Thursday 15 November 2007

Ratings Arguments - Chapter 0

Occasionally when I get into a discussion with someone about a contentious topic, I realise that the position they have taken isn't supported by what I would regard as "facts". When I point this out they often say "Well, what are the facts?". To stop my head from exploding I usually cut the argument off stating "We can only continue this discussion when you know what you are talking about" (or more bluntly "Do you own homework").
I'm currently seeing this in online discussions about Elo v Glicko as a preferable rating system. This is a particularly heated topic in Australia, although the heat mainly comes from those who would use it as a stick to beat the current ACF Ratings Officer Bill Gletsos over non-related issues.
However for those that don't fall into that category here are a couple of links of interest. Parameter estimation in large dynamic paired comparison experiments is Mark Glickman's original paper on the "Glicko" Rating System. (Note You will need a Postscript reader to look at this file). A summary of the Glicko system can be read here. (This is PDF format).
At some point down the track there will be a "Chapter 1", "Chapter 2" and even a "Chapter 978" of this post, but before I get there I'd like to be sure that people who throw around terms like "inferior mathematically" know what they are talking about.

No comments: