While it has huge benefits (not included here), it is subject to abuse, for example:
The top rated blitz player has achieved his mega rating having only played SEVEN games.
As you scroll through the list, you will find many players on the front page have only played a handful of games, and even more have played less than 100 or so.
It seems necessary, then, to introduce some measures against this, such as e.g. a provisional rating which becmes real after so many games, etc.
Mehh, you could get the highest rating in the world by playing 1000 rated players, what good does it do you?
In addition to strategies like reporting ranked lists by rating only among players who have some specified maximum RD, another approach I've advocated is to rank players not by rating, but by a quantity like rating - RD or even rating - 2*RD.
Even another idea is to rank players according to a quantity like
which turns out to be monotonically related to the probability a player's true strength is greater than 2200.
- Mark (a.k.a. glicko)
Looks like a t-value, of sorts, that (rating - 2200)/RD.
Yes, in fact it's a z-value (from a normal distribution), since a player's rating under the Glicko system is assumed to be approximately normally distributed.
I guess the staff have no issue with this :S
My Question is ...
If Im rated 1200 with this glicko,
What would my rating be in the elo system that USCF uses????