Back to the actual tennis ( phew say many ), pity about Laura, but she still has playing so well so comparatively recently to hold on to and hopefully give her the confidence that she is moving in the right direction. Good to see Harriet get a R1 win, which will move her closer to her CH 345.

DF wrote:

---

Katie B defeats Vrljic 6-2 6-1
A fine win 

---

 Vrljic retired last week, so may be was an easier proposition that first thought, hopefully Katie can dish out a bit of Brit revenge in the next round !

While there will clearly be legitmate arguments re UTR methodology and indeed the whole concept of trying to bring together such disparate groups under one rating system, we have here an example of how I said data results can be misinterpreted.

Player A with a lower UTR rating than Player B in one particular match does not show a UTR "error". It just indicates that the lower rated player won, which of course will still often happen no matter even if the concept and the methodology are absolutely fine.



Err... The statistical argument is a bit more interesting than that.

You have to be clear about the question you are asking and I don't think you understand the relative power of the two different types of data used by the different ranking systems.

The Question 

You are comparing two different ranking methodologies and asking which is more accurate at predicting the outcome and why.

One that chelates outcome data that is directly related to the question that is being asked, ie ranking points obtained from playing at the level of competition of the test in the form of ATP points, most importantly points are a surrogate for consecutive wins v a second ranking system that collects data from a variety of levels of competition (some may not be relevant) but more importantly it is only individual wins.

Method A ie. Rank based on ATP points is much more powerful than method B UTR so this an example of method A correctly predicting the outcome and method B failing purely because of the amount of the power of the ranking system.

Method B failed to correctly pick the outcome because it is underpowered in comparison to method A that correctly predicted the outcome of the same event. Method B on this occasion failed to predict the outcome on this occasion because of a type 2 error.

-- Edited by Oakland2002 on Wednesday 5th of October 2016 08:36:06 PM

-- Edited by Oakland2002 on Wednesday 5th of October 2016 08:37:25 PM

tony_orient wrote:

---

Laura doesn't have any events to play next week which might be a good thing. It feels like she hasn't had any sort of break from tournaments since March (apart from a 7 hour holiday), and after 18 months of no tennis and then 9 months of only sporadic play the full time schedule might be catching up with her now.

---

 That is true, and at least this week's and last week's defeats were very tight 3 setters. Plus McKenzie looks like a bright prospect and fresh from qualifying. Will be interesting to see how she fares in the next round - hopefully not too well!!

Possibly getting near time for a UTR thread, because I will respond to points I disagree with and I suspect you will too, Oakland.

And you never know, in a dedicated thread we may get wider input.

Lies, Damned Lies and Statistics