the success rate of a team scoring and preventing opponent scoring throughout the non-garbage-time possessions of a game. FEI represents a team's efficiency value over average.
These FEI ratings are a function of results of games played through September 21st. The ratings for all FBS teams can be found here. Program FEI (five-year weighted) ratings and other supplemental drive-based data can be found here.
11 comments, Last at 29 Sep 2013, 11:20pm by sinkerinthedirt