How much is a knight worth?

Date: 13 Oct 2007 17:25:17
From: Petrovitch
Subject: How much is a knight worth?

Knights have been evaulated as 3 points. A rating difference of 100
points means a player should win 5:8 games. In relative value, how
much is a knight worth in rating points? And the other pieces?

If a player is 100 points higher should he be able to win with a
(pawn, 2 pawns, knight, etc.) advantage?

Date: 15 Oct 2007 13:23:28
From: help bot
Subject: Re: How much is a knight worth?

On Oct 14, 1:09 pm, johnnyt <[email protected] > wrote:
> Petrovitch wrote:
> > Knights have been evaulated as 3 points. A rating difference of 100
> > points means a player should win 5:8 games. In relative value, how
> > much is a knight worth in rating points? And the other pieces?
>
> > If a player is 100 points higher should he be able to win with a
> > (pawn, 2 pawns, knight, etc.) advantage?
>
> As I understand, the value is different based on the understanding of
> the weaker player. Rather than just raw difference.
>
> However, there has been vsrioud odds matched between strong GM's and
> Rybka in the past year, and there has been some discussion about them on
> Rybka's (www.rybkachess.com) website.

One thing I noticed in replaying the games between
Rybka and GM Benjamin is that the program was not
designed or configured properly to make best use of
such odds. In most cases, the human was able to
successfully steer into "book opening" positions where
the missing pawn was a major handicap, yet Rybka
won the match in spite of this fact.

At lower levels of play, there are so many errors
that a handicap of QN may well be overcome by a
very strong player over the course of many moves,
but the higher you go, the more important it is to not
be in "a lost endgame" at the end of every exchange
of pieces.

For example, I would feel reasonably comfortable
playing Rybka at Queen odds, because I know that
the program has not been properly programmed to
compensate for this titanic handicap, so it is not
the same dangerous opponent it would be in more
normal types of positions.

Those players who advocate time-odds want to
keep the character of the game the same, and this
is the simplest way to achieve that end. However,
in many cases the much weaker player will not
make good use of his extra time, and so the
stronger player is giving a sort of "fake" odds, in
one sense; I have seen massive time odds where
in the end, the weaker player is the one in time
pressure (these are the type of players who attract
offers of massive time odds, since it doesn't really
help them very much).

Another issue is an unwillingness to surrender
the detailed knowledge of the opening, the heavy
study of chess openings theory, and again I smell
a rat; top players spend an inordinate amount of
time on the study of openings theory, and in some
lines no amount of additional time OTB is going to
enough to unravel the intricacies, so the receiver
of a time handicap will squander his edge in a vain
attempt to jump to the moon, wearing just his
sneakers.

I want to see Rybka playing IMs at Knight odds,
followed by weak GMs and then, ultimately,
stronger GMs. But first they need to stop mucking
it up the way they did with GM Benjamin; figure out
what contempt factor and other settings will improve,
rather than handicap the program's play.

-- help bot

Date: 14 Oct 2007 11:09:16
From: johnnyt
Subject: Re: How much is a knight worth?

Petrovitch wrote:
> Knights have been evaulated as 3 points. A rating difference of 100
> points means a player should win 5:8 games. In relative value, how
> much is a knight worth in rating points? And the other pieces?
>
> If a player is 100 points higher should he be able to win with a
> (pawn, 2 pawns, knight, etc.) advantage?
>

As I understand, the value is different based on the understanding of
the weaker player. Rather than just raw difference.

However, there has been vsrioud odds matched between strong GM's and
Rybka in the past year, and there has been some discussion about them on
Rybka's (www.rybkachess.com) website.

Date: 13 Oct 2007 22:07:02
From: Petrovitch
Subject: Re: How much is a knight worth?

Let's take the opening first. How could a player determine which
piece to give an opponent to equalize the game given the difference in
ratings. Obviously, a queen's rook would be worth less than a king's
rook. But how can we put this into quantative values?

Date: 14 Oct 2007 08:11:53
From: Ian Burton
Subject: Re: How much is a knight worth?

"Petrovitch" <[email protected] > wrote in message
news:[email protected]...
> Let's take the opening first. How could a player determine which
> piece to give an opponent to equalize the game given the difference in
> ratings. Obviously, a queen's rook would be worth less than a king's
> rook. But how can we put this into quantative values?

It cannot sensibly be done. Forget the idea and give time odds.

--
Ian Burton
(Please reply to the Newsgroup)

Date: 13 Oct 2007 13:53:28
From: Ken Blake
Subject: Re: How much is a knight worth?

On Sat, 13 Oct 2007 17:25:17 -0000, Petrovitch
<[email protected] > wrote:

> Knights have been evaulated as 3 points. A rating difference of 100
> points means a player should win 5:8 games. In relative value, how
> much is a knight worth in rating points? And the other pieces?
>
> If a player is 100 points higher should he be able to win with a
> (pawn, 2 pawns, knight, etc.) advantage?

The answers to all question like this is "it depends on the position."

To take the simplest of examples, if am up a single pawn, in a K+P vs.
K position, I can win against *anybody* in the world (regardless of
who is rated higher or by how much) in some positions, and be unable
to win against almost anybody in other positions.

--
Ken Blake
Please Reply to the Newsgroup