Anyone out there like cricket?

In cricket PriceWaterhouseCooper's Ratings are used
http://www.cricket.pwcglobal.com/cricket/cricket.htm
which give a cricketer a rating between 0 and 1000.

The rating is worked as described here
http://nrich.maths.org/public/viewer.php?obj_id=1385.

I have a feeling that this system is more accurate at determining ability
than the Elo system because it is weighted:
You score more points for scoring runs against stronger bowlers, or for
taking wickets in crucial matches.

So in the Seville match of 1987 Kasparov's Elo rating would have gone down -
because he drew a match with a lesser-rated player.
But his PWC rating - if it existed - would have gone up because he win the
24th game to retain the title.

Alan

Alan O'Brien wrote:
Anyone out there like cricket?

Yup.

In cricket PriceWaterhouseCooper's Ratings are used
http://www.cricket.pwcglobal.com/cricket/cricket.htm
which give a cricketer a rating between 0 and 1000.

The rating is worked as described here
http://nrich.maths.org/public/viewer.php?obj_id=1385.

That doesn't describe how the PWC ratings work in very much detail --
it only goes as far as exponential decay. I can't quote the URL for
PWC's details on it because their site is annoyingly designed to prevent
that. However, if you go to

http://www.pwcratings.com/cricket/cricket.htm

Amd choose `About the ratings' to the left of the photos and then `Ratings
Calculations', you get the details. It's basically an exponentially
decaying average with various fiddle factors to take into account things
like it being less impressive scoring highly against weak opposition or in
a match where everyone else scores highly. There are minor bonuses for
match-winning performances and for not-out batsmen.

Clearly, the case of not-out batsmen has no analogy in chess. Match-
winning performances aren't relevant to chess, either. In an individual
match, the result is the sum of the individual games and it's not clear
that it really makes much difference whether one wins a match +++==-- or
--==+++. In a team competition, there's no scope for a player to make a
match-winning contribution because he can't score more than a single point
per round, just like anyone else on his team. There's no meaningful
comparison between that and the case where a batsman makes 200 runs out of
a team total of 350 or a bowler take seven of the ten wickets in an
innings.


I have a feeling that this system is more accurate at determining
ability than the Elo system because it is weighted:

Not necessarily. Although it uses more of the available data to come up
with the rating, it also introduces more subjectivity.


You score more points for scoring runs against stronger bowlers, or for
taking wickets in crucial matches.

That's not quite true -- you get bonuses for doing well against strong
opposition and for performing well when your team wins (which is, I think,
supposed to model match-winning performances but doesn't).


So in the Seville match of 1987 Kasparov's Elo rating would have gone
down - because he drew a match with a lesser-rated player.
But his PWC rating - if it existed - would have gone up because he win
the 24th game to retain the title.

You're misunderstanding how the PWC ratings work -- there's no bonus for
`performances in crucial games'; it's bonuses for `crucual performances in
games', i.e., doing well when your team does well. As I've explained,
that's not relevant to chess.

I'm not sure that there's anything that should be added into chess ratings
other than the result of the game and the strength of the players. For
example, you could argue that the length of the game should be added
because a quick win as Black is more impressive than a long win as
White. That, however, introduces an entirely subjective value judgement
into the rating scheme by biasing it towards Kasparovs who try to push you
off the board right from the start and against Karpovs who attempt to
accumulate small advantages over a period of time and then win the
endgame. This also means that endgame skill contributes little to the
rating, yet endgame skill is often stated as one of the most significant
differences between a strong player and a weaker player.

You might also want to say that a win precipitated by one's opponent's
blunder is worth less than a win where one's opponent doesn't blunder but
it would be very difficult to define blunder and unclear who is to be the
judge of what is and is not a blunder.


Dave.

--
David Richerby Fluorescent Dangerous Chainsaw (TM):
www.chiark.greenend.org.uk/~davidr/ it's like a lethal weapon but it could
explode at any minute and it'll hurt
your eyes!