I'm a football stat nerd. I always have been. I was never able to play football, but I was always curious about different ways to analyze it.
These days in sports, we have more data than we know what to do with. Baseball is particularly egregious, but the data revolution has invaded football as well. But as I got into sites like Football Outsiders, I started realizing that some of the numbers they use are a bit convoluted. That's not to say they aren't valuable - They are - but sometimes they're hard to understand, and when they're hard to understand, they're harder to articulate to fans that aren't stat nerds like me.
I read Moneyball a few years ago, and I was fascinated with how Sabermetrics started out as simply a way to emphasize things that we already knew and could easily calculate, like on-base percentage over batting average, for example. So the thing I began to wonder was, can I do the same with football?
To me, the fundamental unit of the game is the possession: What do you do when you have the ball, and what do you permit your opponent to do? On offense, the objective is simple: Score points. Likewise, on defense, the primary objective is preventing the other team from scoring.
But we never - NEVER - see any measurement of teams' on-base-percentage, if you will. How often do teams score when they have the ball? How much do they score when they have it? How can we present this information in a way that's easily understandable? Why don't we see this anywhere? Or am I missing it?
That's what I'm going to do this summer with UCF Football.
I'm not pretending to launch some kind of revolution here. I'm reasonably sure someone else has come up with this way of measuring football this way long before me and has probably published it behind an absurdly expensive paywall somewhere. I just like doing the work and trying to see if the numbers are telling me something new.
Go look at any box score and you'll see the drive chart. All it tells you is how each team got the ball, when and where they got it, how far they went with it, how long they had it, and what they did with it. We know time of possession is a key stat, but what about the actual result of the possessions themselves?
Under Scott Frost, UCF didn't post a lot of time of possession, but man, did they score a lot of points. Back in the George O'Leary Era, the goal was nominally the opposite: Keep the ball as much as you can, and score only what you must.
So UCF provides an interesting case study into styles of football and how they differ. The questions are whether they differed at all, and to what extent?
To analyze this, we have to actually look at the drive charts and box scores themselves. There are only eight possible results of a possession:
- TD - Touchdown
- FG - Made field goal
- FGA - Field goal attempted and missed
- PUNT - Duh.
- FUM - Lost fumble
- INT - Interception
- DOWNS - Turnover on downs
- HALF - Possession ended with the end of a half
HALF is the complicated one that throws everything off. You could just run out the clock to end a game once it has been decided. You could run out the clock as a way of deciding the game (think the 4-minute offense). You could run out the clock at halftime. You could attempt an incomplete Hail Mary. You could run out of time getting the field goal team on. Any way you slice it, it's a wildcard, but it still has to factor into an analysis of possessions.
In defining a turnover, I calculated them as fumbles lost + interceptions + turnovers on downs + blocked punts + blocked field goals, since all of those result in less-than-optimal field position.
In addition, you have to compensate for defensive and special teams touchdowns. In those cases, the offense never actually takes the field, so technically it's not a possession at all. We'll factor that in as we go.
What I Wanted to Find Out
So what can we get out of this data? I wanted to find answers to the following questions:
- How often did UCF's offense put points on the board?
- How often did UCF fail to score?
- How often did UCF give the ball away?
I'll answer all of these in a second, but first, let's take the 2017 season game-by-game and break down the possessions:
|GAME||TD||FG||FGA||Punt||Half||Turnover||Def/ST TD||TOTAL||Offensive Points|
A few things to point out:
- The Offensive Points column is the number of points the offense actually scored as a result of touchdowns and field goals. Defensive and special teams TDs (and their resulting extra points - and I had to go back and look at each one individually) and safeties do not count.
- Tunovers = Fumbles lost + Interceptions + Turnovers on downs + blocked punts + blocked field goals, as mentioned above.
- I didn't differentiate between HALF possessions, like if UCF intended to run out the clock or not.
- I also didn't differentiate between types of turnovers. Fumbles and interceptions are genuine errors, but some turnovers on downs could be the result of either errors or just aggression gone awry. It's impossible to determine intention.
Now let's have some fun.
What I Found Out
How often did UCF score?
The offense's job is to score, or at least put you in position to score. For example, it's not the offense's fault if the kicker misses a field goal - the offense did its job, the kicker didn't. We could break that down even further, but let's try and keep it simple here. I want to know how many times UCF scored on its possessions.
So let's take the number of touchdowns and field goal attempts (field goal misses are not the offense's fault) and divide that by the number of possessions:
(77 TDs + 13 FGs + 5 missed FGs) / 176 offensive possessions = 54.0% Scoring Rate
Now let's say you think that field goals are not really in the control of the offense, because #collegekickers, and we only care about touchdowns:
77 TDs / 176 offensive possessions = 43.8% TD Rate
How often did UCF fail to score?
Failed Possession Rate
Now let's see the inverse: How often UCF's possessions failed. We'll remove HALF possessions, since success or failure of a possession that ended with the half is strategically subjective, and only count punts and turnovers:
(46 punts + 25 turnovers) / 176 offensive possessions = 40.3% Failed Possession Rate
What if you only want to see how often UCF turned the ball over? No problem:
25 turnovers / 176 offensive possessions = 14.2% Giveaway Rate
How much did UCF score?
Now we get into the Total Bases part of this analysis, where we figure out how many points UCF's offense scored and divide that by the number of opportunities it had to score:
576 offensive points* / 176 offensive possessions = 3.27 Points Per Possession
*Offensive points = Offensive TDs only + ensuing extra points + field goals. No safeties, defensive TDs, special teams TDs or their ensuing extra points are counted.
But What Does It All Mean?
So now we've got some data. But data without context is meaningless, so I went back in time and figured out these same stats for UCF over the previous five years, including the Fiesta Bowl year of 2013.
I wanted to use this five-year period because it had everything:
- 2013 was a highly successful season - one loss, a BCS bowl, and a conference regular season title (there was no AAC Championship Game).
- 2014 was a pretty darn successful year, more or less the minimum acceptable standard for fans - a share of the conference title and a bowl game.
- 2015 was the absolute bottom of the barrel.
- 2016 was a mediocre year, not up to fans' standards - around .500 with a bowl game loss.
- 2017 was ZOMG AWESOME.
Here's what I found:
|Year||W/L||Possessions||TD||FG||Punts||Turnovers||Def/ST TD||Half||Poss./ Game||Off. Points|
And the big breakdown:
|Year||W/L||Scoring Rate||TD Rate||Failed Poss. Rate||Giveaway Rate||Points/Poss.|
This is a limited sample size, so there are very few conclusions we can definitively draw. But based on what I see here, there are some magic numbers can start to tease out.
To have a successful season, UCF's offense would most likely have to reach the following benchmarks:
- Scoring Rate > ~33%
- TD Rate > ~20%
- Failed Possession Rate < ~50%
- Giveaway Rate < ~20%
- Points per Possession > ~2.0
Now, I know this is incomplete (Stop complaining, stat nerds). I really would like to do this for The American, and all of College Football for that matter, if I have the time. I'm not sure there's an easier way to do this (oh I wish there was though), or even if someone has done it before (if so, please let me know so I can give them some well-deserved credit). And if anyone else has an easier way to calculate this that accounts for HALF possessions, please let me know IMMEDIATELY.
But at least we can see some trends making their way into the light. It will take me some time to fully analyze it to my liking, but that's what summer is for.
Next time, I'll look at the defensive side of the ball using these stats to see what else we can learn.