SedatChess

[Sedat Canbaz]

UPDATE 3

Well, SF-PB 280924 SC is still unstoppable...Congrats to Linrock, Skynet!!
Sure..If I played some role over that SC is Leader..then how happy I am..

By the way,
I am impressed a lot by Corchess 170924 (Bullet Champ)'s points too !
And you may notice or not, so far Corchess has the lowest draws: 89.8%

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 SF-PB 280924 SC                11       8     392   51.5%   94.9% 
   2 Corchess 170924                 5      11     392   50.8%   89.8% 
   3 SOLIDOR 180924                  1       8     395   50.1%   94.7% 
   4 C-Dragon v3.0                   0       8     393   50.0%   94.4% 
   5 SF-PB 120624                   -1       8     395   49.9%   94.7% 
   6 Brainlearn 29                  -1       8     394   49.9%   94.2% 
   7 Kookaburra 3                   -3       7     394   49.6%   95.7% 
   8 Stockfish 17                   -3       8     392   49.6%   94.6% 
   9 Killfish 280624                -4       8     392   49.4%   94.6% 
  10 Rems M 091024                  -5       9     395   49.2%   92.4% 

1967 of 2700 games finished.

[Sedat Canbaz]

Breaking News (10.10.2024):
SC SC SC.. I told you before that I never play for 2nd place..and I said
Recently that we've to pay more attention over, during:: EvalFile CS V
https://sites.google.com/site/computers ... lfile-cs-5
And once more, not only via CM9 Gladiator settings or via Books such as
Perfect, Balsa, Unique etc...plus via organizing 3800+ Elo tours too !)
In short, what does it mean? We have to work hard and smart before,
Our good dreams to come true, no any other explanation of that, right?
As other important issue is that Never loose your motivation..try again!!
Sooner or later the success will be in door.. .but just do not give up...!!
On other hand, finely (after 17 years) I broke the well-known rules..so
I titled my work (SC) to be Champ but sometimes not all in my hands!
As 2nd is Corchess..impressive as well, C- Dragon is 3rd, very good too!
About latest new updated entry Rems 091024, I expected more.. since
It's the newest updated one..but anyhow no any problem..just fun tour!

Blitz Champion: SF-PB 280924 SC - Congratulations to Linrock, Skynet
If I played role over then how happy I'm.. cause SC = nn-86aaa607b31b

FOR FINAL BLITZ STANDINGS:
https://sites.google.com/site/computers ... s-54-blitz

GAMES:
https://mega.nz/file/SkoXnICL#d3MI3zhOv ... eCxMH_syX0


Kind Regards,
Sedat Canbaz

[Sedat Canbaz]

Hello Dear Friends )

Well, here are new strength tests for future Book tours..
My mission is very simple: to find less Drawish engines..
Sure I refer for latest Top engines which are 3800+ Elo!

1st test: RemsEXP via its default EvalFile..where
SF-PB 220324 SC played via nn-b1a57edbea57.nnue

And again and again +6 Elo difference (in favor for SC)
E.g a few days ago..exactly same Elo diff. is recorded!

Code: Select all

1   SF-PB 220324 SC      +48/-29/=969 50.91%  532.5/1046
2   RemsEXP 160824  def  +29/-48/=969 49.09%  513.5/1046

DrawRatio: 92%, not so high, since played via strong lines!

-----------------------------------------------------------------------------------

2nd test: RemsEXP SC via nn-86aaa607b31b.nnue..where
SF-PB 220324 SC is played via nn-b1a57edbea57.nnue too

But this time, it seems Evalfile influence played role
As we see +1 Elo difference (in favor for RemsEXP SC)

Code: Select all

1   RemsEXP 160824 SC  +37/-35/=1000 50.09%  537.0/1072
2   SF-PB 220324 SC    +35/-37/=1000 49.91%  535.0/1072

DrawRatio: 93%, still acceptable, cause via other engines
Are expecting to be appeared much more draws for sure !
-----------------------------------------------------------------------------------

3rd test: the target to test which are different in NN styles!
But in same time to be very close in strength..this is not all,
And same time the target is also that to be not so drawish each!!

SF-PB 220324 SC = nn-b1a57edbea57.nnue
SF-POLY 020624 SC = nn-ab69aa4c1cd8.nnue

Code: Select all

1   SF-POLY 020624 SC  +60/-59/=1229 50.04%  674.5/1348
2   SF-PB 220324 SC    +59/-60/=1229 49.96%  673.5/1348

DrawRatio: 91%, the lowest record produced.. just great !!
As other good news that they are identical in performance!

-----------------------------------------------------------------------------------

GAMES:
https://mega.nz/file/OgxHGLqY#18tOt-3ww ... XEHY7DQ5U4

Conditions:
2x Epyc 7B12, CuteChess1.2.0, 1 Core, Ponder OFF, 30s+0.5s, Balsa/Unique, 64 Hash, 5-MEN

As final words,
You may know, only who are with 'weak characters' blame the strong
Opening theory... yes..it's all in out hands, as we know also that
Climbing to Everest is a very hard work..but otherwise..if preferring
The easy roads..can't be remained as unforgettable records, right ?
Because some programmers, when can't find how to improve..often
Accuse, blame the strong openings.. but via latest tours, test etc.
I demonstrated (even as non-programmer) that everything is possible!
Yes..all these tests are useful..but anyhow I'm still not much satisfied...
Because just in case BIN vs CTG matches: may appear high draw values!
Ok..in next days, I will see again what I can do...Btw, what about you ?)
E.g can you help here please regarding 3800+ via BIN+CTG future and
In same time to be not so Drawish each other..let's say close to 90% draws?
Cause so far especially 3800+ CTG future ones are much more Drawish each other...
Thanks in advance

Ok.. that's all for now see you later )


Greetings

[Sedat Canbaz]

UPDATE

Well, today let me start with this great saying please:
Don't just be part of the herd

By the way, without to not mention this I can not as well, e.g:
If we were to say that a large social group consists of people,
Who don't have their own thoughts, who constantly need guidance,
Who try to adopt what is accepted..instead of better original ideas,
And I think here only a few people would object all over...right ?)

On other hand, other good notes, which perfectly suits:
In 2008, Prof. Jens Krause of the University of Leeds,
Conducted an experiment into human herd mentality and
How people, quite literally, follow the crowd..source:
https://www.progressbb.co.uk/dont-just- ... f-the-herd

Groups of people were asked to walk randomly around a hall.
They were not allowed to communicate with one another but
Had to stay within arm’s length of another person at all times.
Among the groups, a few individuals were given more specific
Instructions on where to walk. The researchers found that,
Very quickly, people started to follow the select few who
Appeared to know where they were going.
The researchers found that it only took 5% of confident
Looking people to influence the direction of the other 95%.

In other words, it's all right that when to be in nature,
Most of animals show a herd mentality for safety reasons!

But what about Humans, Chess? NNUE testings etc.
How many of us are trying something new, better?
If it's 'green' net: 27944 downloads.. hehe ))
If not..for example: nn-86aaa607b31b, just: 497

Again I ask to all people here, what does it mean?
Actually go this way...if not 'SC' would not win !
Go + test the engines only with weak lines too..!!
Then just don't come and say again latest SF dev.
Is best..cause not 'true'.. at least in SCCT..!!
Even if future tours.. SF will be Champ..will be great..
But still will be not enough..cause what about draws?
What about rest SF's options: CTG, BIN, EXP etc. ?
For these reasons. we've have think more 'deeply'
When we are going to talk..or testing any file etc.!

And final words (once more, for anyone missed):
One day, the Dog says to a Wolf:
If you were loved, they'd take you into their homes and love you.
The Wolf replies:
If we were curious about every lap, we'd be on streets like you,
Moreover, you can't find us but we can find you in every street.

Ok friends..thanks for reading.. and please stay tuned,
Very soon new / more strength NN results are coming!

Best,
Sedat

[Sedat Canbaz]

Meanwhile (about nn-86aaa607b31b.nnue statistics):
Before organizing EvalFile CS V... the downloads were
Minus plus 400 (sure I refer for nn-86aaa607b31b.nnue)
So it seems (via SCCT LIV too) played some influences!
As we see not so much..but played some role for sure..!!
Very likely here only SedatChess visitors showed interest..
I mean if was in past.. I was active in many chess forums..
Then the number of downloads would be much increased...
But not so important...even +1 (one) is all right.. actually
If all people follow what I do..that means I'm doing it wrong!

[Sedat Canbaz]

Hello Friends,

Well, after so many attempts..even I will have difficulties to count..
But believe Me..I played many 3800 Elo testings...and as conclusion:
I could not find a Top engine 3800+ where to produce low draw-ratio..
The goal was minus plus 90-92 % but .. no such numbers are recorded
Sure I refer in case testing the current mentioned Top opening books...

But anyhow, I have good news,
I realized to try with other Top 3750 Elo engine: Raubfish GTZ SC

And why Raubfish GTZ SC?
Because is still super strong, capable to use BIN books!
Another as one of most important parts: one of least Drawish NN engines!
Sure this is not all, stable as stone..besides Raubfish's produced points
Are closest to overall ranks of latest recent organized Book NN CS XXXVI
https://sites.google.com/site/computers ... k-nn-cs-36
What we can need more ? almost perfect, right?..exc.Raubfish is not capable
To use CTG books...but SF-SCCT (by Tanick Ramz) can be as solution here...

A new Unofficial Ranking by Raubfisch GTZ SC:

Code: Select all

1   Chucaro 130424  100.0 - 100.098.0 - 102.0107.0 - 93.0115.5 - 84.5**     420.5/800
2   SENTINEL 2407   100.0 - 100.0102.5 - 97.5105.5 - 94.5104.0 - 96.0 **    412.0/800
3   Broth 36        102.0 - 98.097.5 - 102.598.5 - 101.5106.5 - 93.5  **   404.5/800
4   SENTINEL 2409   93.0 - 107.094.5 - 105.5101.5 - 98.5109.0 - 91.0   **  398.0/800
5   Geralt          84.5 - 115.596.0 - 104.093.5 - 106.591.0 - 109.0    ** 365.0/800

Note: Draw-Ratio: 86%

Meanwhile here is latest Raubfish Strength Test:
As we see about 5+ Elo in favor for GTZ version !

Code: Select all

1   Raubfisch GTZ SC  +158/-132/=1394 50.77%  855.0/1684
2   Raubfish  Std SC  +132/-158/=1394 49.23%  829.0/1684

Draw-Ratio, just: 83% and this is not as joke !!

Conditions:
2x Epyc 7B12, CuteChess1.2.0, 1 Core, Ponder OFF, 30s+0.5s, Balsa/Unique, 64 Hash, 5-MEN

GAMES:
https://mega.nz/file/D1hxVDYJ#qh5nzwJvU ... 8qBSUjSZng

In other words,
Very likely next tour, will be played as main engine: Raubfisch GTZ SC

Greetings

[Sedat Canbaz]

Dear Chess Friends,

As usual, I am very pleased to announce that
New book CS is coming very soon, tour start:
It's expecting in next days..but I hope to work
Over it for many days, so in case of interest to
Submit a opening book, you can Contact Me
And in case ... please just don't be too late,
So from today, you can submit new book/s..

What's New:
Almost all BIN Book Entrants will be played by
Raubfish GTZ27c3 SC AVX2, for more: Details
CTG books will be played by SF-SCCT engine

Tournament Rules:
All opening books are allowed: Private plus Public!
Several books (BIN+CTG) by same author allowed,
Note also each other have to be different in styles..!
Be aware for CTG is suggested green lines tuning!
If not..then many CTG's lines will be played randomly..
Up to 250 MB for BIN / 500 MB for CTG (compressed)
As Bonus (for frequent) I can accept even larger sizes
Btw, against cheating: from unknown authors recent
Games required, played on Lichess, Playchess etc.
A rule can be called VISA, thanks for understanding!

That's all for now exc. be ready for the real ARENA!
And via this tour..let's see who will say the 'last' word..
Sure just in case, if your book will be ranked at last...
Not so important as well, the most important is FUN!

And as a final note.
I hope to see many registrations, thanks in advance )

For Official Tournament Page:
https://sites.google.com/site/computers ... k-nn-cs-37

Best Wishes,
Sedat Canbaz

[Sedat Canbaz]

UPDATE

1st. Test: Raubfisch GTZ - Itself Matches:

Code: Select all

1   Raubfisch (Copy)  +87/-86/=839 50.05%  506.5/1012
2   Raubfisch GTZ SC  +86/-87/=839 49.95%  505.5/1012

Note: Two copy of GTZ eng vers. are used...where the
Draw-Ratio is again 83%, not so high numbers for sure!

----------------------------------------------------------------------------------------

2nd.Test (in case of CTG book entries):
SF-SCCT 250122 vs Raubfisch GTZ SC: identical in strength !!
Sure I've expected to appear close points, but not so much )
And frankly I can not remember to appear identical points..
Even if running as same Eng copies !!

In other words,
That shows not only engines strength, plus hardware stability,
And this is not all, the played/used openings etc. as well !!

Code: Select all

1   Raubfisch GTZ SC    +145/-145/=1710 50.00%  1000.0/2000 
2   SF-SCCT 250122 def  +145/-145/=1710 50.00%  1000.0/2000 

Draw-Ratio is 86% slightly higher.. but not too bad!

GAMES:
https://mega.nz/file/69gSUSoY#5_1mUDBbY ... iKcr01-s6c

Conditions used: sure adapted to planning New CS XXXVII...
As main difference, Balsa + Unique openings are played...

Greetings )

[Sedat Canbaz]

UPDATE 2

Finely the Tour is started.. and here are 1st played results,
Where for right now DONFree leads...

Btw, If newer entries,
Sure later these books will be played In Gauntlet mode...
And in the next days many more entries are expecting too

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 DONfree 090724                 44      30      72   56.3%   84.7% 
   2 Unicornio Upd2                 38      28      74   55.4%   86.5% 
   3 AdrianOliva v3                 24      25      72   53.5%   90.3% 
   4 Tunafish 3.2                   24      31      73   53.4%   84.9% 
   5 SOLIDOR Dual                   24      28      73   53.4%   87.7% 
   6 Chucaro 130424                 19      26      72   52.8%   88.9% 
   7 Esculapio                      19      29      73   52.7%   86.3% 
   8 MegaTron RevA                  15      25      71   52.1%   90.1% 
   9 Vorden B1                      14      16      72   52.1%   95.8% 
  10 HCAN 160524                    14      28      72   52.1%   87.5% 
  11 Skynet 180724                  14      27      74   52.0%   87.8% 
  12 Rasp 070924                    10      30      72   51.4%   86.1% 
  13 DeadLy CRiMe                    5      21      71   50.7%   93.0% 
  14 UpAndDown                       5      21      72   50.7%   93.1% 
  15 Dara 141024                     5      24      74   50.7%   90.5% 
  16 The Big Castle                  0      38      71   50.0%   77.5% 
  17 Knuckle Shuffle v1              0      34      69   50.0%   82.6% 
  18 OlisFish 141024                -5      16      74   49.3%   95.9% 
  19 Castigo 050224                 -5      21      73   49.3%   93.2% 
  20 SappHire 260924                -5      41      72   49.3%   73.6% 
  21 Cevdet 131024                  -5      34      72   49.3%   81.9% 
  22 HCanFree v2                    -5      21      71   49.3%   93.0% 
  23 Rodent Z                      -10      43      71   48.6%   71.8% 
  24 Rebel 020624                  -10      30      71   48.6%   85.9% 
  25 Geralt 051024                 -10      24      70   48.6%   91.4% 
  26 OPTIMUS 2409                  -14      25      73   47.9%   90.4% 
  27 OrionPax 2407                 -19      30      72   47.2%   86.1% 
  28 LesChats 240824               -25      25      71   46.5%   90.1% 
  29 Broth 131024                  -29      33      71   45.8%   83.1% 
  30 SENTINEL 2409                 -34      39      71   45.1%   76.1% 
  31 MiniFish                      -44      30      72   43.8%   84.7% 
  32 Yoruba 081024                 -59      40      71   41.5%   74.6% 

1151 of 9920 games finished.

GOOD LUCK TO ALL )

[Sedat Canbaz]

Breaking News (15.10.2024):
XXXVII CS already started... and here are the latest results (after 9920 games)
Via this update many trend books entered where mostly of them are quite new..
E.g all of them created in 2024 that means they represent latest strong theory!
What I can add more: 620 games is good data but not enough so let's see what
Will be next ranks, but there is one true, Messi is Messi!! as we see again many
His books are on Top..sure Tunafish, Skynet..are ones which are dangerous too!!
Be aware that too please, in the next days..newer DON book vers may entered..
You may know, Mr. Vasid Chouhan is another Top book master, who never play
For 2nd place..In other words, a lot of fun is waiting us..Ok friends, now I'm busy..
So more info in next updates..good luck to all authors..and let's best book wins!
Btw, for moment. CTG are NOT accepted as reason even the defeats are played

Current Book Leader: AdrianOliva v3 - My Congratulations to Angel Morano!

For Latest Standings:
https://sites.google.com/site/computers ... k-nn-cs-37