View Single Post
Old 02-01-2013, 07:13 PM   #38
dallasoliver's Avatar
Join Date: Aug 2007
Posts: 297

Originally Posted by jma View Post
It all comes down to the judgments made about how to "weight" certain results over others. That master list, when you look at it, is pretty scary. But if it is at all related to TRN's ranking and you had the time to look into it, it would tell you a lot about how they weight certain results. Judgments are necessary to take lots of variables and produce a single list, but they also have significant influence on how that list ultimately falls.
I think that Internet discussion is healthy - and everyone is entitled to his or her opinion. But I wanted to make two comments.

(1) To address jma's comment above... I will state - emphatically - that there is no "weighting" of results at Tennis Recruiting. Every week, we take all results for a gender and pass them to our ranking system, and our system produces a single objective ranking from those results. There is no weighting - a win is a win. And every Monday, we throw out the previous week's rankings and start fresh with a new 52-week set of data.

(2) As a more general comment, no ranking system can get everything right., and most ranking systems do a better job with players who have long records. We like to think that our system does a good job of being predictive based on past results. Our system also employs a time decay factor so that recent play counts more than play from a year ago - to take into account that players improve/decline over the course of a year.

dallasoliver is offline