RB Improvement with Experience

OK, here's the update - I've got the last three season's (2007, 2008, and 2009) worth of games into my database and I can finally start to play around with some stats.

So now the question becomes, what is the best way to project a player's improvement?

I'm starting by looking at running backs and yards per carry. I am breaking down everything by years of experience. In the chart below, I added up all the carries for first year players, then added up all the yards that first year players gained, then divided yards by carries to get yards per carry for first year players. Then I did the same for second year players, third year players, etc.

The first 8 or 9 years of this chart looked like I expected, but there is a very unexpected spike in year 10. This is due to some pretty good year 10's lately from Fred Taylor, Sammy Morris, Kevin Faulk, and Ricky Williams. Those guys also got many more carries than other year 10 guys, so the average was heavily weighted towards their performances.

So to take the number-of-carries bias out of the equation, I just decided to average up all the individual yard per carries for each year of experience. That looked like this:

Hmmm... same shape, just the numbers are a little lower, owing to the fact that guys with better yards per carry tend to get more carries.

OK, lets try another approach. Lets see how individual players improve from year-to-year with experiance. In other words, how much does the average year 2 player improve over his year 1 performance, how does the average year 3 player improve over his year 2 performance, and so on.

Well, we're just not going to get over the late-career bias in this one, but I still think this one might turn out to be the most useful. This would give us the ability to use last season's stats to project a player's performance in the next season. However, no way I am projecting huge improvements for running backs going into years 9 and 10. (Sorry, LaDainian.)

Up next, we will look at the same charts ordered by age as opposed to experience, then we will look at ways to figure out which method is the most accurate.

Another interesting idea, if we have time, is to use a player's total carries entering a season to predict performance.