SedatChess

[Sedat Canbaz]

Thank you very much Sedat!

You are welcome my friend

And you are one of good people that
I realized to be active member ...

Also many thanks to Openchess admin,
For his kind support too!

Greetings

[Sedat Canbaz]

Unofficial Test - Lichess 2024 played by SF-PB 290824 SC

For more information, please see 'More Details'

Code: Select all

Rank Name              Elo    +    - games score oppo. draws 
   1 Skynet 180724    3830    4    4  1000   64%  3769   72% 
   2 DON 261024       3828    4    4  1000   63%  3769   72% 
   3 SOLIDOR Dual     3828    4    4  1000   63%  3769   73% 
   4 Kill Bookson 2   3828    4    4  1000   63%  3769   73% 
   5 Esculapio        3826    4    4  1000   62%  3769   75% 
   6 Cevdet 131024    3825    4    4  1000   62%  3769   75% 
   7 Turkey 2024      3824    4    4  1000   62%  3769   76% 
   8 LesChats 240824  3824    4    4  1000   62%  3769   77% 
   9 RubiX            3822    4    4  1000   61%  3769   75% 
  10 UpAndDown        3816    4    4  1000   59%  3769   80% 
  11 Dara 141024      3815    4    4  1000   59%  3769   81% 
  12 OPTIMUS 2409     3815    4    4  1000   59%  3769   81% 
  13 No Book NN       3811    4    5  1000   58%  3769   83% 
  14 Lichess 2024     3810    4    5  1000   57%  3769   77% 
  15 Redfish 240122   3778    4    4  1400   42%  3822   83% 
  16 Obsidian 13      3774    4    4  1400   40%  3822   78% 
  17 BrainLearn 13.1  3772    4    3  1400   40%  3822   78% 
  18 Cfish 120821     3771    4    3  1400   40%  3822   78% 
  19 Caissa 1.21      3768    3    3  1400   39%  3822   77% 
  20 ShashChess 20    3767    3    3  1400   38%  3822   76% 
  21 Honey v14        3766    3    3  1400   38%  3822   75% 
  22 Berserk 12       3766    3    3  1400   38%  3822   74% 
  23 Alexandria 7.1   3765    3    3  1400   38%  3822   74% 
  24 K. Dragon 3      3764    3    3  1400   37%  3822   74% 

GAMES:
https://mega.nz/file/S5IHiRqS#TsDE_LayC ... ZYKBe2wzvw

More Details:
Just out of curiosity, I realized to test exactly the same book,
Which are used to play all the weaker (constant) chess engines!
As I expected, Lichess 2024 performed not as best, but for better
Performance required special tuning and beta testings..BTW, if
I'm not wrong.. I spent approx. 15 min..anyhow not a weak book,
Because as we see close to successful ones of previous XXXVII CS

Note also that,
This time Lichess 2024 book's games are recorded as 30sec,
Since I used Chessbsase program, copy-paste from pgn to pgn
Reason,I needed to share only 1000 games out of 14000 games
Actually there are some other ways, as without Chessbase, e.g
Each time I have to set for games a new PGN output..but will
Be too complicated for each test etc. so now I hope you give
Understanding.. otherwise, sharing all database (14000 games)
Will be pointless and meaningless, right? that's why you have
To be in my shoes (as TD) to understand about what's going on )

Greetings

[Sedat Canbaz]

Hello Chess Friends,

As usual, I'm pleased to announce:
A new tournament idea is born:
I am going to organize a new CS with engines,
Which are capable to learn... since recently
We talked much about learning chess engines )
If not running it..there will be a BIG miss...!)

Yes..we will see about which ones will be best,
You know..we can't know until we try, right?)
Moreover, after checking...no any new data is
Available in latest months, actually available
But only in SCCT! anyhow these tours are already
Old-dated! Sure if I am missing something here,
Please correct Me... regarding all over...
At least I may save time, I mean if available ..

On other hand,
Rather I'd say a new Rating (not tour) will be..
If short tours (e.g small number games per player)
No much ways to see which are strongest..! But
If running Rating (with a lot of games per player),
Then no doubt that the error margin will be less!

By the way,
Why often I mention all these important tips etc?
Simply because to gain...e.g some TDs miss a lot..
Sure anyone is free, but it will be not bad idea
At least once, to work over a good rating, e.g if using
Strong openings, run about 1000+ games (per player),
If weak lines run min. 4000-5000+ games (per player)
Otherwise, believe it or not it is as mystery to see
Regarding which is better / stronger etc
And just in case...please don't forget to include a lot
Lot of players too, as we know, the more, the better!

Be aware too, for better eng learning effect:
A lot players will be played in Round-Robin!

And please stay tuned, more info soon )

Greetings

[Sedat Canbaz]

UPDATE

Well, the Championship is Started...! and for right
Now, Brainlearn 30 is current Leader! Impressive !!
Sure we are just in beginning.. let's see after 1000
Games...about what will be next new standings...

Note also that,
I realized to include many former Champions as well..
And stated earlier, all are playing as learning enabled!

Be aware too please,
Special version of Unique opening suite is used..
Based on 60 openings.. as you see , I pick small
Number of openings to increase learning effects!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Brainlearn 30                  36      29     137   55.1%   75.2% 
   2 Rems EXT 160824                31      30     135   54.4%   73.3% 
   3 Spectral 5.0                   23      26     138   53.3%   79.0% 
   4 SunLight 3 sl                  20      29     138   52.9%   75.4% 
   5 Stockfish 220823 sl            18      28     137   52.6%   75.9% 
   6 Incognito 5 Pro                16      28     134   52.2%   77.6% 
   7 Eman 10                        13      31     136   51.8%   71.3% 
   8 Eman 9.51                       8      28     135   51.1%   77.0% 
   9 Tactical 281023                 3      27     135   50.4%   78.5% 
  10 RapTora 2.2                     0      30     136   50.0%   73.5% 
  11 Cool Iris 13.90                -5      28     136   49.3%   76.5% 
  12 BrainLearn 26                  -5      26     136   49.3%   80.9% 
  13 Predator X5                   -10      32     136   48.5%   70.6% 
  14 AIO 1.02                      -15      26     136   47.8%   79.4% 
  15 Raubfisch X48e                -28      27     136   46.0%   78.7% 
  16 SolistaAttack v2              -29      27     134   45.9%   78.4% 
  17 ProteusSF Aureo               -31      30     137   45.6%   73.7% 
  18 AI 28.0                       -44      29     136   43.8%   74.3% 

1224 of 9180 games finished.

More Details:
In the end, after the competition...the Winner will be played
As without learning...then we may have some new data about
The influences: Learning vs Non-Learning...

And according to my experience over long past years:
Best learning engines are much less potential to play same
Blunders, while the ones, which are not capable to learn,
As result, (sure here I refer in case of critical positions...)
They may repeat exactly the same mistakes over and over!

In other words,
Never stop 'learning' if wish you stay always young and strong!)

And more new updates soon as possible..

Best,
Sedat

[Sedat Canbaz]

UPDATE 2

Well, Brainlearn 30 is leading as before,
Where 2nd Spectral 5 and 3rd Eman 10

Sure still too early for final conclusion,
But I guess Brainlearn will be Champion!
Because as we see, it's points clearly ahead ...

On other hand, interesting to point out,
Some older versions perform even better
Than many newer releases..!!

But this is also true that,
The oldest ones (Raubfish, SolistAtack etc) are at last places..
I mean here no any BIG surprise..

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Brainlearn 30                  34      15     436   54.8%   78.0% 
   2 Spectral 5.0                   22      15     436   53.2%   78.0% 
   3 Eman 10                        21      16     439   53.1%   74.3% 
   4 Incognito 5 Pro                18      15     436   52.5%   77.5% 
   5 Rems EXT 160824                14      16     436   52.1%   75.7% 
   6 Stockfish 220823 sl            11      14     434   51.6%   81.6% 
   7 SunLight 3 sl                   6      16     434   50.9%   75.1% 
   8 BrainLearn 26                  -1      14     439   49.9%   80.2% 
   9 Cool Iris 13.90                -1      15     437   49.9%   78.7% 
  10 AIO 1.02                       -2      15     438   49.7%   78.8% 
  11 Eman 9.51                      -2      16     437   49.7%   76.4% 
  12 Tactical 281023                -2      15     435   49.7%   79.1% 
  13 RapTora 2.2                    -3      16     436   49.5%   77.1% 
  14 ProteusSF Aureo               -17      16     436   47.6%   75.5% 
  15 AI 28.0                       -18      17     436   47.5%   72.9% 
  16 Predator X5                   -20      16     436   47.1%   75.5% 
  17 SolistaAttack v2              -29      15     437   45.8%   77.3% 
  18 Raubfisch X48e                -31      16     436   45.5%   74.5% 

3927 of 9180 games finished.

[Sedat Canbaz]

IMPORTANT UPDATE (ABOUT ERROR MARGIN)

Check please Raubfish ranks between
UPDATE 1 vs UPDATE 2

Where you will notice that they are not same...
And why? over long past years what I say ?)
1000+ games (not 100 or 200 or 300 etc ) per play needed!
Sure for accurate, true metrics...otherwise it's hard to
Determine the right standings...at least current,
Actıually mostly test results indicate this, right ?)

Greetings

[Sedat Canbaz]

UPDATE 3

Here are latest Standing after 6562 games,
1st Brainlearn 2nd Incognito, 3rd Spectral

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Brainlearn 30                  39      12     729   55.6%   77.1% 
   2 Incognito 5 Pro                22      12     727   53.2%   76.5% 
   3 Spectral 5.0                   20      12     729   52.9%   76.7% 
   4 Eman 10                        16      12     730   52.3%   76.3% 
   5 Rems EXT 160824                15      12     729   52.2%   75.6% 
   6 Stockfish 220823 sl             9      11     729   51.2%   79.4% 
   7 Tactical 281023                 6      11     730   50.9%   80.4% 
   8 SunLight 3 sl                   6      13     727   50.8%   75.0% 
   9 Cool Iris 13.90                 3      12     729   50.4%   78.9% 
  10 BrainLearn 26                   2      11     726   50.3%   79.3% 
  11 Eman 9.51                       0      12     728   50.1%   77.3% 
  12 AIO 1.02                       -2      11     730   49.7%   79.9% 
  13 RapTora 2.2                    -8      12     729   48.8%   76.8% 
  14 AI 28.0                       -16      13     732   47.7%   74.7% 
  15 Predator X5                   -16      12     730   47.7%   76.6% 
  16 ProteusSF Aureo               -19      12     730   47.3%   76.2% 
  17 SolistaAttack v2              -31      12     731   45.5%   76.7% 
  18 Raubfisch X48e                -46      13     729   43.5%   71.1% 

6562 of 9180 games finished.

[Sedat Canbaz]

Breaking News (24.11.2024):
A new championship organized...Congrats and many thanks to all authors!
And as I stated earlier, as participants are included, which are capable to
Learn...In other words, all of the engines are played with Eng Learning ON..
Exception, the Winner (Brainlearn 30) is played as without learning as well
As we see, Brainlearn's learning effect is played not so BIG influence over!
Who knows? maybe more games required for better points differences ?!
Anyhow, Brainlearn's performance is just great, managed to be Champion!
Interesting to point out Incognito 5 Pro, even as old-dated but still among
Best ones...in the end, a master piece and created by Eduard Nemeth, is
This all? not of course his engines are great in Anayzing mode too Results
Also good performance come by other engines Rems, Eman, Spectral etc.
Btw, not sure about over but according to some members, Eman 10 is fake
I downloaded it from Openchess, shared by SOLİDOR member..and sorry
For the trouble.I mean if this is true regarding not the real Eman 10 engine
One thing more, Shashchess Santiago's learning did not work..so was out
And without to not mention this I can't, sorry (special note for all Trollers):
Everyone has a PC..What do you wait? go organize better tours, good luck..
Btw, What is TROLL (in Chess)? Answer: they sit in front of computer..and
Wondering how to attack but a small note, only via papers! on real Arena,
Hm..they're invisible like Ghost) if not clear, come I'm here but all together!
Yes..these 'clever' guys are so cheap! after checking theirs activities..even
1% they're not..sure if going comparing them with SCCT...reality is reality!
Frankly after all, what makes Me happy?) No doubt that I 'm in theirs eyes:
So good )) as we see, since over years: Sedat or SCCT etc. what a name!!
And without the orders of bOsS oN PaPeR (M.Z) these Trollers are 0 (zero)!
And hope helps )

LV's Champion: Brainlearn 30 - Congratulations to K. Kiniama, A. Manzo!

FOR FINAL STANDINGS :
https://sites.google.com/site/computers ... ning-on-55

GAMES:
https://mega.nz/file/elwEUaiS#rf4Ad0ID- ... b6ECp9crus


Best,
Sedat

[Sedat Canbaz]

Breaking News (25.11.2024):
I have good news, SCCT NNUE CS LV - 3750+ resumed!
Incognito 5 Pro Off (Learning disabled) is entered..
And let's see what will be influences as ON and OFF?
Since we have a good data about recent played result:

Code: Select all

Rank Name                  Elo    +    - games score oppo. draws 
   3 Incognito 5 Pro      3804    9    9  1080   53%  3787   77% 

By the way,
Here are LIVE standings, playing Incognito as learning OFF:

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   0 Incognito 5 Pro Off            -5      22     225   49.3%   77.3% 
   1 Incognito 5 Pro                97     102      11   63.6%   72.7% 
   2 Spectral 5.0                   64      84      11   59.1%   81.8% 
   3 Rems EXT 160824                64      84      11   59.1%   81.8% 
   4 AIO 1.02                       50     124      14   57.1%   57.1% 
   5 SolistaAttack v2               35     122      10   55.0%   70.0% 
   6 ProteusSF Aureo                29      55      12   54.2%   91.7% 
   7 Eman 9.51                      29     100      12   54.2%   75.0% 
   8 Eman 10                        29     100      12   54.2%   75.0% 
   9 SunLight 3 sl                   0       0      12   50.0%  100.0% 
  10 Stockfish 220823 sl             0       0      11   50.0%  100.0% 
  11 Brainlearn 30 Off               0      82      12   50.0%   83.3% 
  12 BrainLearn 26                   0      82      12   50.0%   83.3% 
  13 Cool Iris 13.90               -25      85      14   46.4%   78.6% 
  14 Raubfisch X48e                -29     100      12   45.8%   75.0% 
  15 Brainlearn 30                 -29     100      12   45.8%   75.0% 
  16 Tactical 281023               -32      60      11   45.5%   90.9% 
  17 Predator X5                   -32     110      11   45.5%   72.7% 
  18 AI 28.0                       -54     106      13   42.3%   69.2% 
  19 RapTora 2.2                   -89     165      12   37.5%   41.7% 

225 of 1140 games finished.

After all,
It's a good idea to check, to learn all over, right ?)
As we know,
The ones who never learns unfortunately die too young!

Is this all? not of course, in Russian:
Vek Zhivi, Vek Uchis = Live and Learn!

Is this all? not of course:
Check SCCT Results and Learn !)

Otherwise,
You are a pawn that you don't know it's a piece of chess!

Greetings

[Sedat Canbaz]

UPDATE 4

Well, Incognito 5 Pro (Learning Off)'s Gauntlet test ended, where
Performed about 7 (seven) Elo weaker than Incognito 5 Pro learning

For Final Results:
https://sites.google.com/site/computers ... ning-on-55

Games:
https://mega.nz/file/TgxDAaCZ#yguNRZbGl ... 62DMQ358Gk

After all, What does it mean? Answer, simply:
Incognito 5 Pro's learning effect is just impressive!
As same words are going for many other engines too..!

But I admit that too some suffer..and let's hope in
Next new releases their learning to be more effective...

Have a nice day!

Best,
Sedat