Sedat Canbaz

[Sedat Canbaz]

Latest SCCT B/N XXV's Strength Test Results:

1 Elo Difference:

Code: Select all

1   Zeus 26.5      +43/-40/=917 50.15%  501.5/1000
2   ShashChess 32  +40/-43/=917 49.85%  498.5/1000

-----------------------------------------------------
2 Elo Difference (in both tests):

Code: Select all

1   Zeus 26.5  +30/-26/=544 50.33%  302.0/600
2   Kayra 1.5  +26/-30/=544 49.67%  298.0/600
                   
1   Kayra 1.5      +20/-17/=563 50.25%  301.5/600
2   ShashChess 32  +17/-20/=563 49.75%  298.5/600

GAMES:
https://mega.nz/file/S4R2zK6B#d-yCZbk3r ... ZXgc0eoYRs

Some Notes:
Zeus 25.5 engine is played via own nn.zeus file,
Where Kayra 1.5 played with nn-335a9b2d8a80.nnue
And ShashChess 32 with nn-ad9b42354671.nnue file

Otherwise, just in case of using default Evalfiles,
For example, then ShachChess 32 performs stronger..
For this reason, via my current setup.. they are very
Close in points, I mean almost identical in strength!

Btw, for anyone missed, interested, before start...
Just for the current new BOOK / NNUE XXV CS:
I've tested over than 100+ Top NNUE engines...
And I managed to produce 40.000 strength games!
Yes... this is not as joke... many have no clue...
Really...but I am pretty sure also in that,
Still there are good chess friends, who pay
Attention to my activities....and without them,
SCCT would not excist... thanks a lot again!

Btw, what a pitty that, there are also such guys,
Without any reason.. they blame SCCT's used engines..
Such as I use wrong conditions.... maybe that's why
They can not be as frequent Number Ones ?)
Ok.. now you have a chance... not Raubfish, not Cfish...
I use Zeus and ShachChess..let's see how good you are?!
But as we see, even via latest tour: again DON is Ledidng...
Maybe, the real, actual problem is DON ??)) just tell Me.. ))

Remember that too,
There is one way... work harder...and just do not
GIVE UP...sooner or later.. the success is in door!
If other/s managed to do that...why not you ?)

What I can say more over latest tested 100+ engines:
Almost all of them are super strong... really... but
One small point, which is one of most important:
They produced very very high draw percentages !!
E.g many of them performed close to 98%-99% draws..
And these high-draw values via Bullet cond. 1m+0.5s
Just imagine.. if via Blşitz, Rapid TC + MP+ Ponder ON...
Very likely almost all games will be ended as draw !!

And without to not mention this I can not as well,
As usual, several latest NNUE engines are crashed...
This is already known...due to high concurrency...
But these crashes are related with learning engines!
Even disabling their learning... they crash again...
It seems the programmers have difficulties to fix..
Very sad...I assume many of them are aware, but...

But I admit also that,
Mostly of my tested Top engines are very stable!
No any single crash is recorded....

Continuing....(about high draw percentages)

Sure, mostly engines are not as main reason, the real
Problem is in the Evalfiles.... and no doubt that too:
E.g the NNUE (Evalfiles) bring better/stronger quality!
But in same time, they kill the fun! did you know that?
At least I'm talking for my side...not sure about you ?!)
But I am sure in that exc. Me, many are affecting too!
Because almost all People love to see many wins ...right ?)
Plus, almost all People also love to see many games (per player)
Otherwise, the Error Elo margin can be very very high...
Just saying...

In other words,
if we'll pick one of the latest strongest Evalfiles..
The situation seems to be terrible..really...
Too much boring chess for sure...now you may blame
My current conditions too... but can you offer
Something better?... I'd be happy to hear...thanks )
If you can not.. please...
Btw, if possible via many many strength engine games..
Only via comments.. you may know.. too boring too!

And as final words (to be more clear),
I mean all about in case of running book tours!

And I hope all the current posting to be useful..

Ok.. that's all for now, thanks for reading!


Greetings )

[Sedat Canbaz]

Breaking News - 07.01.2024:
This time only 2 entries, but quite strong vers: Unicornio 2 and played also
As alternative by Kayra 1.5 engine as well... so now is much more clear that
Kayra 1.5 seems to be more aggressive..at least the draw statistics confirm..
Btw, Raubfish SC is also very aggressive, otherwise no explanation of that!
Note also that I managed to run other new strength testings, where now it's
Already clear that current Bullet tour can be counted Blitz via Raubfish SC..
Plus today I've organized another new book tour, with the current high-rated
Ones but as main diff.is used: Raubfish SC (22 books, 420 games per player)
Sure in my next updates, I hope to share all these new strength competitions..
So after that you can compare the influences, and what I can say at moment:
The error margin (22 books, 420 games) play as serious role, for this reason
I suggest a lot of games... e.g 1000+ games (per player) seems to be ideal..
Sure vs many various books, otherwise no way to see who is Best of Best
Ok, friends..that's all for now, thanks a lot for your interest.. good luck )

For Latest Standings:
https://sites.google.com/site/computers ... t-bn-cs-25

Greetings

[Sedat Canbaz]

UPDATE

Some notes more...

Code: Select all

Rank Name                Elo    +    - games score oppo. draws
  30 Grendel All        3759    5    4   940   50%  3760   97% 
  35 Grendel W          3756    5    5   660   49%  3763   94% 
  37 Solista W          3756    5    5   720   49%  3763   94% 
  56 Grendel CTG        3748    5    5   640   47%  3764   92% 
  57 Solista All        3747    5    5   580   47%  3765   91% 

More Details (for anyone missed, interested):
I managed to create derivative BIN books (based on Grendel's games):
Grendel All = Optimized to prefer latest both Grendel books versions
Grendel W = Optimized to prefer Grendel's wins (especially as White)
Sure as Blacks, all games are included, I mean the wins plus the draws
Grendel CTG = The official Grendel 081023.ctg (original, untouched)

Interesting to note that,
My created Grendel derivatives are performed better than original!

Note also that
As same words are going for Solista W book.. mainly on wins etc.
Btw, Solista CTG is not planning to be tested, as main reason:
According to my latest tests, some CTG opening books suffer...
They are safe up to 10 moves, but in deeper pos. no guarantee..


Best,
Sedat

[Sedat Canbaz]

Hello there )

Well, here are one of latest new Strength results:

1st of all,
Raubfish SC is used to play Blitz Time control (2m+1s),
Where his opponents are played via Bullet TC (1m+0.5)

Code: Select all

Kayra 1m+0.5s vs Raubfish 2m+1s: 10 Elo difference 
                    
1   Kayra 1.5          +21/-11/=308 51.47%  175.0/340
2   Raubfish x48c3 SC  +11/-21/=308 48.53%  165.0/340

Draw Ratio: 90%

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

ShachChess 1m+05s vs Raubfish 2m+1s: 2 Elo difference

Code: Select all

1   ShashChess 32      +14/-12/=314 50.29%  171.0/340
2   Raubfish x48c3 SC  +12/-14/=314 49.71%  169.0/340

Draw Ratio: 93%

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

Zeus 1m+05s vs Raubfish 2m+1s: 3 Elo difference

Code: Select all

1   Raubfish x48c3 SC  +14/-11/=315 50.44%  171.5/340
2   Zeus 26.5          +11/-14/=315 49.56%  168.5/340

Draw Ratio: 92%

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

The Overall performance (all 3 players merged in one),
Triple Brain 1m+05s vs Raubfish 2m+1s: 3 Elo difference

Code: Select all

1   Triple Brain 1m05s  +46/-37/=937 50.44%  514.5/1020
2   Raubfish SC 2m1s    +37/-46/=937 49.56%  505.5/1020

The Overall Draw Ratio (based on 1020 games): 92%

GAMES:
https://mega.nz/file/TsY2kJ7Z#trS_kLXJU ... K07GMGiRd4


More Details:
Now is already 100% clear that Raubfish SC vs Kayra 1.5 have
Really very Aggressive, more Dynamic of playing styles!
To be sure, see 1st test, where the draw ratio is lowest: 90%
Btw, actually lowest draw record of 3700+ engines belongs to
Raubfish SC vs Raubfish SC e.g usually, when matches each other:
Raubfish X48c3 SC's overall draw stats are minus plus: 80%-88%
E.g no any other engine, to be so strong as Raubfish SC and
To produce so low draw values...proved already many times!

But this is also true that,
No any other engine to be so strong as Kayra, ShachChess, Zeus
And in same time, to produce these lower draw perc. values!!

On other hand,
Here (in all recent 3 tests): all players used the same Balsa suite
Btw, for Raubfish SC: 128 MB / TripleBrain ones: 64 MB Hash used

And be aware, I realized to run all these new strength testings
To see in reality.. e.g current tests can be used as useful data
About B/N XXV Bullet games can be counted Blitz via Raubfish SC!

As final words,
'Papers' and 'Reality' are 2 (two) different things!
If still not so clear, we should not be too naive and
We should not mix Apples with Oranges, to be more clear:
The Evalfiles and Engines are two different things!

Or as other example:
Books and Engines are two different things as well!
E.g if X Evalfile is stronger...it does not mean that
Raubfish is bad, poor, crab.. etc and etc..

Or other example:
If X Book is weaker, it does not mean also that
Raubfish is bad etc and etc.

Have a nice day and hope helps...

Best,
Sedat

[Sedat Canbaz]

[Sedat Canbaz]

Breaking News - 09.01.2024:
Well, the championship is already over.. congrats and many thanks to all!
As usual, all of the played tournament games are available (in PGN): XXV
Via this update, 5 entries...after checking..their points seems to be not bad..
Haze TR by Tanick Ramz, Rodent Z by Garry Law, Notyet Dual+Taram by
Cevdet Sarı and MobY by Ian Hurnavich, sorry that I can't count all rest...
Due to many participants...no free time for all...thanks for understanding..
As I promised, all other new SCCT testings are shared: Goi + OpenChess

Private Champion: DON Hercules - Congratulations to Vasid Chouhan!
Public Champion: OPTIMUS - Congratulations to Roger Constantinesco!

For Final Standings:
https://sites.google.com/site/computers ... t-bn-cs-25

GAMES:
https://mega.nz/file/G0RCXaIS#i2iXhrpRk ... VSJj_JSe0Y

Have fun,
Sedat

[Orgz]

Breaking News - 09.01.2024:
Well, the championship is already over.. congrats and many thanks to all!
As usual, all of the played tournament games are available (in PGN): XXV
Via this update, 5 entries...after checking..their points seems to be not bad..
Haze TR by Tanick Ramz, Rodent Z by Garry Law, Notyet Dual+Taram by
Cevdet Sarı and MobY by Ian Hurnavich, sorry that I can't count all rest...
Due to many participants...no free time far all...thanks for understanding..
As I promised, all other new SCCT testings are shared: Goi + OpenChess

Private Champion: DON Hercules - Congratulations to Vasid Chouhan!
Public Champion: OPTIMUS - Congratulations to Roger Constantinesco!

For Final Standings:
https://sites.google.com/site/computers ... t-bn-cs-25

GAMES:
https://mega.nz/file/G0RCXaIS#i2iXhrpRk ... VSJj_JSe0Y

Have fun,
Sedat

Congratulations to the winners as usual. If i had started with the tour i believe i would have been in a more better position. Thanks again director!

[Sedat Canbaz]

Breaking News - 09.01.2024:
Well, the championship is already over.. congrats and many thanks to all!
As usual, all of the played tournament games are available (in PGN): XXV
Via this update, 5 entries...after checking..their points seems to be not bad..
Haze TR by Tanick Ramz, Rodent Z by Garry Law, Notyet Dual+Taram by
Cevdet Sarı and MobY by Ian Hurnavich, sorry that I can't count all rest...
Due to many participants...no free time far all...thanks for understanding..
As I promised, all other new SCCT testings are shared: Goi + OpenChess

Private Champion: DON Hercules - Congratulations to Vasid Chouhan!
Public Champion: OPTIMUS - Congratulations to Roger Constantinesco!

For Final Standings:
https://sites.google.com/site/computers ... t-bn-cs-25

GAMES:
https://mega.nz/file/G0RCXaIS#i2iXhrpRk ... VSJj_JSe0Y

Have fun,
Sedat

Congratulations to the winners as usual. If i had started with the tour i believe i would have been in a more better position. Thanks again director!

You are welcome )

Just a moment.. you are trying to say that,
If your Haze TR would be played 1260 games
Then would be in better rank position, right ?

If so.. this is as mystery... maybe higher or maybe
It would be at lower rank, who knows ?)

But however, according to my experience,
After 1000 games ( per player ), the error margin is less...
Sure depending on book entries..the points are changing
But not so much as it looks...

Actually I wonder more,
What about if all books would be new, updated...
Then very likely the current positions would be not same..
But I admit that this is simply impossible,
Because not all the book makers are active...
But in the past...many authors books were great..
Actually still some very old-dated ones (Cerebellum, CF1977, Vitamin etc)
Performed better than many of the new ones, which are released in 2023

Greetings )

[Orgz]

i guess we will never know . But i have great faith in this one.

[Sedat Canbaz]

i guess we will never know . But i have great faith in this one.

Ah ah... Mr. Tanick ) ok, Haze TR's mystery solved... e.g
+100 extra unofficial games played, now your book contains 1260 games
And now +1 point gained, nice. but unfortunately Haze could not be Winner...
Even if we will continue testing Haze with 1000 games more, there is
No much possibility to be Champion...believe Me, reality is reality!
Sure everything is possible, but before, I think/suggest:
You have to add / update with newer stronger lines... only then...
Actually if you will manage to create newer lines that never played before
Sure...if superior strong... your book will be in better positions...
Maybe even better than DON Hercules...it is all in your hands !)

And once more, as I stated earlier:
I see no point of running all weaker books over than + 1000 games
I mean same number of games such every book to include 1260 games,
Especially when creating/running huge rating lists... it will be
Just waste of time...because the Electricity is not so cheap !)

If still not so clear:
Actually if Haze TR performed in Top 5 or 10 (in Official ranking)
I'd run it as same number of games as DON Hercules, OPTIMUS etc.
Because then any highest ranked book (especially who are in Top 3)
They have much more chances to be Champion!

In other words,
If no much chances to be in Top 5 or 10...I prefer less games..
Because there are such critical weak lines in some opening books,
Really just waste of time...btw, who cares much about them ?)
Sure I test them.. and I feel BIG pleasure... but if +1000 games..
For them is too much...

And here is the Unofficial Ranking (via latest extra Haze 100 games)

Code: Select all

Rank Name                Elo    +    - games score oppo. draws 
   1 DON Hercules 3     3780    3    3  1260   54%  3760   90% 
   2 OPTIMUS 2311       3778    3    4  1260   53%  3760   92% 
   3 Red Dragon         3777    3    4  1260   53%  3760   89% 
   4 ORIONPAX4          3777    3    4  1260   53%  3760   92% 
   5 Everest 300923 Ka  3774    4    4  1060   52%  3760   88% 
   6 Everest 300923     3773    4    4  1240   52%  3759   93% 
   7 SENTINEL 4         3772    4    4  1260   52%  3760   94% 
   8 Unicornio 2 Ka     3770    4    4  1040   52%  3760   94% 
   9 Unicornio 2        3769    4    4  1120   52%  3760   95% 
  10 CachoSCCT opt2     3769    4    4  1260   52%  3760   95% 
  11 Unicornio          3769    4    4  1140   52%  3759   95% 
  12 OlisFish 201123    3768    4    4  1260   51%  3760   95% 
  13 Haze TR            3768    4    4  1260   51%  3760   90% 
  14 Eman 061223        3768    4    4  1260   51%  3760   93% 
  15 Leschats 291223    3767    4    4  1260   51%  3760   93% 
  16 Rodent Z           3767    4    5  1000   51%  3761   91% 
  17 Leschat 291223p    3766    4    4  1260   51%  3760   95% 
  18 C.B. Conqueror     3766    5    5   900   50%  3763   96% 
  19 Cubail 2024        3765    4    4  1260   51%  3760   93% 
  20 Drifter            3765    4    4  1120   51%  3761   94% 
  21 Hamish 5           3764    4    4  1260   51%  3760   96% 
  22 Xperimental        3764    5    5   920   50%  3763   95% 
  23 DragonTooth        3764    4    4  1260   51%  3760   94% 
  24 HCan UPD           3764    4    4  1220   51%  3760   96% 
  25 Africaata 311223   3763    5    5  1000   50%  3761   94% 
  26 Donmez Dual        3762    5    5   820   50%  3763   90% 
  27 Beast              3761    5    5   940   50%  3762   95% 
  28 Goddess            3761    4    4  1040   50%  3761   93% 
  29 Rebel              3760    6    5   660   49%  3765   97% 
  30 Sun                3760    5    5  1000   50%  3761   94% 
  31 Zaytsev            3760    5    5   800   50%  3762   97% 
  32 Notyet Dual        3759    5    5   760   49%  3764   94% 
  33 V by Kerveros      3759    5    5   840   49%  3763   95% 
  34 Grendel All        3758    5    5   960   50%  3760   97% 
  35 MobY v1            3758    5    5   780   49%  3764   94% 
  36 Hubble 3.3         3758    5    5   920   49%  3762   94% 
  37 Goi forum          3757    5    5   940   49%  3762   90% 
  38 LooseCanon         3757    5    5   820   49%  3763   93% 
  39 Sari 300923        3756    5    5   860   49%  3762   94% 
  40 CallmeX            3756    5    5   860   49%  3763   95% 
  41 Grendel W          3756    5    5   780   49%  3763   94% 
  42 Sambala            3755    5    5   720   48%  3765   93% 
  43 Ingkong 230923     3755    5    5   900   49%  3763   94% 
  44 Cevdet 241223      3755    5    5   780   49%  3764   91% 
  45 StudentChess       3755    5    5   900   49%  3762   93% 
  46 Solista W          3754    5    5   840   49%  3763   93% 
  47 Cerebellum X       3754    5    5   740   48%  3764   94% 
  48 Dara Light         3753    5    5   720   48%  3764   94% 
  49 HCanFree Upd       3753    5    5   720   48%  3764   92% 
  50 Echo               3753    5    5   760   48%  3763   92% 
  51 CF1977             3753    5    4   900   49%  3762   93% 
  52 Taram              3752    5    5   720   48%  3764   91% 
  53 Vitamin 290320     3751    5    5   740   48%  3764   94% 
  54 Bloom 090723       3751    5    5   640   47%  3767   90% 
  55 GutLess Git        3751    5    5   740   48%  3763   93% 
  56 Feta X             3750    5    5   680   48%  3765   92% 
  57 No Book            3750    5    5   600   47%  3765   95% 
  58 Scorpio            3750    6    5   520   47%  3766   92% 
  59 Yoruba 311223      3749    5    5   700   47%  3765   88% 
  60 Mini Mighty        3748    5    4   760   47%  3763   92% 
  61 Grendel CTG        3748    5    5   680   47%  3764   92% 
  62 Solista All        3746    5    5   600   47%  3765   91% 
  63 Bigfoot            3742    4    4   740   46%  3764   88% 
  64 Lion 110423        3739    4    4   640   45%  3765   82%