Tuesday, October 29, 2013

Interesting USTA League Stats - Histogram of NTRP ratings down to the tenth

Next in my series of posts on interesting stats from the data I collect for the Estimated Dynamic NTRP Rating reports I generate, I took a look at how many players are at each rating level down to the tenth.

Why down to the tenth?  There was a discussion on Talk Tennis about the issues with the new "plus" leagues where it was hypothesized that there were clusters of players in certain ranges and how, naturally or otherwise, a slight movement of the range for each NTRP level might result in better alignment with those groups of players.  With my ratings, we should be able to chart the ratings rounded to the tenth and clearly see if and where there are any clusters.

Here is the chart:
From this, it doesn't appear there are any abnormal peaks or valleys that would indicate such clusters exist.  There is a peak at 3.2 a slight dip at 3.4, then another peak at 3.5, but this could potentially be explained by the fact that this chart includes both men and women, and perhaps women have their peak at 3.2 and men at 3.5 and when put together, you get the chart above.

So we don't see any odd clusters.  It could be that my ratings aren't accurate and so don't show them, but given my accuracy in predicting DQs and bumps, I think it is more likely that my ratings are pretty accurate and there aren't any significant clusters of players around a particular rating other than the normal large group of players at the NTRP 3.5 and 4.0 levels.