I like to imagine college football as a set of swinging pendulums. With each drive you attempt to swing your ball as close as you can get it to scoring, and then if you turn it over or punt the ball your opponent has a chance to swing it back over to their side. Only when one team scores does the pendulum get reset and it starts back over again. What I set out to do was to put some numbers behind the mechanics of this pendulum.
Every post I write has to start out with a big thanks to CFBstats, who has provided a treasure trove of data on college football. Using that data I already looked at how having the ball at a given yard line and down affects the outcome of your own drive. In this post I wanted to look at how having the ball at a given yard line and drive outcome affects the outcomes of your opponent's next drive. In order to do this I had to create a dictionary of both the ending of your current drive and the ending of the next drive. Then I had to verify that the next drive listed was from the same game and that your current drive and the next drive did not result in the end of half. Once that was done I filtered for non-garbage time plays only, and then went about tabulating the average start spot of the next drive and the drive outcome. Now, we can look at some charts.
First, I wanted to see what kind of numbers involved with the "second level" drives. If I am looking at each yardline and outcome combination, I am inevitably going to have some small sample sizes. For example, no one from 2005-11 punted the ball after having the ball one yard from the endzone, and thank goodness for that. Here are the rest of the totals:
You should read this plot like this: This is the total number of opponent drives that result from you having the ball at a given distance to the endzone and ending your drive in a certain outcome. The only point that is missing is having the ball at your own 20 and ending your drive in a punt, there were over 10,000 opponent drives of those and I wanted to make everything else easier to see. For some of the outcomes it will be hard to draw too many conclusions, but I will make adjustments when it is necessary, for now I'll just keep all of the raw data.
One of the first things I wanted to see was how did ending your drive in a certain outcome affect your opponent's starting spot for their next drive. Here is a plot showing just that:
Just so we are clear, each grid represents how you might have ended your own drive, and the points represent the average starting spot of your opponent's next drive. The x-axis represents you having a play at that yard line in your drive that ended in a given outcome. And a value of '0' for you opponent's start spot means there were no instances of that occurring (looking at you Safeties). Here are some of my observations:
- No matter what yard line you have the ball at, if you score a touchdown or field goal then your opponent's average start spot is constant, which is good. Both of these result in kickoffs so they should be the same. Sanity check complete.
- You can see the difference between "having the ball at a yard line and ending your drive in an outcome" and "ending your drive in an outcome at a yard line" in the Downs grid. For the first 10 yards or so the slope of the line is about at a 45 degree angle. This means that the yard line you end on downs is generally the same yard line (going the other way) that your opponent starts on. After that though the slope slightly increases. This is because knowing you are at the 20 yard line and turning it over on downs doesn't mean that you turned it over at the 20, you had more plays available to gain more yards and move down the field to then turn the ball over.
- Everything else looks about as you would expect, anything I missed?
- Fumbling the ball in the redzone seems to not be as bad as throwing a pick, or at least your opponent will punt the ball back to you a little more often. I would guess this is because of the returns for interceptions near the endzone are returned farther than fumbles, maybe? Something for a future post.
- The plots look similar except for the dropoff in the percentage of your opponent's drive that end in field goals when you fumble the ball. Here is my explanation: Fumbles occur generally at or near the line of scrimmage, so your opponent probably gets the ball near where you fumbled it. When you throw a pick that throw is downfield, for the most part, so teams probably get the ball farther back than if they had recovered a fumble on the play. So having the ball near your own endzone and ending on a fumble probably means your opponent started with better field position. We can look at the start spot plots and sure enough that is the case, in regards to your opponent's start spot. I'm not sure if this explains the decrease in FG % when you fumble, but it could.
- What is also interesting to me is that if you fumble the ball there is a slight uptick in fumble percentage (green line) towards when you have had the ball near your own endzone. Could this be because the conditions that had a hand in causing you to fumble the ball (weather) also would make your opponent's fumble percentage higher, and having the ball near your own endzone means you have to have more plays to move the ball down the field? I think that is certainly a factor, whether or not its measurable is another thing.