Wednesday, November 30, 2005

NFL: using statistics in articles

There is an interesting article on ESPN.com about the NFL. The article states 5 Sins of the NFL - 5 statistics that rate highest in correlation to losing in the NFL. While this is interesting, there is one big flaw in the article.

First, here is the link to the article:

5 Sins in the NFL on ESPN.com

To summarize, the five sins and their correlation to wins/losses listed in the article are:

1. Trailing after the first quarter (75 percent)
2. Losing the turnover battle (81 percent)
3. Allowing a 100-yard runner (75 percent)
4. Allowing more sacks (70 percent)
5. Losing time of possession (67 percent)

They researchers at ESPN.com did a fine job researching. Greg Garber did a fine job making the article interesting and fun to read. But their problem is that all of these 5 sins are related.

One of the objectives of teams that are ahead is to shorten the game. That means running the ball, taking time off the clock. Not throwing incompletions. Throwing short and high percentage completion passes are fine (screens, flats) because they are the same as running the ball as they keep the clock ticking. Meanwhile teams that are behind are often throwing the ball because they want to lengthen the game without the clock ticking to their zero. Most importantly, teams that are ahead don't need to score more to win the game. While teams that are behind need to score to catch up. So teams that are behind are willing to take risks. They are willing to risk throwing an INT if it means it increases the chance of scoring a TD. They want volatility.

Imagine how this
A team starts the 1Q by getting behind. Say they get behind by 2 touchdowns. That's Sin #1. The team losing starts to pass the ball more, thus increasing the chance that they will get sacked. You can only get sacked if you are trying to pass the ball, not when you are rushing. That's Sin #4. Teams that are ahead will run the ball more, and teams that are behind will run the ball less - all for obvious reasons. So the team that is already ahead in the 1Q is more likely to run the ball in the 2Q, 3Q and 4Q, and thus more likely to produce a 100-yard runner. That's Sin #3. The team that is behind and needs to catch up is throwing the ball, thus increasing the chances of a INT. That's Sin #2. And lastly, the team that has the lead in the 1Q is more likely to produce a higher Time of Possession in the latter quarters because they are chewing up the clock running the ball.

So you see, all those correlation numbers are correct, but once we get the first piece of information - Sin #1 (Trailing after the first quarter), the other pieces of information are not worth very much by themselves. These 5 sins are all really 1 big sin combined.

No comments: