datamining and chess

Date: 17 Oct 2005 15:14:41
From: Folkert van Heusden
Subject: datamining and chess

A few weeks ago I did a little data-mining on the game of chess.
I analyzed (using some scripting, of course) approx 350000 games by
grandmasters and rookies.
The results can be found here:
http://www.vanheusden.com/misc/datamining_on_chess.html

Date: 25 Oct 2005 02:37:57
From: Predictor
Subject: Re: datamining and chess

Folkert van Heusden wrote:
> A few weeks ago I did a little data-mining on the game of chess.
> I analyzed (using some scripting, of course) approx 350000 games by
> grandmasters and rookies.
> The results can be found here:
> http://www.vanheusden.com/misc/datamining_on_chess.html

It seems you've collected a number of sumy statistics on a large
database of games. I wouldn't call this "data mining". My one
suggestion would be to provide proportions as well as raw counts. For
instance, the number of times a pawn was moved by white is 2618154,
which I can see is greater than for any other piece for white, but what
percentage does that represent?

As an occasional chess player, I find it vaguely interesting, but I
doubt much of this would be useful in indicating what to do in any
particular game or situation.

-Will Dwinnell
http://will.dwinnell.com

Date: 17 Oct 2005 09:44:26
From: Predictor
Subject: Re: datamining and chess

Folkert van Heusden wrote:
> >> A few weeks ago I did a little data-mining on the game of chess.
> >> I analyzed (using some scripting, of course) approx 350000 games by
> >> grandmasters and rookies.
> >> The results can be found here:
> >> http://www.vanheusden.com/misc/datamining_on_chess.html
> > That document includes the following statement:
> > "White wins 1.30 more often then black."
> > Do you mean that white wins 1.3 times for every 1 time won by black?
>
> Ehr yes: 1.3 times more often. If black wins 10 times, white wins 13 times.

That's why I wrote: this would be 1.3 times as often.

-Will Dwinnell
http://will.dwinnell.com

Date: 17 Oct 2005 08:17:22
From: Predictor
Subject: Re: datamining and chess

Date: 17 Oct 2005 18:13:57
From: Folkert van Heusden
Subject: Re: datamining and chess

>> A few weeks ago I did a little data-mining on the game of chess.
>> I analyzed (using some scripting, of course) approx 350000 games by
>> grandmasters and rookies.
>> The results can be found here:
>> http://www.vanheusden.com/misc/datamining_on_chess.html
> That document includes the following statement:
> "White wins 1.30 more often then black."
> Do you mean that white wins 1.3 times for every 1 time won by black?

Ehr yes: 1.3 times more often. If black wins 10 times, white wins 13 times.

I was very surprised as everybody always told me there's no difference.

Date: 18 Oct 2005 00:25:31
From: CeeBee
Subject: Re: datamining and chess

Folkert van Heusden <[email protected] > wrote in
rec.games.chess.analysis:

> I was very surprised as everybody always told me there's no difference.

I wonder who that "everybody" is. Every random large data collection will
give you an approximate white win number of around 37-40%, a black win
around 29-31% and the rest draw; no matter what random selection you make.

--
CeeBee

***Ancient Wisdom in a Crunchy Treat***

Date: 19 Oct 2005 14:29:22
From: David Richerby
Subject: Re: datamining and chess

CeeBee <[email protected] > wrote:
> Folkert van Heusden <[email protected]> wrote:
>> I was very surprised as everybody always told me there's no difference.
>
> I wonder who that "everybody" is. Every random large data collection
> will give you an approximate white win number of around 37-40%, a black
> win around 29-31% and the rest draw; no matter what random selection you
> make.

Seventy-odd percent of games being decisive seems too high for grand-
master games and too low for patzer games. Between complete beginners,
each colour wins about half the games, with very few draws. Between
grandmasters, White does better, with lots of draws.

Dave.

--
David Richerby Hungry Peanut (TM): it's like a
www.chiark.greenend.org.uk/~davidr/ roasted nut but it'll eat you!

Date: 18 Oct 2005 08:00:36
From: Few Good Chessmen
Subject: Re: datamining and chess

"Folkert van Heusden" <[email protected] > wrote in message
news:[email protected]...
> >> A few weeks ago I did a little data-mining on the game of chess.
> >> I analyzed (using some scripting, of course) approx 350000 games by
> >> grandmasters and rookies.
> >> The results can be found here:
> >> http://www.vanheusden.com/misc/datamining_on_chess.html
> > That document includes the following statement:
> > "White wins 1.30 more often then black."
> > Do you mean that white wins 1.3 times for every 1 time won by black?
>
> Ehr yes: 1.3 times more often. If black wins 10 times, white wins 13
times.
>
> I was very surprised as everybody always told me there's no difference.

Other than this I found rest of the tabled figures useless even you have the
sample of million games. And if you added another 10 folds after that still
the table is worthless, etc. A well earned Chess Knowledge could easily
predicts similiar result relatively. As for winning games - the figures show
the only games won at conclusion not from the turning point when they are
decisive to the end.

Date: 18 Oct 2005 01:04:55
From: Vasileios Zografos
Subject: Re: datamining and chess

"Few Good Chessmen" <[email protected] > wrote in message
news:[email protected]...
> "Folkert van Heusden" <[email protected]> wrote in message
> news:[email protected]...
>> >> A few weeks ago I did a little data-mining on the game of chess.
>> >> I analyzed (using some scripting, of course) approx 350000 games by
>> >> grandmasters and rookies.
>> >> The results can be found here:
>> >> http://www.vanheusden.com/misc/datamining_on_chess.html
>> > That document includes the following statement:
>> > "White wins 1.30 more often then black."
>> > Do you mean that white wins 1.3 times for every 1 time won by black?
>>
>> Ehr yes: 1.3 times more often. If black wins 10 times, white wins 13
> times.
>>
>> I was very surprised as everybody always told me there's no difference.
>
> Other than this I found rest of the tabled figures useless even you have
> the
> sample of million games. And if you added another 10 folds after that
> still
> the table is worthless, etc. A well earned Chess Knowledge could easily
> predicts similiar result relatively. As for winning games - the figures
> show
> the only games won at conclusion not from the turning point when they are
> decisive to the end.
>
>
>

useless?
worthless?
wow...you are really a very nice and polite person aren't you?
Is that what you do in here? Look around in the postings trying to offend
people?
This guy was good enough to do so much hard work and generate all this
statistical data and kind enough to publish it for us and you call it
useless and worthless?
Who do you think you are anyway?

Date: 18 Oct 2005 05:10:42
From: Michael Vondung
Subject: Re: datamining and chess

On Tue, 18 Oct 2005 01:04:55 +0100, Vasileios Zografos wrote:

> wow...you are really a very nice and polite person aren't you?

Not wanting to get involved in this, but it's a common observation that
good chess players tend to lack in "social skills". I don't know if that
makes them "impolite" or "not nice", though, perhaps it's just "no fluff"
bluntness and efficiency. I don't necessarily agree that it is efficient,
but I don't think the goal is to offend. It's more of a "no bullshit"
approach.

I'm a sucky chess player, so I'm much more social. ;)

M.

Date: 18 Oct 2005 12:22:31
From: Few Good Chessmen
Subject: Re: datamining and chess

"Michael Vondung" <[email protected] > wrote in message
news:[email protected]...
> On Tue, 18 Oct 2005 01:04:55 +0100, Vasileios Zografos wrote:
>
> > wow...you are really a very nice and polite person aren't you?
>
> Not wanting to get involved in this, but it's a common observation that
> good chess players tend to lack in "social skills". I don't know if that
> makes them "impolite" or "not nice", though, perhaps it's just "no fluff"
> bluntness and efficiency. I don't necessarily agree that it is efficient,
> but I don't think the goal is to offend. It's more of a "no bullshit"
> approach.
>
> I'm a sucky chess player, so I'm much more social. ;)

Everyone know for the fact that Capablanca, Jose Raul is Great Chess Player
and he also a womanizer...Chess Players might lack the "social skills" among
peers (understandable for tournaments player) but nothing was known within
close circle of friends and private lifes.

Social skills on the Internet is a very touchy subject (we don't know the
true gender, intentions, etc behind the scene). I'd be careful if caught too
attached, ganging up, etc...

Date: 18 Oct 2005 04:17:47
From: Vasileios Zografos
Subject: Re: datamining and chess

Hi Michael,

you know what, I would normaly agree with you. And that is the case for more
"geniouses", they do lack certain social skills. In the case of this guy
however, I had hard time believing he is a good chess player, because he
does not understand basic things about chess (see his other postings), and
not only that but he insists he is right even after many people try to
convience him otherwise. There is nothing with being wrong, many people are.
But what really is amazing with him is that he is so stubborn and aroggant
to understand when he is wrong.

Anyways, I promised myself I will not give him the pleasure of starting
another flame war.

"Michael Vondung" <[email protected] > wrote in message
news:[email protected]...
> On Tue, 18 Oct 2005 01:04:55 +0100, Vasileios Zografos wrote:
>
>> wow...you are really a very nice and polite person aren't you?
>
> Not wanting to get involved in this, but it's a common observation that
> good chess players tend to lack in "social skills". I don't know if that
> makes them "impolite" or "not nice", though, perhaps it's just "no fluff"
> bluntness and efficiency. I don't necessarily agree that it is efficient,
> but I don't think the goal is to offend. It's more of a "no bullshit"
> approach.
>
> I'm a sucky chess player, so I'm much more social. ;)

:) what can I say? Both my chess and social skills are virually
non-existent.

>
> M.

Date: 18 Oct 2005 08:10:24
From: Few Good Chessmen
Subject: Re: datamining and chess

"Vasileios Zografos" <[email protected] > wrote in message
news:1129593899.b9109cd77c8f9fb086024047b8a6a7ee@teranews...
>
> "Few Good Chessmen" <[email protected]> wrote in message
> news:[email protected]...
> > "Folkert van Heusden" <[email protected]> wrote in message
> > news:[email protected]...
> >> >> A few weeks ago I did a little data-mining on the game of chess.
> >> >> I analyzed (using some scripting, of course) approx 350000 games by
> >> >> grandmasters and rookies.
> >> >> The results can be found here:
> >> >> http://www.vanheusden.com/misc/datamining_on_chess.html
> >> > That document includes the following statement:
> >> > "White wins 1.30 more often then black."
> >> > Do you mean that white wins 1.3 times for every 1 time won by black?
> >>
> >> Ehr yes: 1.3 times more often. If black wins 10 times, white wins 13
> > times.
> >>
> >> I was very surprised as everybody always told me there's no difference.
> >
> > Other than this I found rest of the tabled figures useless even you have
> > the
> > sample of million games. And if you added another 10 folds after that
> > still
> > the table is worthless, etc. A well earned Chess Knowledge could easily
> > predicts similiar result relatively. As for winning games - the figures
> > show
> > the only games won at conclusion not from the turning point when they
are
> > decisive to the end.
> >
> >
> >
>
> useless?
> worthless?
> wow...you are really a very nice and polite person aren't you?
> Is that what you do in here? Look around in the postings trying to offend
> people?
> This guy was good enough to do so much hard work and generate all this
> statistical data and kind enough to publish it for us and you call it
> useless and worthless?
> Who do you think you are anyway?

Tell me then numbskull, how do you find it useful? I stressed my points of
the figures worthlessness could you even suggest about yours?

Date: 17 Oct 2005 19:22:06
From: Ron
Subject: Re: datamining and chess

In article <[email protected] >,
Folkert van Heusden <[email protected] > wrote:

> I was very surprised as everybody always told me there's no difference.

The only "everybody" who would have told you there was no difference
would have been people who don't know chess at all.

Every serious player knows white has an advantage.

Date: 17 Oct 2005 10:38:54
From: Angelo DePalma
Subject: Re: datamining and chess

I don't get it.

"Folkert van Heusden" <[email protected] > wrote in message
news:[email protected]...
>A few weeks ago I did a little data-mining on the game of chess.
> I analyzed (using some scripting, of course) approx 350000 games by
> grandmasters and rookies.
> The results can be found here:
> http://www.vanheusden.com/misc/datamining_on_chess.html
>

Date: 17 Oct 2005 18:14:18
From: Folkert van Heusden
Subject: Re: datamining and chess

What exactly?

> I don't get it.
>
>
> "Folkert van Heusden" <[email protected]> wrote in message
> news:[email protected]...
>>A few weeks ago I did a little data-mining on the game of chess.
>> I analyzed (using some scripting, of course) approx 350000 games by
>> grandmasters and rookies.
>> The results can be found here:
>> http://www.vanheusden.com/misc/datamining_on_chess.html
>>