Draught Engines to comment WC draughts in June

Discussion about development of draughts in the time of computer and Internet.
User avatar
FeikeBoomstra
Posts: 306
Joined: Mon Dec 19, 2005 16:48
Location: Emmen

Post by FeikeBoomstra »

There has to be at least one draw, otherwise we can't compute the rating difference.

But anyway, a very good result Jelle.
BertTuyt
Posts: 1592
Joined: Wed Sep 01, 2004 19:42

Post by BertTuyt »

Jelle, congratulations with this result.

So far I have seen that you played 7 games and 7 wins.
Were there more games played?

Can you say something about the program/hardware you used, where there changes compared with the "France" Version?

Was your program anywhere in "problems" according to the eval?
And what was the main reason, for the wins (specific human errors?, specific positions?, specific phase in the game?)

Bert
Piet Bouma
Posts: 3574
Joined: Sun Nov 02, 2003 13:05
Location: Harlingen

Post by Piet Bouma »

BertTuyt wrote:Jelle, congratulations with this result.

So far I have seen that you played 7 games and 7 wins.
Were there more games played?

Can you say something about the program/hardware you used, where there changes compared with the "France" Version?

Was your program anywhere in "problems" according to the eval?
And what was the main reason, for the wins (specific human errors?, specific positions?, specific phase in the game?)

Bert
"Gorredijk Open" is a 'moyenne-system' tournament, with 15 rounds in the summer (every tuesday). The players have to play 10 rounds to be classified.
Sjende Blyn has to play another three rounds to win the tournament.
Let's hope Mr. Goedemoed can play a draw (I think he wants more), so we can measure the rating of Sjende Blyn!
wellnesswrotter
Posts: 323
Joined: Mon May 21, 2007 15:10
Location: www.snukenkuzco.nl
Contact:

Post by wellnesswrotter »

Jelle still uses the same computer for this tournament as he always does. His laptop has a issue with the fan, so he always has to run two times to setup his computer. But he also tweaks his computer during the week. For example: he now uses a openingsbook.

Is it nescisary to have at least one draw??
User avatar
FeikeBoomstra
Posts: 306
Joined: Mon Dec 19, 2005 16:48
Location: Emmen

Post by FeikeBoomstra »

A 100% score is an anomaly in the rating system (if you look at the tournament rating).

In the Dutch system, the tournament rating is only used to give an initial rating to players, based on their first 25 games against rated players. After that an increment method is used, and then there is no problem anymore with the 100% scores.

For the tournament rating:
MyRating = AverageOpponentRating + delta_rating(score).

See the article (in Dutch):
http://www.kndb.nl/cms/index.php?option ... Itemid=126 and
http://www.kndb.nl/cms/index.php?option ... Itemid=126

For the tournament rating, in practice, the problem is solved by saying a 100% score gives a rating difference of 965 i.s.o infinity. Another way of solving it, is to include a "phantom match", playing a draw against yourself.
Piet Bouma
Posts: 3574
Joined: Sun Nov 02, 2003 13:05
Location: Harlingen

Post by Piet Bouma »

FeikeBoomstra wrote:Another way of solving it, is to include a "phantom match", playing a draw against yourself.
Because Sjende Blyn has no rating, a "phantom match" is not possible.
wellnesswrotter
Posts: 323
Joined: Mon May 21, 2007 15:10
Location: www.snukenkuzco.nl
Contact:

Post by wellnesswrotter »

and if it participated in a tournament later on?
User avatar
FeikeBoomstra
Posts: 306
Joined: Mon Dec 19, 2005 16:48
Location: Emmen

Post by FeikeBoomstra »

Piet Bouma wrote:
FeikeBoomstra wrote:Another way of solving it, is to include a "phantom match", playing a draw against yourself.
Because Sjende Blyn has no rating, a "phantom match" is not possible.
Sorry, not true.

Say, he played nine matches against average 1100 players
including the phantom match his score is 95%.
The related rating difference is 470, so:

Code: Select all

0.1 x SB + 0.9 x 1100 = SB - 470  - ->
0.9 x SB = 0.9 x 1100 + 470 -->
SB = 1622

Piet Bouma
Posts: 3574
Joined: Sun Nov 02, 2003 13:05
Location: Harlingen

Post by Piet Bouma »

FeikeBoomstra wrote:
Piet Bouma wrote:
FeikeBoomstra wrote:Another way of solving it, is to include a "phantom match", playing a draw against yourself.
Because Sjende Blyn has no rating, a "phantom match" is not possible.
Sorry, not true.

Say, he played nine matches against average 1100 players
including the phantom match his score is 95%.
The related rating difference is 470, so:

Code: Select all

0.1 x SB + 0.9 x 1100 = SB - 470  - ->
0.9 x SB = 0.9 x 1100 + 470 -->
SB = 1622

I don't know if that was the purpose of the "phantom match". I believe it was more to calculate a TPR in a tournament of player X (with rating) who gets the 100% score.
Then the formula is: 0.1 x ratingX + 0.9 x 1100 = SB - 470.

But Sjende Blyn has also played a simultane during the last day of the Frisian Youth Championship. If we count this 4 games (1h30min timecontrol) with the games of Gorredijk, we have a rating:
11 games, 19 points 86,36%, average rating 1193 (4 games KNDB-rating 1-7-2008, 7 games provisonal rating KNDB 1-7-2009).
Rating (on KNDB-rules) Sjende Blyn: 1506. FMJD-equivalent: (1506 + 850) = 2356.
wellnesswrotter
Posts: 323
Joined: Mon May 21, 2007 15:10
Location: www.snukenkuzco.nl
Contact:

Post by wellnesswrotter »

De Gorredijk Open Zomercompetitie 2009 is afgelopen.
Sjende Blyn is eerste geworden met 21 uit 12.
Weer even tijd voor een ratingberekening.
b.Als wij wel de rating van Wiersma kennen en de einduitslag van de match, maar niet de rating van Nikhila, kunnen we deze laatste rating op basis van het resultaat in de match berekenen. De behaalde score van Nikhila is 14/40 ofwel 30%. Uit de tabel kunnen wij bij 30% (kolom D) aflezen dat hierbij een ratingverschil hoort van 149 (kolom A). Uitgaande van de bekende rating van Wiersma, namelijk 1606, kunnen we de rating van Nikhila berekenen op 1606 - 149 = 1457. Dit is dus de rating die behoort bij het resultaat in
deze match.

Voor spelers die nog geen rating hebben of aan het begin van het seizoen minder dan 25 wedstrijden hebben gespeeld wordt de rating anders bepaald. Alle afzonderlijke uitslagen worden verzameld evenals de rating van de tegenstanders. Op basis van methode b. wordt uit de gemiddelde tegenstanderrating en het behaalde resultaat de eigen rating berekend. Dit
kunnen resultaten zijn uit het lopende seizoen maar ook uit vorige seizoenen. De correctiefactor bedraagt 5 voor spelers die meer dan 125 wedstrijden in totaal hebben gespeeld. Voor spelers die minder dan 125 wedstrijden hebben gespeeld geldt een correctiefactor van 7,5.

Code: Select all

Pieter Dijkstra          ( 982)  2 
Reinout Sloot            (1159)  2 
Sietse Nagel             (1196)  2 
Willem van der Velde     ( 885)  2 
Jouke de Haan            (1090)  2 
Johan Teake Dekker       (1357)  2 
Anton Schotanus          (1321)  2 
Erwin Heslinga           (1269)  2 
Hans Zondervan           (1254)  1 
Tjalling Goedemoed       (1395)  1 
Jan Marten van der Reest (1248)  2 
Oege Dijkstra            (1318)  1 
================================== 
Total                    14474  21 
Average: 14474/12=1206
De behaalde score van Sjende Blyn is 21/24 ofwel 87.5%
Ratingpuntenverschil = +322
Toernooiprestatierating: 1528

Dit zou theoretisch een plek betekenen tussen Virny (1532) en Thijssen (1520). De nummer 9 en 10 van de Nederlandse ratinglijst.
Post Reply