Being a sports statistician/economist blogger has to be the best gig on the planet if you make money from it.
Here is a blog worth visiting:
Brian Burke's NFL Stats site. Burke looks exclusively at the NFL and does a lot of regression analyis.
I came across Burke's site by Googling "NFL regression to the mean." I believe the NFL has more parity than any other league, and the season outcomes depend greatly on luck. Because of this, a team can go 1-15 one year and 10-6 the next. Teams regress to .500 very rapidly.
I was reading Dean Oliver's book about the NBA and regression to the mean. He found that winning franchises revert to the mean more slowly than losing ones. He determined what the chances were of a 25 win team winning 55 games the next year and so on. I wanted to do the same analysis for the NFL.
Burke found that Vegas tends to ignore regression to the mean in setting its over/under before the season. A team's performance in the previous season doesn't hold as much prediction value for their record the following season as it does in other sports. Burke did the math and figured pretty simply how to beat Vegas. Here's Sabermetrics (another great sports stats blog) review of it.
Burke kindly pointed me to another website that shows the average change in NFL wins per given record. It made a prediction for last year's teams against the Vegas over/under. This is what I wanted to look at, but I wanted to apply Oliver's logic and see how long it takes for teams to regress towards the mean. I also wanted to go back to '93, which is when the current free agency system went into effect. Standard deviation is also important, and other sites don't often address that.
So, I'll post my results tomorrow in an attempt to be a real sports economist/statistician blogger.