A Chess forum. ChessBanter

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » ChessBanter forum » Chess Newsgroups » rec.games.chess.misc (Chess General)
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

Highest ELO performance



 
 
Thread Tools Display Modes
  #21  
Old October 12th 05, 11:28 PM
Taylor Kingston
external usenet poster
 
Posts: n/a
Default


Mike Murray wrote:
On 12 Oct 2005 13:42:52 -0700, "Larry Tapper"
wrote:
You may be thinking of a puzzle book GM Nunn published in 1999. One
chapter features an error-checking study he did of games played at
Karlsbad 1911.


Thanks. That's exactly the source I was trying to remember.

Nunn's logic and conclusions were strongly criticized by IM Richard
Forster in a May 1999 article at ChessCafe.com, "Jewels from Carlsbad
1911." A sample quote:
"Nunn feels confident to state ... that the average strength of the
tournament was a mere 2130 Elo. Now some simple mathematics show that
this is quite ridiculous. If the tournament's strength was 2130 then
[winner Richard] Teichmann's score of 18/25 (+11) corresponds to a
performance of about 2300. You can have many reservations about the Elo
system and the calculation of historical ratings, but something must be
very rotten in the state of Denmark if Teichmann's performance in what
was undoubtedly the best tournament of his life is 270 points below his
peak five-year average." (Which Elo gives as 2570 in "The Rating of
Chessplayers Past and Present.")
Forster goes on to challenge Nunn's contention that of the 325 games
played at Carlsbad 1911, only two were very good. Forster presents
several fine games, and shows how the ideas and techniques demonstrated
in them presaged modern GM games.
The article is, alas, no longer in the ChessCafe archives, but can be
found in print form in the "Heroic Tales" anthology (Russell
Enterprises, 2002).

  #22  
Old October 13th 05, 12:49 AM
Mike Murray
external usenet poster
 
Posts: n/a
Default

On 12 Oct 2005 15:28:32 -0700, "Taylor Kingston"
wrote:

Mike Murray wrote:
On 12 Oct 2005 13:42:52 -0700, "Larry Tapper"
wrote:


You may be thinking of a puzzle book GM Nunn published in 1999. One
chapter features an error-checking study he did of games played at
Karlsbad 1911.


Thanks. That's exactly the source I was trying to remember.


Nunn's logic and conclusions were strongly criticized by IM Richard
Forster in a May 1999 article at ChessCafe.com, "Jewels from Carlsbad
1911." A sample quote:
"Nunn feels confident to state ... that the average strength of the
tournament was a mere 2130 Elo. Now some simple mathematics show that
this is quite ridiculous. If the tournament's strength was 2130 then
[winner Richard] Teichmann's score of 18/25 (+11) corresponds to a
performance of about 2300. You can have many reservations about the Elo
system and the calculation of historical ratings, but something must be
very rotten in the state of Denmark if Teichmann's performance in what
was undoubtedly the best tournament of his life is 270 points below his
peak five-year average." (Which Elo gives as 2570 in "The Rating of
Chessplayers Past and Present.")


It seems to me that Nunn was attempting, by Fritzing a set of games
from representative tournaments of two different eras, to do something
Elo ratings can't: to compare the relative strengths of two disparate
sets of rated players. Having done this using blunder-ratio, he then
expressed, in today's Elo numbers, his estimate for the ratings of
several of these older players.

If my understanding of this is correct, it's a mistake to criticize,
as you report Forster did, Nunn's estimate (given in today's Elo) of
Teichmann's tournament performance rating relative to a lifetime Elo
gleaned from play in his own era.

From one I know of Nunn's article, he didn't dispute that "several
fine games" were played at Carlsbad, 1911. His claimed that the ratio
of blunders over the whole set of games was higher than in a modern
high-category tournament.

For either of us to get in the middle of this by analyzing the games
themselves would be like a couple of cats trying to intervene in a
battle between two elephants.

Forster goes on to challenge Nunn's contention that of the 325 games
played at Carlsbad 1911, only two were very good. Forster presents
several fine games, and shows how the ideas and techniques demonstrated
in them presaged modern GM games.
The article is, alas, no longer in the ChessCafe archives, but can be
found in print form in the "Heroic Tales" anthology (Russell
Enterprises, 2002).

  #23  
Old October 13th 05, 12:57 AM
Ron
external usenet poster
 
Posts: n/a
Default

In article .com,
"Taylor Kingston" wrote:

Forster presents
several fine games, and shows how the ideas and techniques demonstrated
in them presaged modern GM games.


That's sort of a misdirection, though, isn't it?

I mean, nobody doubts that today's players are standing on the shoulders
of giants.

But blundercheck is just that - a blunder check. It's not looking at
complex strategic ideas. It's looking at how often you hang a piece to a
three-mover.

-Ron
  #24  
Old October 13th 05, 01:33 AM
Taylor Kingston
external usenet poster
 
Posts: n/a
Default


Mike Murray wrote:
If my understanding of this is correct, it's a mistake to criticize,
as you report Forster did, Nunn's estimate (given in today's Elo) of
Teichmann's tournament performance rating relative to a lifetime Elo
gleaned from play in his own era.


Mike, I would respectfully submit that your understanding is not
correct. Unless one wants to chuck Elo's historical ratings completely
out the window, Nunn's estimate of the tournament's category makes no
sense.
Keep in mind that Teichmann did not finish atop a bunch of stiffs.
You may recognize some of the other participants, whom I give with
their historical Elos: Rubinstein (2640), Schlechter (2600), Marshall
(2570), Nimzovitch (2615), Vidmar (2600), Alekhine (2690), Duras
(2580), Tartakower (2560), Spielmann (2560), Levenfish (2540), just to
name those most famous now. That's an average of 2595. While not all of
these were at their peaks in 1911, they were all certainly way above
2130, or even 2300.
However, if you accept Nunn's logic, you would have to conclude that
these greats, who include a future world champion (Alekhine), two
actual challengers (Schlechter, Marshall), and three serious title
contenders (Rubinstein, Nimzovitch, Duras), with the rest of them all
regulars in very high-level tournaments, all performed *below*
Teichmann's supposed 2300 TPR. Seems just a tad implausible to me.
And keep in mind that Teichmann never had another result comparable
to Carlsbad 1911, either before it or after. It was the only time he
ever finished so high in so prestigious an event. So if he only scored
a lousy 2300 TPR there, where did his 2570 historical Elo come from?
I recommend that you buy "Heroic Tales" (I get royalties on it) and
read Forster's article. You will enjoy it, and the book as a whole.

  #25  
Old October 13th 05, 02:07 AM
Ron
external usenet poster
 
Posts: n/a
Default

In article .com,
"Taylor Kingston" wrote:

Keep in mind that Teichmann did not finish atop a bunch of stiffs.
You may recognize some of the other participants, whom I give with
their historical Elos: Rubinstein (2640), Schlechter (2600), Marshall
(2570), Nimzovitch (2615), Vidmar (2600), Alekhine (2690), Duras
(2580), Tartakower (2560), Spielmann (2560), Levenfish (2540), just to
name those most famous now. That's an average of 2595. While not all of
these were at their peaks in 1911, they were all certainly way above
2130, or even 2300.
However, if you accept Nunn's logic, you would have to conclude that
these greats, who include a future world champion (Alekhine), two
actual challengers (Schlechter, Marshall), and three serious title
contenders (Rubinstein, Nimzovitch, Duras), with the rest of them all
regulars in very high-level tournaments, all performed *below*
Teichmann's supposed 2300 TPR. Seems just a tad implausible to me.


There are a couple of different problems here, all hinging on the
definition of Elo.

Elo's are nothing more or less than a measure of the performance of a
player compared to his contemporaries. Through some hocus-pocus, we can
guesstimate "historical" Elos for them, but:

Those numbers are fundamentally different than the rating you would
expect those players to achieve if they were playing against modern
players. The two numbers are not comparable.

Nunn's measure was simple: players who blunder at such-and-such a rate
tend to perform as such-and-such a level in today's tournaments. I think
it's a mistake to conflate, too much, his analysis with the entirely
different analysis which produces historical Elos.

-Ron
  #26  
Old October 13th 05, 02:57 AM
Mike Murray
external usenet poster
 
Posts: n/a
Default

On 12 Oct 2005 17:33:48 -0700, "Taylor Kingston"
wrote:

Mike, I would respectfully submit that your understanding is not
correct.
...
keep in mind that Teichmann never had another result comparable
to Carlsbad 1911, either before it or after. It was the only time he
ever finished so high in so prestigious an event. So if he only scored
a lousy 2300 TPR there, where did his 2570 historical Elo come from?


And I equally respectfully suggest it's *you* who misunderstands.

Taylor, the 2300 TPR is based on Nunn's mapping (which largely depends
on Fritz Blundercheck performance) of Teichmann's and the other
competitors' ratings to *contemporary* Elo. It relates in no
significant way to Teichmann's 2570 historical Elo which came from his
performance against peers.

It's akin to a strong player looking through a few of your games
against various opponents with no ratings specified and guesstimating
your rating from the quality of the play. Not precise, but not
nonsense either.

If Nunn went through his analysis and ratings estimates for ALL the
players of Teichmann's vintage, they'd ALL get lower ratings, but,
their predicted performance against each other should stay the same.
That's all that Elo purports to measure. Since they're all long since
dead, of course, this is hard to verify empirically :-(

I recommend that you buy "Heroic Tales" (I get royalties on it) and
read Forster's article. You will enjoy it, and the book as a whole.


Available new for $19.95 (discounted from $24.95) or *damaged* for
$9.95 from USCF/Chess Cafe. Or available brand new for $8.50 from a
couple of affiliated booksellers through Amazon!

And this book is published by Russell Enterprises. Hmmmm. Do you
get your royalties on the sales price or per unit ?
  #27  
Old October 13th 05, 09:37 AM
David Richerby
external usenet poster
 
Posts: n/a
Default

Mike Murray wrote:
David Richerby wrote:
The top ten from today's larger pool of players are statistically
likely to be stronger than the top ten from yesteryear's smaller pool
so they're also harder to beat.


Good point.


Actually, thinking about it, I'm not sure it is a good point. :-) The Elo
system is set up so that it's the rating difference that determines how
likely the win is, so it's supposed to be as hard for a 2700 to win a
tournament where the average rating is 2600 as it is for a 1700 to win
against a field averaging 1600.


Dave.

--
David Richerby Chocolate Lotion (TM): it's like
www.chiark.greenend.org.uk/~davidr/ a soothing hand lotion that's made
of chocolate!
  #28  
Old October 13th 05, 09:53 AM
David Richerby
external usenet poster
 
Posts: n/a
Default

Taylor Kingston wrote:
Mike Murray wrote:
If my understanding of this is correct, it's a mistake to criticize,
as you report Forster did, Nunn's estimate (given in today's Elo) of
Teichmann's tournament performance rating relative to a lifetime Elo
gleaned from play in his own era.


Mike, I would respectfully submit that your understanding is not
correct. Unless one wants to chuck Elo's historical ratings completely
out the window, Nunn's estimate of the tournament's category makes no
sense.

Keep in mind that Teichmann did not finish atop a bunch of stiffs.
You may recognize some of the other participants, whom I give with
their historical Elos: Rubinstein (2640), Schlechter (2600), Marshall
(2570), Nimzovitch (2615), Vidmar (2600), Alekhine (2690), Duras
(2580), Tartakower (2560), Spielmann (2560), Levenfish (2540), just to
name those most famous now.


Mike is, as far as I can see, absolutely correct. You (and, it seems,
Forster) are forgetting that Elo explicitly states that you can't use
Rubinstein's rating of 2640 in 1911 (or whenever that rating is from) to
say that he is as good as Alexander Onischuk and Zoltan Almasi, who have
that rating today. As you say, it's also wrong to use the players' peak
ratings rather than their ratings in 1911 -- that is the case for two
readons: firstly because, as you say, they might not have been at their
peak in 1911 and also, to a lesser degree, because Rubinstein's peak
rating is compared to the strength of other players at the time Rubinstein
peaked, not compared to the strength of other players in 1911.

Rubinstein's rating is relative to his contemporaries, as are Onischuk and
Almasi's ratings. What Nunn is trying to do is estimate how ratings have
changed since 1911. Now, when you say that Nunn's estimate of the
tournament's category makes no sense, you do have a point. If ratings and
categories were in use at that point, the tournament would have a high
category because it featured most of the best players in the world. What
Nunn's analysis shows, though, is that a Category N tournament in 1911 had
a much lower standard of play than a Category N tournament does now.


Dave.

--
David Richerby Surprise Mouldy Smokes (TM): it's
www.chiark.greenend.org.uk/~davidr/ like a pack of cigarettes but it's
starting to grow mushrooms and not
like you'd expect!
  #29  
Old October 13th 05, 01:37 PM
Taylor Kingston
external usenet poster
 
Posts: n/a
Default


Mike Murray wrote:
Taylor, the 2300 TPR is based on Nunn's mapping (which largely depends
on Fritz Blundercheck performance) of Teichmann's and the other
competitors' ratings to *contemporary* Elo. It relates in no
significant way to Teichmann's 2570 historical Elo which came from his
performance against peers.


Well, I can't really discuss this in any informed way without Nunn's
book, which I do not have now. When I saw it years ago, I did not go
through the section on Carlsbad 1911 in any depth. Perhaps if I get the
chance to review it fully, we could take up the discussion again.

Available new for $19.95 (discounted from $24.95) or *damaged* for
$9.95 from USCF/Chess Cafe. Or available brand new for $8.50 from a
couple of affiliated booksellers through Amazon!

And this book is published by Russell Enterprises. Hmmmm. Do you
get your royalties on the sales price or per unit ?


I am paid per unit on Heroic Tales. It has helped me become a
thousandaire.

  #30  
Old October 13th 05, 05:32 PM
Larry Tapper
external usenet poster
 
Posts: n/a
Default


Mike Murray wrote:

MM ...And I equally respectfully suggest it's *you* who
misunderstands.

MM Taylor, the 2300 TPR is based on Nunn's mapping (which largely
depends on Fritz Blundercheck performance) of Teichmann's and the other

competitors' ratings to *contemporary* Elo. It relates in no
significant way to Teichmann's 2570 historical Elo which came from his
performance against peers.

MM It's akin to a strong player looking through a few of your games
against various opponents with no ratings specified and guesstimating
your rating from the quality of the play. Not precise, but not
nonsense either.

MM If Nunn went through his analysis and ratings estimates for ALL the

players of Teichmann's vintage, they'd ALL get lower ratings, but,
their predicted performance against each other should stay the same.
That's all that Elo purports to measure. Since they're all long since
dead, of course, this is hard to verify empirically :-(

Mike,

I mostly agree with your take on all this.

Another way to look at Nunn's study is that it measures playing
strength along one dimension only, which we might call "error
avoidance". This has the conspicuous merit of allowing us to compare
across large time gaps. But of course error avoidance is not all there
is to success at chess; so the blundercheck approach may be
questionable as an indicator of anyone's _overall_ strength. It ignores
important qualities such as tenacity, imagination, strategic insight,
etc.

Even so, I think Nunn came up with a real finding there, and an
interesting one too. His results suggest that surprising as it may
seem, the average 2400 player of today might well be able to hold his
own in an error-avoidance contest against Teichmann at his peak
strength. And this does not seem so outlandish when I pause to think
about it --- it would be a sign that professional standards have
tightened due to competitive pressure and so on.

Larry T.

 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Forum Jump


All times are GMT +1. The time now is 07:47 PM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.Content Relevant URLs by vBSEO 2.4.0
Copyright 2004-2017 ChessBanter.
The comments are property of their posters.