Ramz Tours & TESTS

[Sedat Canbaz]

Dear Sedat great idea!
My cutechess settings
Depth: 13 plies
8 concurrent games

UHO_2022_8mvs_+170_+179.pgn
https://pixeldrain.com/u/Bx4KaYtK

Dear Tanick,

Thanks again....

Yes.. the idea is not so so bad... )

But at least I can say,
The planning new test is never seen before for sure )
But you know, my tours, testings do not looks to others..

I wonder really...and in my opinion is lottery to see, to determine
the real Elo difference.. if two Top engines and close in strength...
Sure I mean if running via UHO and just 150 games (per player)

But anyhow, after all,
Time will tell about who will win each time.. right ? )

Btw, after quick checking...(your previous test)
I noticed very bad games..just one example,
Which is ended in 12 moves....and the other game
Also ended in 12 moves.. I mean same opening as twice..

Code: Select all

[Event "My Tournament"]
[Site "?"]
[Date "2024.03.24"]
[Round "66"]
[White "Stockfish 240322_avx2"]
[Black "Marauders 3.0_avx2"]
[Result "1-0"]
[ECO "A10"]
[PlyCount "24"]
[EventDate "2024.??.??"]
[TimeControl "60+1"]

1. c4 {book} d6 {book} 2. d4 {book} f5 {book} 3. Nc3 {book} Nf6 {book} 4. Bg5 {
book} g6 {book} 5. Bxf6 {book} exf6 {book} 6. e3 {book} Bg7 {book} 7. Bd3 {book
} f4 {-1.62/19 4.2s} 8. exf4 {+1.89/19 3.7s} f5 {-1.83/20 4.3s} 9. Nf3 {
+1.91/19 2.3s} O-O {-1.90/19 4.7s} 10. h4 {+2.05/17 1.4s} c5 {-1.82/19 7.0s}
11. d5 {+1.77/20 5.3s} Re8+ {-1.82/19 5.1s} 12. Kf1 {+1.82/19 2.1s} h5 {
-1.82/19 3.0s, White wins by adjudication} 1-0

What I can say more,
FUN is important....but at least the openings should
A little bit more serious... otherwise...what a pity that,
I can not say such as FUN.. just waste of time...

And let's hope to see, to appear less similar games ))
But via UHO openings... no any guarantee..right ?? ))

Greetings

[Sedat Canbaz]

1st Test is ended... in favor for Stockfish
And very soon more same tests are coming..

Code: Select all

Score of Marauders 3 vs Stockfish 240322: 32 - 33 - 85 [0.497]
...      Marauders 3 playing White: 32 - 1 - 42  [0.707] 75
...      Marauders 3 playing Black: 0 - 32 - 43  [0.287] 75
...      White vs Black: 64 - 1 - 85  [0.710] 150
Elo difference: -2.3 +/- 36.7, LOS: 45.1 %, DrawRatio: 56.7 %
150 of 150 games finished.

[Sedat Canbaz]

2nd Test is ended too, where this time in favor for Marauders
And very soon 3rd of the same kind test is coming as well...

Code: Select all

Score of Marauders 3 vs Stockfish 240322: 24 - 22 - 104 [0.507]
...      Marauders 3 playing White: 23 - 0 - 52  [0.653] 75
...      Marauders 3 playing Black: 1 - 22 - 52  [0.360] 75
...      White vs Black: 45 - 1 - 104  [0.647] 150
Elo difference: 4.6 +/- 30.8, LOS: 61.6 %, DrawRatio: 69.3 %
150 of 150 games finished.

[Sedat Canbaz]

3rd Test is ended too, this time again in favor for Stockfish
But in current 3rd test, SF managed to win via BIG margin.. 16 ELO..
Well, If I was new or so naive in chess.. perhaps I'd believe but I am not so...
Right ?) or maybe I am missing something...?)

Code: Select all

Score of Marauders 3 vs Stockfish 240322: 26 - 33 - 91 [0.477]
...      Marauders 3 playing White: 25 - 1 - 49  [0.660] 75
...      Marauders 3 playing Black: 1 - 32 - 42  [0.293] 75
...      White vs Black: 57 - 2 - 91  [0.683] 150
Elo difference: -16.2 +/- 34.9, LOS: 18.1 %, DrawRatio: 60.7 %
150 of 150 games finished.

And here is overall performance (based on 450 games): + 5 Elo in favor for SF

Code: Select all

1   Stockfish 240322  +88/-82/=280 50.67%  228.0/450
2   Marauders 3       +82/-88/=280 49.33%  222.0/450

More Details:
For all above matches are used same UHO opening suite,
Which is played/adapted in close cond. as Ramz test...

Anyhow, another test idea is born as well,
via Balsa...let's see what will be situation...?)
Be aware that all games will be shared soon...

And please stay tuned

[Cengiz-Baba]

3rd Test is ended too, this time again in favor for Stockfish
But in current 3rd test, SF managed to win via BIG margin.. 16 ELO..
Well, If I was new or so naive in chess.. perhaps I'd believe but I am not so...
Right ?) or maybe I am missing something...?)

Code: Select all

Score of Marauders 3 vs Stockfish 240322: 26 - 33 - 91 [0.477]
...      Marauders 3 playing White: 25 - 1 - 49  [0.660] 75
...      Marauders 3 playing Black: 1 - 32 - 42  [0.293] 75
...      White vs Black: 57 - 2 - 91  [0.683] 150
Elo difference: -16.2 +/- 34.9, LOS: 18.1 %, DrawRatio: 60.7 %
150 of 150 games finished.

And here is overall performance (based on 450 games): + 5 Elo in favor for SF

Code: Select all

1   Stockfish 240322  +88/-82/=280 50.67%  228.0/450
2   Marauders 3       +82/-88/=280 49.33%  222.0/450

More Details:
For all above matches are used same UHO opening suite,
Which is played/adapted in close cond. as Ramz test...

Anyhow, another test idea is born as well,
via Balsa...let's see what will be situation...?)
Be aware that all games will be shared soon...

And please stay tuned

My dear friend Sedat,

As far as we understand, it seems that thousands or even tens of thousands of matches are necessary to reach the closest to the correct conclusion.

I have fun doing the tests, but I get a little sad when I get different results. Anyway, I guess I should be Kasparov and beat the machines myself, example deep blue

Best regards,
Fringe

[Sedat Canbaz]

3rd Test is ended too, this time again in favor for Stockfish
But in current 3rd test, SF managed to win via BIG margin.. 16 ELO..
Well, If I was new or so naive in chess.. perhaps I'd believe but I am not so...
Right ?) or maybe I am missing something...?)

Code: Select all

Score of Marauders 3 vs Stockfish 240322: 26 - 33 - 91 [0.477]
...      Marauders 3 playing White: 25 - 1 - 49  [0.660] 75
...      Marauders 3 playing Black: 1 - 32 - 42  [0.293] 75
...      White vs Black: 57 - 2 - 91  [0.683] 150
Elo difference: -16.2 +/- 34.9, LOS: 18.1 %, DrawRatio: 60.7 %
150 of 150 games finished.

And here is overall performance (based on 450 games): + 5 Elo in favor for SF

Code: Select all

1   Stockfish 240322  +88/-82/=280 50.67%  228.0/450
2   Marauders 3       +82/-88/=280 49.33%  222.0/450

More Details:
For all above matches are used same UHO opening suite,
Which is played/adapted in close cond. as Ramz test...

Anyhow, another test idea is born as well,
via Balsa...let's see what will be situation...?)
Be aware that all games will be shared soon...

And please stay tuned

My dear friend Sedat,

As far as we understand, it seems that thousands or even tens of thousands of matches are necessary to reach the closest to the correct conclusion.

I have fun doing the tests, but I get a little sad when I get different results. Anyway, I guess I should be Kasparov and beat the machines myself, example deep blue

Best regards,
Fringe

Dear Fringe,

1st of all,
You run too small number of games.. such as
100-150 games (per player), why ?))

On other hand,
What I am trying to tell all of you..sure via current tests..
Do not waste your time...sure if using UHO ...exc.
5.000+, 6.000+ games (per player ) is required..
But if running not repeating same opening as twice..
Then suggested over than 20.000 games (per player)

In other words,
It's time to learn...otherwise will be be pity..
Sure my fingers are already tired to explain...
But anyhow time to time I will do my best as before...

Hope helps...

Best,
Sedat

[Cengiz-Baba]

3rd Test is ended too, this time again in favor for Stockfish
But in current 3rd test, SF managed to win via BIG margin.. 16 ELO..
Well, If I was new or so naive in chess.. perhaps I'd believe but I am not so...
Right ?) or maybe I am missing something...?)

Code: Select all

Score of Marauders 3 vs Stockfish 240322: 26 - 33 - 91 [0.477]
...      Marauders 3 playing White: 25 - 1 - 49  [0.660] 75
...      Marauders 3 playing Black: 1 - 32 - 42  [0.293] 75
...      White vs Black: 57 - 2 - 91  [0.683] 150
Elo difference: -16.2 +/- 34.9, LOS: 18.1 %, DrawRatio: 60.7 %
150 of 150 games finished.

And here is overall performance (based on 450 games): + 5 Elo in favor for SF

Code: Select all

1   Stockfish 240322  +88/-82/=280 50.67%  228.0/450
2   Marauders 3       +82/-88/=280 49.33%  222.0/450

More Details:
For all above matches are used same UHO opening suite,
Which is played/adapted in close cond. as Ramz test...

Anyhow, another test idea is born as well,
via Balsa...let's see what will be situation...?)
Be aware that all games will be shared soon...

And please stay tuned

My dear friend Sedat,

As far as we understand, it seems that thousands or even tens of thousands of matches are necessary to reach the closest to the correct conclusion.

I have fun doing the tests, but I get a little sad when I get different results. Anyway, I guess I should be Kasparov and beat the machines myself, example deep blue

Best regards,
Fringe

Dear Fringe,

1st of all,
You run too small number of games.. such as
100-150 games (per player), why ?))

On other hand,
What I am trying to tell all of you..sure via current tests..
Do not waste your time...sure if using UHO ...exc.
5.000+, 6.000+ games (per player ) is required..
But if running not repeating same opening as twice..
Then suggested over than 20.000 games (per player)

In other words,
It's time to learn...otherwise will be be pity..
Sure my fingers are already tired to explain...
But anyhow time to time I will do my best as before...

Hope helps...

Best,
Sedat

My dear friend Sedat,

The reason why I play 100-150 games per engine is all about time loss. I have several work computers and I work with all of them, I run tests against my chess curiosity whenever I can find a free moment, unfortunately I can't keep the tests open during working hours.

Lately, if you are aware, I have been playing between 100-150 games per engine, totalling 560 games in tournaments. My newly started tournament now covers 900 games.
You know that new engine updates are coming very fast. I am not very interested in testing the old one. I want to stay up to date, but I understand you, please don't get tired.

I will try to make tournaments with more games as much as I can.

Best regards,
Fringe

[Sedat Canbaz]

I will try to make tournaments with more games as much as I can.

Best regards,
Fringe

Sounds good... and wishing you good luck

Greetings

[Sedat Canbaz]

The test via Balsa openings is ended as well and as I expected,
Almost identical... just 1 Elo Difference (in favor for Marauders)

Code: Select all

1   Marauders 3       +8/-7/=435 50.11%  225.5/450
2   Stockfish 240322  +7/-8/=435 49.89%  224.5/450

GAMES:
https://mega.nz/file/Sl41naaT#jLY8Lxgiu ... rQR7o79KvA

More Details,
Balsa openings give us a good indication of real strength!
Besides, it's not required running min 5.000, 6.000 + games
As wee see... even 450 games is enough to see the true,
Reality and accurate measuring of real strength, right ?

And once more,
Anyone is free to run at any cond..but I am free too,
To correct, to show all of you what is going on...!!

If still not so clear,
Please do not mix apples with oranges...because
Via weak (handicapped) openings such as running 100-150 games:
- It is simply such as disaster in our great Computer Chess hobby !


Best,
Sedat

[OrgZ]

Thanks again for the wonderful tests Sedat!

and it is important to note that Hash size doesn't always favour both engines. some perform better in low hash and others in larger ones. and this is a bit difficult to do bcz then you would have to run a series of tests to try to find each engines best hash size....