| George John 2005-04-07, 6:08 pm |
|
Mike Nolan wrote:
quote:
> "George John" <george@neosoft.com> writes:
>
unrated[vbcol=seagreen]
is[vbcol=seagreen]
>
> Not as rare as you might think. Of 71,024 players with provisional
> ratings earned after 1/1/2004, 5406 or 7.6% have either all wins or
> all losses. (Of those, 5033 have all losses.)
Wow, that's an interesting figure, one which is higher than I would
have thought. If you take into account those who have played in more
than one event, what is the percentage? The calculator only gets into
trouble if there is more than one event. It can handle the first event
correctly (but sometimes an accurate age in necessary).
IIRC, the trick with these players is for those with all losses to
compute a new rating which is either 400 rating points less than their
lowest rated opponent *ever* (or their age-based seed rating if less),
and if all wins 400 points above their highest rated player *ever*, or
their age-based seed rating if more.
[SNIP]
quote:
>
> Using everyone's post-event ratings does result in a much closer
estimate
quote:
> of your own post-event rating, but of course if you know everyone's
> else's post-event rating, you know your post-event rating, too,
making
quote:
> the calculator kind of unnecessary. :-)
The online calculator can be used to calculate post-event ratings for
opponents to get a closer estimate, but it's tedious. Provided
opponents' ratings aren't changing too much, using pre-event ratings
gets one fairly close, and is often good enough.
Best regards,
George
|