2012 Game Splits

Brian Fremeau, publisher of the FEI Ratings in college football and writer for Football Outsiders, has posted "game splits" for every game this season athttp://www.bcftoys.com/results/. What are game splits you ask? In short it's a statistical way to break up the total score difference in a game into contributions from the offense, defense, and special teams (in short...let's not get into the statistical details).

So, for example, rather than saying subjectively that against Rutgers our defense inflicted a Mack Truck of pain and our offense was a steaming..., we can more objectively say our defense was worth 27 points and the offense -25 points. See? It's just a numerical way of saying the same thing.

To save you time, here is a breakdown of VT's season...

Let’s start with the offense so we know what Pep Hamilton is inheriting if things were as bad as they seemed:

It should be pointed out here that these numbers are not opponent adjusted - posts of that nature will come in the future. For this one, just note that while scoring 37 on Bowling Green gets you a high offensive point value, that could just as easily be attributed to their defense being poor. In fact, one team's offensive point value will be the negative of their opponent's defensive value and vice-versa. Looking at their season averages, we actually did well against Bowling Green and were by far their defense's worst performance. Against Duke? We were slightly better than average.

Next let's look at defense:

Here it is easy to see how the mid-to-late season adjustments really paid off. We were Virginia's worst offensive performance of the season by a small margin, and easily Rutger's worst offensive performance (our defensive value was 25.5; the next highest against them was 10.4).

The state of our special teams is probably best for a separate post, but I think most agree with me that we are a fairly average special teams performer at this point. Here are the values:

Obviously special teams tend to affect the game less, so the magnitude's here are less than offense and defense. But they certainly center around 0, confirming my belief that there is nothing special about our special teams.

Those three are the core components, but a subset of each which is also posted would be field position and turnovers. So here are each of those as well, which again don't contribute large magnitudes but are interesting anyway:

Turnovers and field position are correlated, but still I think this highlights what we all saw against Clemson. Similarly the brutal interceptions against Pitt stick out as well. More broadly, we were on the losing end of both field position and turnovers more often than the winning end, and that's not a recipe for winning.

Finally, here are the three core components plotted together on a single graph:

As a final thought, beat Bama.

Tags: 

Comments

Nice Post

This is some really cool stuff. The defensive graph is extremely telling of how good a coach Bud is.

VERY COOL, question though:

What is defined as a "point" in this sense? For example, you say that against Rutgers "our defense was worth 27 points and the offense -25." Obviously 27 + -25 doesn't equal the 13 points we scored, does that mean that our special teams scored the additional 11 points? Or does that mean that of the 3 point difference, special teams accounted for 1 point? Or do these "game split points" not directly reflect the game's final score?

Thanks in advanced for the clarification, this stuff is very interesting.

Twitter me

Good question

There is an "expected points" component as well that I didn't put because it's not interesting (think of it as home field advantage...regardless of who is playing you already expect one team to outscore the other by a certain amount). The offensive split plus the defensive split plus the special teams split plus the expected points equals the score difference: 3 for the Rutgers game for example.

The only caveat to that is that he uses "non garbage time" score...depending on the score difference and time left on the clock, at some point he considers it garbage time as one team could not reasonably come back. He does not include points scored during this time in the final score. That affects two of our games.

For Rutgers, the true values are:

Offense: -23.5
Defense: +25.5
Special Teams: +1.0
Expected Value: 0.0

Those add up to 3, which is how much we won by.

Thanks! Very interesting stuff. Love to see more stuff like this.

Twitter me

If I am looking at this right...

...this means our defense lost us the Cincinatti and UNC games? Am I reading that right? I remember thinking that our defense looked horrible against Cincinatti and had a lot of missed tackles. i just remember being frustrated against UNC. They really came a long way as French pointed out in his latest French on the Bench article.

Defense did lose it for us those games

Given our style of play, and our team's identity, 25 points should be enough to win us any game. The 47th best defense this year was Kent State, giving up an average of 25.08 pts/game. Has this program ever been satisfied with the 47th best defense.

Twitter me

Great coaches are judged on the adjustments they make, whether it be at halftime or game to game. This just shows me how bad our offensive staff is and how good Foster and the defensive guys are. A team should be peaking at the end of the year, the defense was but the offense was terrible.

Rip his freaking head off!

Special Teams loses?

Clemson and Miami look like the special teams had the biggest impact.

I'd like to see this going back several years. Also would like to compare it to other "elite" teams.

Agreed!

These comparisons would be very interesting!

Twitter me