PDA

View Full Version : putting ELO ratings to the test


BeHappy
03-25-2008, 01:18 PM
What if, (wuornos) before every match for one single competition, you were to apply your ELO ratings to predict the winner, and we can compare the accuracy of the results to the accuracy of predicting the results with conventional rankings, ie, predict the higher ranking player wins?

BeHappy
03-26-2008, 01:29 PM
wuornos ?

Wuornos
03-26-2008, 01:38 PM
What if, (wuornos) before every match for one single competition, you were to apply your ELO ratings to predict the winner, and we can compare the accuracy of the results to the accuracy of predicting the results with conventional rankings, ie, predict the higher ranking player wins?

Hmmm. Interesting, although ELO would produce a % chance to win. It might be easier to pick out the matches where the ELO rank differs from the official rank and see which is correct. E.g. Andy Roddick is higher in the ELO rankings.

I'm quite busy at the moment with the new Hybrid system but perhaps we could pit the hybrid against ELO against the official rankings when I have finished this. It should be fun.

Take care

Tim

BeHappy
03-26-2008, 01:41 PM
No hurry, but if this worked you could email the results to the atp.It would also make a good case study if you ever wanted to write an artivle/book about it.

Wuornos
03-26-2008, 01:42 PM
No hurry, but if this worked you could email the results to the atp.It would also make a good case study if you ever wanted to write an artivle/book about it.

I agree. Thanks for that.

Tim