Cheating suspicion at the Zadar Open in Croatia

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
aturri
Posts: 84
Joined: Wed Dec 30, 2009 10:35 pm

Re: Cheating suspicion at the Zadar Open in Croatia.

Post by aturri » Thu Jan 10, 2013 10:12 pm

Hood wrote:Do you know the story about Skynet :-) ?
I am shocked how people dare to made allegations with no proofs.
It does not give them good opinion.
Hood wrote:Polish National tragedy in Smoleńsk. President and all delegation murdered or killed. Cui bono?
Does these two sentences have any relation?

Hood
Posts: 652
Joined: Mon Feb 08, 2010 11:52 am
Location: Polska, Warszawa

Re: Cheating suspicion at the Zadar Open in Croatia.

Post by Hood » Fri Jan 11, 2013 7:04 pm

The traces of TNT were found on the parts of crashed aeroplane.
Polish National tragedy in Smoleńsk. President and all delegation murdered or killed.
Cui bono ?

There are not bugs free programs.
There are programs with undiscovered bugs.




Ashes to ashes dust to dust. Alleluia.


User avatar
simonhue
Posts: 26
Joined: Fri Sep 14, 2012 4:28 pm
Contact:

Re: Cheating suspicion at the Zadar Open in Croatia.

Post by simonhue » Sat Jan 12, 2013 8:51 pm

http://www.chess-db.com/public/perfprob ... =Calculate
Probability to achieve >= 6.0 points: 0.000003271902477007, ~1/305632

User avatar
Ajedrecista
Posts: 1406
Joined: Wed Jul 13, 2011 7:04 pm
Location: Madrid, Spain.
Contact:

Tournament Performance Luck Calculator.

Post by Ajedrecista » Sat Jan 12, 2013 9:05 pm

Hello Simon:
simonhue wrote:http://www.chess-db.com/public/perfprob ... =Calculate
Probability to achieve >= 6.0 points: 0.000003271902477007, ~1/305632
This tool is great!

http://www.chess-db.com/public/perfprob.jsp

I did not know about its existance. I wonder what are the algorithm for this calculator... I remind that I used a probability of draw of 22% that could be too high (it was only an assumption).

I use a trinomial distribution; any info about this online tool will be much appreciated. Thank you very much for the link!

Regards from Spain.

Ajedrecista.

User avatar
simonhue
Posts: 26
Joined: Fri Sep 14, 2012 4:28 pm
Contact:

Re: Tournament Performance Luck Calculator.

Post by simonhue » Sat Jan 12, 2013 10:36 pm

Ajedrecista wrote:
This tool is great!

http://www.chess-db.com/public/perfprob.jsp

I did not know about its existance. I wonder what are the algorithm for this calculator... I remind that I used a probability of draw of 22% that could be too high (it was only an assumption).

I use a trinomial distribution; any info about this online tool will be much appreciated. Thank you very much for the link!

Regards from Spain.

Ajedrecista.
The draw rates are derived from practical statistics:
http://www.chess-db.com/public/research/draw_rate.html

In Ivanov's case ~18% range.

The tool works by enumerating all possible outcomes, and calculating the probability for each to occur. Then summing all this.

User avatar
Ajedrecista
Posts: 1406
Joined: Wed Jul 13, 2011 7:04 pm
Location: Madrid, Spain.
Contact:

Re: Tournament Performance Luck Calculator.

Post by Ajedrecista » Sun Jan 13, 2013 10:44 am

Hello again:
simonhue wrote:
Ajedrecista wrote:
This tool is great!

http://www.chess-db.com/public/perfprob.jsp

I did not know about its existance. I wonder what are the algorithm for this calculator... I remind that I used a probability of draw of 22% that could be too high (it was only an assumption).

I use a trinomial distribution; any info about this online tool will be much appreciated. Thank you very much for the link!

Regards from Spain.

Ajedrecista.
The draw rates are derived from practical statistics:
http://www.chess-db.com/public/research/draw_rate.html

In Ivanov's case ~18% range.

The tool works by enumerating all possible outcomes, and calculating the probability for each to occur. Then summing all this.
Thank you very much: it is what I wanted to know. ;) My tool uses an average of opponent's ratings (which is only an approximation) instead of each individual rating but I also knew that lowering my input probability of draw will bring closer results to ones of that online tool:
Probability to achieve exactly 0.0 points: 0.138820124747056877, ~1/7
Probability to achieve exactly 0.5 points: 0.238072629609693951, ~1/4
Probability to achieve exactly 1.0 points: 0.251974802537574517, ~1/3
Probability to achieve exactly 1.5 points: 0.185493573683177635, ~1/5
Probability to achieve exactly 2.0 points: 0.107187749981418433, ~1/9
Probability to achieve exactly 2.5 points: 0.049993388386886888, ~1/20
Probability to achieve exactly 3.0 points: 0.019559795430256514, ~1/51
Probability to achieve exactly 3.5 points: 0.006478139355314301, ~1/154
Probability to achieve exactly 4.0 points: 0.001847704350396042, ~1/541
Probability to achieve exactly 4.5 points: 0.000454071696103204, ~1/2202
Probability to achieve exactly 5.0 points: 0.000096905701835080, ~1/10319
Probability to achieve exactly 5.5 points: 0.000017842617809472, ~1/56045
Probability to achieve exactly 6.0 points: 0.000002839414544516, ~1/352185
Probability to achieve exactly 6.5 points: 0.000000384105179786, ~1/2603453
Probability to achieve exactly 7.0 points: 0.000000043967536769, ~1/22744053
Probability to achieve exactly 7.5 points: 0.000000004093525465, ~1/244288207
Probability to achieve exactly 8.0 points: 0.000000000305136562, ~1/3277221165
Probability to achieve exactly 8.5 points: 0.000000000015997155, ~1/62511114977
Probability to achieve exactly 9.0 points: 0.000000000000556751, ~1/1796132763944
---
Probability to achieve >= 0.0 points: 0.999999999999999927, ~1/1
Probability to achieve >= 0.5 points: 0.861179875252943049, ~1/1
Probability to achieve >= 1.0 points: 0.623107245643249098, ~1/1
Probability to achieve >= 1.5 points: 0.371132443105674580, ~1/2
Probability to achieve >= 2.0 points: 0.185638869422496945, ~1/5
Probability to achieve >= 2.5 points: 0.078451119441078511, ~1/12
Probability to achieve >= 3.0 points: 0.028457731054191623, ~1/35
Probability to achieve >= 3.5 points: 0.008897935623935108, ~1/112
Probability to achieve >= 4.0 points: 0.002419796268620807, ~1/413
Probability to achieve >= 4.5 points: 0.000572091918224764, ~1/1747
Probability to achieve >= 5.0 points: 0.000118020222121559, ~1/8473
Probability to achieve >= 5.5 points: 0.000021114520286479, ~1/47360
Probability to achieve >= 6.0 points: 0.000003271902477007, ~1/305632
Probability to achieve >= 6.5 points: 0.000000432487932491, ~1/2312203
Probability to achieve >= 7.0 points: 0.000000048382752704, ~1/20668522
Probability to achieve >= 7.5 points: 0.000000004415215934, ~1/226489488
Probability to achieve >= 8.0 points: 0.000000000321690469, ~1/3108578264
Probability to achieve >= 8.5 points: 0.000000000016553906, ~1/60408700663
Probability to achieve >= 9.0 points: 0.000000000000556751, ~1/1796132763944
Doing a little of trial and error, if I input D = 15.4% (there are better values, for sure):

Code: Select all

Probabilities_in_a_trinomial_distribution, ® 2013.

--------------------------------------------------------------------
Probabilities of all possible scores in a match between two engines.
--------------------------------------------------------------------

Write down the number of games of the match (up to 50):

9

Write down the engines rating difference (between -800 Elo and 800 Elo):

-345.44

Probability of a draw (%) between 0 % and  24.08 %

15.4

Write down the clock rate of the CPU (in GHz), only for timing the elapsed time of the calculations:

3

The results have been saved to Probabilities.txt file.

End of the calculations. Approximated elapsed time:   40 ms.

Thanks for using Probabilities_in_a_trinomial_distribution. Press Enter to exit.

Code: Select all

Probabilities for a match of  9 games (rounded up to 0.0001%):
 
Rating difference (rounded up to 0.01 Elo): -345.44 Elo.
 
Probability of a win  = W ~   4.3414 %
Probability of a draw = D ~  15.4000 %
Probability of a lose = L ~  80.2586 %
 
----------------------------------------------------
 
Points:  9.0/ 9
 
+ 9 = 0 - 0
P ~   0.0000 %
 
Probability of win  9.0 points out of  9:   0.0000 %
 
----------------------------------------------------
 
Points:  8.5/ 9
 
+ 8 = 1 - 0
P ~   0.0000 %
 
Probability of win  8.5 points out of  9:   0.0000 %
 
----------------------------------------------------
 
Points:  8.0/ 9
 
+ 7 = 2 - 0
P ~   0.0000 %
 
+ 8 = 0 - 1
P ~   0.0000 %
 
Probability of win  8.0 points out of  9:   0.0000 %
 
----------------------------------------------------
 
Points:  7.5/ 9
 
+ 6 = 3 - 0
P ~   0.0000 %
 
+ 7 = 1 - 1
P ~   0.0000 %
 
Probability of win  7.5 points out of  9:   0.0000 %
 
----------------------------------------------------
 
Points:  7.0/ 9
 
+ 5 = 4 - 0
P ~   0.0000 %
 
+ 6 = 2 - 1
P ~   0.0000 %
 
+ 7 = 0 - 2
P ~   0.0000 %
 
Probability of win  7.0 points out of  9:   0.0000 %
 
----------------------------------------------------
 
Points:  6.5/ 9
 
+ 4 = 5 - 0
P ~   0.0000 %
 
+ 5 = 3 - 1
P ~   0.0000 %
 
+ 6 = 1 - 2
P ~   0.0000 %
 
Probability of win  6.5 points out of  9:   0.0000 %
 
----------------------------------------------------
 
Points:  6.0/ 9
 
+ 3 = 6 - 0
P ~   0.0000 %
 
+ 4 = 4 - 1
P ~   0.0001 %
 
+ 5 = 2 - 2
P ~   0.0002 %
 
+ 6 = 0 - 3
P ~   0.0000 %
 
Probability of win  6.0 points out of  9:   0.0003 %
 
----------------------------------------------------
 
Points:  5.5/ 9
 
+ 2 = 7 - 0
P ~   0.0000 %
 
+ 3 = 5 - 1
P ~   0.0003 %
 
+ 4 = 3 - 2
P ~   0.0011 %
 
+ 5 = 1 - 3
P ~   0.0006 %
 
Probability of win  5.5 points out of  9:   0.0020 %
 
----------------------------------------------------
 
Points:  5.0/ 9
 
+ 1 = 8 - 0
P ~   0.0000 %
 
+ 2 = 6 - 1
P ~   0.0005 %
 
+ 3 = 4 - 2
P ~   0.0037 %
 
+ 4 = 2 - 3
P ~   0.0055 %
 
+ 5 = 0 - 4
P ~   0.0008 %
 
Probability of win  5.0 points out of  9:   0.0106 %
 
----------------------------------------------------
 
Points:  4.5/ 9
 
+ 0 = 9 - 0
P ~   0.0000 %
 
+ 1 = 7 - 1
P ~   0.0005 %
 
+ 2 = 5 - 2
P ~   0.0080 %
 
+ 3 = 3 - 3
P ~   0.0260 %
 
+ 4 = 1 - 4
P ~   0.0143 %
 
Probability of win  4.5 points out of  9:   0.0487 %
 
----------------------------------------------------
 
Points:  4.0/ 9
 
+ 0 = 8 - 1
P ~   0.0002 %
 
+ 1 = 6 - 2
P ~   0.0094 %
 
+ 2 = 4 - 3
P ~   0.0691 %
 
+ 3 = 2 - 4
P ~   0.1015 %
 
+ 4 = 0 - 5
P ~   0.0149 %
 
Probability of win  4.0 points out of  9:   0.1950 %
 
----------------------------------------------------
 
Points:  3.5/ 9
 
+ 0 = 7 - 2
P ~   0.0048 %
 
+ 1 = 5 - 3
P ~   0.0980 %
 
+ 2 = 3 - 4
P ~   0.3599 %
 
+ 3 = 1 - 5
P ~   0.2115 %
 
Probability of win  3.5 points out of  9:   0.6741 %
 
----------------------------------------------------
 
Points:  3.0/ 9
 
+ 0 = 6 - 3
P ~   0.0579 %
 
+ 1 = 4 - 4
P ~   0.6383 %
 
+ 2 = 2 - 5
P ~   1.1253 %
 
+ 3 = 0 - 6
P ~   0.1837 %
 
Probability of win  3.0 points out of  9:   2.0053 %
 
----------------------------------------------------
 
Points:  2.5/ 9
 
+ 0 = 5 - 4
P ~   0.4528 %
 
+ 1 = 3 - 5
P ~   2.6612 %
 
+ 2 = 1 - 6
P ~   1.9550 %
 
Probability of win  2.5 points out of  9:   5.0690 %
 
----------------------------------------------------
 
Points:  2.0/ 9
 
+ 0 = 4 - 5
P ~   2.3600 %
 
+ 1 = 2 - 6
P ~   6.9346 %
 
+ 2 = 0 - 7
P ~   1.4555 %
 
Probability of win  2.0 points out of  9:  10.7501 %
 
----------------------------------------------------
 
Points:  1.5/ 9
 
+ 0 = 3 - 6
P ~   8.1995 %
 
+ 1 = 1 - 7
P ~  10.3259 %
 
Probability of win  1.5 points out of  9:  18.5254 %
 
----------------------------------------------------
 
Points:  1.0/ 9
 
+ 0 = 2 - 7
P ~  18.3140 %
 
+ 1 = 0 - 8
P ~   6.7268 %
 
Probability of win  1.0 points out of  9:  25.0408 %
 
----------------------------------------------------
 
Points:  0.5/ 9
 
+ 0 = 1 - 8
P ~  23.8613 %
 
Probability of win  0.5 points out of  9:  23.8613 %
 
----------------------------------------------------
 
Points:  0.0/ 9
 
+ 0 = 0 - 9
P ~  13.8173 %
 
Probability of win  0.0 points out of  9:  13.8173 %
Then, my results are much more similar than before with respect to the web you linked to.

Regards from Spain.

Ajedrecista.

Lavir
Posts: 263
Joined: Sun Oct 28, 2012 10:45 am

Re: Cheating suspicion at the Zadar Open in Croatia.

Post by Lavir » Sun Jan 13, 2013 10:59 am

Hood wrote:Innocent until proven guilty. That is the rule which shall be followed.
That he was guilty is self-evident just by looking at his games on the whole. Have you studied them? An human individual doesn't suddenly became a chaotic individual when he is usually very ordained in thought, nor he becomes suddenly perfect while he is instead usually very lousy.

You can see this perfectly on the games. The games played recently are completely different in nature than those he played before. Moreover all the recent moves matches 98% of Houdini 1.5 moves and many of the strategies behind the game are obviously non-human like (an human for example very difficulty will suddenly change side of the board while he has an ongoing plan, that's not how an human mind works, while for computers this is natural).
Hood wrote: He was checked nth found, it is prove of his innocence.
"Checking" is surely not the way to find one cheating if the cheater is at the least a minimal smart. There are too many ways to bypass a search.

User avatar
geots
Posts: 4790
Joined: Fri Mar 10, 2006 11:42 pm

Re: Cheating suspicion at the Zadar Open in Croatia.

Post by geots » Tue Jan 15, 2013 12:08 pm

Lavir wrote:
Hood wrote:Innocent until proven guilty. That is the rule which shall be followed.
That he was guilty is self-evident just by looking at his games on the whole. Have you studied them? An human individual doesn't suddenly became a chaotic individual when he is usually very ordained in thought, nor he becomes suddenly perfect while he is instead usually very lousy.

You can see this perfectly on the games. The games played recently are completely different in nature than those he played before. Moreover all the recent moves matches 98% of Houdini 1.5 moves and many of the strategies behind the game are obviously non-human like (an human for example very difficulty will suddenly change side of the board while he has an ongoing plan, that's not how an human mind works, while for computers this is natural).
Hood wrote: He was checked nth found, it is prove of his innocence.
"Checking" is surely not the way to find one cheating if the cheater is at the least a minimal smart. There are too many ways to bypass a search.




Fabio, I have followed your analysis of some positions and games, and I have to admit you are as good or better than anyone I have ever seen around here at it. I don't have a dog in this hunt- but you stopped short above. All of a sudden playing way over his head is certainly telling- and I understand where you are coming from. But are you saying that in itself, with no concrete evidence, ie catching him with the goods on him- is enough to convict or find him guilty. You never really said, and I just wondered.



Best,

george

Lavir
Posts: 263
Joined: Sun Oct 28, 2012 10:45 am

Re: Cheating suspicion at the Zadar Open in Croatia.

Post by Lavir » Wed Jan 16, 2013 10:39 am

geots wrote: But are you saying that in itself, with no concrete evidence, ie catching him with the goods on him- is enough to convict or find him guilty. You never really said, and I just wondered.
No, naturally, that's not enough at all. If they did search him and didn't find nothing and that was the only way they had at their disposal then one is obviously innocent until proven guilty.

However what I'm saying is that searching alone is an abysmal method of cheating detection in itself, and so abiding to it as a way to prove someone guilty or innocent is completely at fault from beginning; i.e. you would never know for sure one way or another, and you could actually commit a terrible error on both fronts.

There are much better options and I find it strange that the only thing the organizers could come up with was searching the guy as proof of something.

Post Reply