| Angelo De Pa1ma 2005-04-07, 6:08 pm |
|
OK, so it appears that new ratings are calculated in increments, one
opponent at a time, as opposed to the old system that used average opponent
ratings or PR.
All well and good. Do you think it's fair? I seem to have stumbled on an
example that shows it may not be. Or at least that it's a disincentive to
playing everybody you're paired with.
Intuitively -- and I've taken quite a bit of math, including statistics --
it would seem that a better rating-PR difference over more games should be
worth more points, period. Play better, longer, and your rating goes higher.
What's happening here is players are penalized for being paired down. In
other words it's "better" for me to play 4 1950s than two 2350s and two
1550s.
This on top of the fact that of of the low-rated players I faced at the
USATE only the highest-rated, a 1700-ish, was in my oppinion accurately
rated. Almost every player in the 1100-1700 range under the age of 25 (even
more true for teens) that I have faced in the last 6 years has been
under-rated. Playing these guys is a heads-I-lose, tails-he-wins
proposition.
Angelo
"Mike Nolan" <nolan@gw.tssi.com> wrote in message
news:d2qh2d$ll7$1@gw.tssi.com...
quote:
> "Angelo De Pa1ma" <adpspammersgotohell@tellurian.net> writes:
>
>
> If you've read the paper on the website describing the ratings formula,
> you should have noticed that performance rating isn't really a factor
> anymore, it isn't even computed except for provisionally rated players.
>
> The pre-event rating is a major component in computing the K factor,
> and the effective number of games, and the difference in rating between
> the player and each opponent determines the expected probability of
> each game.
>
> The bonus is essentially the same formula, but the maximum bonus available
> drops as the number of games increases.
> --
> Mike Nolan
|