I can get sucked up into numbers. Does that happen to anyone else? No? Huh. Anyway, ever since I started looking at football from a "why & how?" perspective, the advanced stats have always interested me, especially Bill Connelly's S&P+, and Brian Fremeau's FEI. A lot more can be done on the pro side of things, with the resources and available data they have. One of my favorite blogs to keep tabs on is Chase Stuart's blog, Football Perspective.
One interesting thing I got from him is this idea of a knockout in a game of football. Though not what I would call an advanced stat, it tries to answer the question, "When was the game actually over, even if there was time left on the clock?" The simplest way to define a knockout (and the method I'm using in this post) is,
A knockout occurs when the winning team scores more points than the losing team's final score.
So, when Nick Marshall ran for a 19 yard touchdown in the third quarter vs Arkansas, Auburn knocked Arkansas out since that gave the Tigers 28 points and the Razorbacks would only score 21. When Marshall threw a TD pass to Sammie Coates to go up 10-0 in the first quarter vs LSU, LSU had been knocked out since it would only score 7 points by the end of the game. As soon as the winning team's score passes the eventual final score of the losing team, it's over.
Knowing this, we can track when each game was over and how much time was left in the game. A last second, game-winning field goal is still a knockout but there is no time left. If a team scores two minutes into the first quarter and goes on to shut out the opponent, there were 58 minutes remaining after that knockout.
Finally, we can average these times together for each team and find when their average knockout time was. Doing the knocking out results in positive time. Getting knocked out is shown as negative time. And this is a way to measure a team's dominance over its schedule. It answers the question, "On average, when did a team have its game in hand?"
The knockout times for every game of 2014 is here, and the full 2014 rankings are here, but the tables below shows the top ten teams,
|Top Ten Teams||Wins||Losses||Avg KO||rk||Avg KO in Wins||rk||Avg KO in Losses||rk|
and the bottom ten teams.
|Bottom Ten Teams||Wins||Losses||Avg KO||rk||Avg KO in Wins||rk||Avg KO in Losses||rk|
Auburn came in at #35.
Here's a look at when the knockouts occurred throughout the season.
As you can see, a lot of knockouts occurred in the last two minutes of the game. There's also a decent spike in the last two minutes of the first half. My guess is that a lot of scoring happens in those four minutes anyway, so a knockout is more likely to happen in that time. Plus, any lead change in the last two minutes is likely to be a game-winning score, too.
And now for the disclaimers.
This is not meant to be a stand-alone ranking. No, I don't think Marshall was the second best team in the country last year. After all, there is no strength of schedule adjustment to these numbers. Despite my numerical fanhood, my lack of statistical training means these numbers end here.
Instead this is really just another way to look at margin of victory. Compare my list to the MOV ranking found here. Pretty similar. After all, they share eight of the top ten, though not in the same order. Instead of measuring how much one team dominated the other, my ranking tries to show long one team dominated the other. In fact, if the winning points are scored with no time left or in overtime, the winner and loser get the same time, 0:00:00.
Finally, this isn't the best definition of a knockout. It's full of fatalism, assuming that the losing team was destined to score what they scored and once the winning team surpassed it, the game was essentially over. This method says that Auburn knocked Mississippi State out of their 2008 matchup after Wes Byrum's field goal in the second quarter, but we all know that the Bulldogs could have taken the lead at any time had their offense not been somehow worse than Tony Franklin's that night.
That's why this is labeled as Method 1. This method is relatively easy to do because it only relies on scoring summaries, data which is easily found on the internet (and easily manipulated when taken from sports-reference.com).
I have an idea for a better definition of a knockout, but it will take more data than scoring summaries. If I'm able to pull it off, I'll share it here and probably update it throughout next season. Let me know what you think!