[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Suggested Idea for Performance Grading

Posted By: Rick Janowski
Date: Friday, 3 October 2014, at 6:43 p.m.

In Response To: Suggested Idea for Performance Grading (Keene)

I agree that ideally results (or other win-rate related criteria like Elo rating) are the best way of measuring success. Results are the output in the process whereas error rate (or pr) is only an input (though I think obviously the most important one). The difficulty with using win-rate as a measure in backgammon is that it needs a very large sample size to be reliable. Even when this large sample size is available, ongoing improvements in a player’s level of play will usually be shrouded. As a measuring device it’s a bit too crude to pick up say an elo rating difference of say 50 (equivalent to pr difference of 1.5). Error rate, in comparison has much smaller variance making it a much more useful measuring tool for small differences.

Error Rates used by Snowie or Gnu can be approximated to equivalent prs and vice versa, and similarly whatever measurement adopted by a new bot can be dealt with by recalibrating the norm levels. We are not stuck with any bot really, but merely using the best tools currently available.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.