Data Dump: Who’s Really Good and Who’s Really Bad? Let’s Get Nerdy With It!
I’ve been running some analysis this week using the following process:
- Calculating FBS teams’ offensive and defensive points per game against other FBS competition.
- Finding national averages in each of those figures.
- Defining Offensive and Defensive Coefficients to each team that will “normalize” their offensive and defensive points per game to the national averages.
- Using these Coefficients to Analyze games.
Here are some interesting tidbits regarding Conferences:
- Conference USA is the worst conference in the country according to pure offensive and defensive points per game statistics. Of the 12 member institutions only two (UCF and Tulsa) have an offensive scoring average that is higher than their points allowed average. In other words 10 of 12 Conference USA schools would lose their average game. The ACC is the weakest “big” conference with only 5 of 12 teams outscoring opposition averages.
- While we’re on the subject of conferences, the Big 12 fields the best offenses according to Offensive Coefficients. A lower figure is good here as it represents an outperformance (a higher scoring average) relative to the national average. With a league-wide average coefficient of 0.8997, the Big 12 is the only conference that as a whole scores more than average. In layman’s terms: a Big 12 teams points scored need to be multiplied by 0.8997 to be brought down to the national average, every other conference needs to be brought up to hit the national average.
- Defensively the SEC boasts the strongest figures in Defensive Coefficient where a high number is good. Points scored against SEC defenses need to be multiplied by 1.3743 to be brought up to the national average.
- If you subtract the Offensive Coefficient (a small number is good) from Defensive Coefficient (a big number is good) the strength of the SEC really shows through. With a Differential Coefficient of 0.2860 the SEC is the strongest conference with the Big 12 coming in second a 0.1156. The worst conference in America goes to the WAC with a Differential Coefficient of -0.1940.
On the team side of things here are some observations:
- Oregon has the nation’s best offense with a Coefficient of 0.5245. In other words, the Ducks score almost twice the national average. And to normalize their points relative to an opposing score their total should be multiplied by .5245. That may sound confusing but think of it this way: If Oregon is averaging almost 54 points per game (which they are) wouldn’t holding them to 35 points be a pretty good defensive effort? Sure the national average is just over 28 points per game. But relative to their season average 35 is a small number. That 35 points would be the equivalent of about 18 points against an average offense (35 points * Oregon’s Offensive Coefficient of 0.5245).
- Conversely UMass has the worst offense in the country with a Coefficient of 2.5237. If UMass scored the national average (28 points) that would be the equivalent of an average team scoring 71 points (28 points * UMass’s Offensive Coefficient of 2.5237).
- On the other side of the ball a larger Defensive Coefficientis better. Alabama and Notre Dame are tied for the nation’s lead with a Coefficient of 2.5560. In other words, a single point scored on the Irish or the Crimson Tide is the equivalent of 2.556 points score on an average defense.
- Points scored against Colorado – the worst defense in the nation – are discredited to the value of 0.5751 points each against an average defense.
Since a high Defensive Coefficient is a good thing and a high Offensive Coefficient is a bad thing, subtracting the Offensive number from the Defensive number can give you a pretty decent picture of how a team is performing.
But, these numbers aren’t full-proof by themselves. The most obvious flaw to this system (there are admittedly many) is the lack of accounting for difficulty of competition. By adding each team’s respective Conference Differential Coefficient to the individual Differential Coefficients we can give credit to teams who play in tougher conferences and discredit teams who do not. This is not fool-proof as every game is not a conference game but it does shore up the data.
This data yields a top10 as follows:
- Notre Dame
- South Carolina
- Kansas State
- Texas A&M
*BYU is a clear outlier here at 6-4. However they’ve managed to somehow win some games by huge margins and lose four games by close margins. This is a scheduling irregularity as it is the Cougar’s first year as an independent and thus no Conference Schedule weighting.
The real benefit to these Coefficients is “normalizing” individual games. In other words, they can be cross-referenced with actual scores to say “If team A had played a team with the nation’s average offense and the nation’s average defense – instead of team B – the score would have been X to Y.”
Some high points in this data include:
- Georgia’s 48-3 win over Vanderbilt (a respectable team with a stout defense and slightly below average offense) – if normalized – would have seen a final score of 83-3.
- Notre Dame defeated Purdue early in the season by a score of 20-17. Had Purdue boasted both an average defense and an average offense, the Irish would have suffered a 17-18 defeat.
- Not surprisingly, Oregon takes the cake in the highest normalized point total thanks to their 62-51 win over USC. If both offenses and defenses were normalized, the score of that game would have been 92-41.
What’s interesting is using the formulas to generate hypothetical scores in matchups. If you divide Team A’s Points Per Game by Team B’s Defensive Coefficient and then Divide Team B’s Points Allowed Per Game by Team A’s Offensive Coefficient and average the two numbers you can generate an expected score for Team A. Doing the opposite will generate a score for Team B.
So how about the Top five teams?
If all five teams played each other here are the outcomes you would see:
- Kansas State would go 1-3 with a 36-34 loss to Oregon, a 17-16 loss to Notre Dame and a 24-16 loss to Alabama. The Wildcats would defeat Georgia 27-24.
- Oregon would go 2-2. The Ducks would beat Kansas State 36-34 and beat Georgia 36-30. But the Ducks would lose to Notre Dame in a tight game (21.4 – 21.1 to be exact) and to Alabama by a score of 30-21.
- Notre Dame would defeat Kansas State (17-16) and Oregon (21.4-21.1) by narrow margins and defeat Georgia 17-14. But the Irish would lose to Alabama 15-10.
- Alabama would go undefeated, beating K-State by 8, Oregon bu 9, Notre Dame by 5 and Georgia by 10.
- Georgia would go 0-4.
But as I alluded to earlier in the week, all is not lost for Georgia. These numbers represent season-long data. Georgia is playing much better over the past three weeks.
Here is how each team has performed (Defensive Coefficient – Offensive Coefficient) in each of their last four games. Here come the Bulldogs, baby.
That’s all I got/