*Team Aging Curve

[speculadora]

A brief explanation for those unfamiliar with what an aging curve is: for teams that played in consecutive seasons, you take the change in some number and average that out in certain intervals – in this case season. So essentially you could look at, say, the typical increase in receiving yards from a player’s first season to his second to project future performance. Rather than look at that on a player level (because data collection would be a bitch), I’m going to look at it on a team level.

What I did was, for every season in a team’s history, I collected their win percentage in the following season and took the difference between that and their previous season's win percentage. I then averaged those differences out to get an expected change in win rate from one season to the next. The nice thing is we can group this by season, team, number of wins, and general team quality to get all kinds of different outputs. Of course some of these will end up being useless because every team plays the others twice, but it’ll be nice to have the data for the future.

The first thing I tried was grouping by season which, didn’t produce much of anything because of the whole balanced schedule thing. In any event, here are the results of that:



Next I decided to look at it on a team level. Have any teams been on a steady rise/decline? Are certain teams consistently outputting similar records? This was a little more interesting. Arizona, somewhat predictably, has been regressing at the greatest rate. After their initial run that featured three seasons of fairly similar records, they have been on a downward trend culminating in a 5-9 finish in Season 8. The most consistent team from season to season has been the Otters, averaging a change in win percentage of just over 1% each season. Lastly we have the Second Line who have been on a strong upward swing ever since Bovo took over, averaging what amounts to an extra win each season. That number is sure to look even better following this season.



Let’s go a little more granular now and look at the change in win rate by previous season’s record. This is about what you would expect with the exception of the dip at the 3-win mark, skewed by the Second Line and Yeti both failing to improve upon the only 3-win seasons in league history. Unfortunately, most of these samples are pretty small, but we still get some sense of what’s going on.



If we ignore the 3-win outlier, the trend looks something like this:



We could also knock out the 2-win and 10-win data points that seem out of place to get a trend line that fits a little bit better, but arbitrarily removing data points for that reason is a pretty bad practice. In any event, we get a trend line for expected record based on a previous season. Expected Increase in Win % = -0.0234*PreviousWins + 0.168. So a 7-win team, for example, could would project for just about no additional wins. Which we can obviously identify as a flaw in taking this equation as gospel. It would be a much better fit for data points on the extremes, like 1 or 12 wins. A 12-win input would output an 11.3% decrease in win percentage (about 1.5 wins), for example.

The last thing I want to do is look a little more generally at team records and group teams together by winning percentage. Essentially all teams with losing records will be grouped and all teams with winning records will be grouped. Here’s what that table looks like:



So we see that, generally speaking, losing teams typically improve by a little more than a win. Teams with an even record improve by a little over half a win, and – perhaps the most surprising – teams with winning records drop of by a little less than half a win per season. The moral of the story being that it appears to be much simpler to stay competitive than it is to blow things up and rebuild.

[bovovovo]

10/10 good numbers