Super Tournament XXXVII

[lkaufman]

Here you are talking about making unbalanced human openings based on game results, something which has already been done.

The first book I have to Graham for this tourney was same idea, only using comp-comp games and with a high eval cull (0.70). This did no better than a random book for decreasing draw rate.

To lower the draw rate, one MUST not cull high evals, one must cull low evals (out of book).
Alternatively, no culling of high nondrawrate from results, culling of lownondrawrate.

Culling low nondrawrate is identical to culling high draw rate as I proposed, so we agree on that. My point is that this can be combined with a maximum Score (with draws counting half), to avoid starting with likely won positions, which I think are the ones Graham means when he says "unfair". Combining the two basically gives you unbalanced books but not unreasonably unbalanced books, which I think most of us favor.

[lkaufman]

For what it's worth, once I stop running tournaments on the i7-4770k towards the end of September, tournaments will then be run on the 5950x with incremental time controls and 6 piece tablebases.

However, there will not be a new rating list.
The 40/15 list will become 40/15 (or 15m10s), although on the 5950x the time control is likely to be 12m10s.

The draw adjudication score will be lowered as well, possibly to 10cp or 15cp for 10 consecutive moves rather than 0.80cp used at present.

10 or 15 cp is pretty reasonable; I would go with 10 myself. Regarding time control, perhaps you meant 12m8s on your hardware to keep the same ratio as 15 + 10s, taking 2/3 for each? The ratio should be consistent I think. Together with a lower-draw book, this should be a big improvement, I look forward to it. Good news!

[Graham Banks]

For what it's worth, once I stop running tournaments on the i7-4770k towards the end of September, tournaments will then be run on the 5950x with incremental time controls and 6 piece tablebases.

However, there will not be a new rating list.
The 40/15 list will become 40/15 (or 15m10s), although on the 5950x the time control is likely to be 12m10s.

The draw adjudication score will be lowered as well, possibly to 10cp or 15cp for 10 consecutive moves rather than 0.80cp used at present.

10 or 15 cp is pretty reasonable; I would go with 10 myself. Regarding time control, perhaps you meant 12m8s on your hardware to keep the same ratio as 15 + 10s, taking 2/3 for each? The ratio should be consistent I think. Together with a lower-draw book, this should be a big improvement, I look forward to it. Good news!

I just liked 10s rather than 8s.
Shouldn't make too much difference?

[lkaufman]

For what it's worth, once I stop running tournaments on the i7-4770k towards the end of September, tournaments will then be run on the 5950x with incremental time controls and 6 piece tablebases.

However, there will not be a new rating list.
The 40/15 list will become 40/15 (or 15m10s), although on the 5950x the time control is likely to be 12m10s.

The draw adjudication score will be lowered as well, possibly to 10cp or 15cp for 10 consecutive moves rather than 0.80cp used at present.

10 or 15 cp is pretty reasonable; I would go with 10 myself. Regarding time control, perhaps you meant 12m8s on your hardware to keep the same ratio as 15 + 10s, taking 2/3 for each? The ratio should be consistent I think. Together with a lower-draw book, this should be a big improvement, I look forward to it. Good news!

I just liked 10s rather than 8s.
Shouldn't make too much difference?

Admittedly not a big difference, but just for consistency it is better to keep a constant base to increment ratio. Some programs might be better than others with a larger or smaller ratio, so having all testers use the same ratio minimizes such random effects. Developers usually use 100 to 1 ratio I believe, which is right in the middle of your 120 to 1 ratio for blitz and the 90 to 1 ratio proposed (at 15' + 10") for Rapid. A 72 to 1 ratio is getting pretty far away from the "standard" 100 to 1. In your specific situation, 12' + 8" will already average longer (for 60 moves) than your current tc on your hardware, so using a larger increment would raise the quality even further above what you have been using, which is not desirable if the two are to be mixed. In principle, if you are now using 40/11, that is 16.5 minutes for 60 moves (a good number to based things on), so base plus inc. should be close to that number. Ten minutes plus six seconds on your hardware, the standard 100 to 1, would be almost perfect.

[Graham Banks]

SUPER TOURNAMENT XXXVII

Intel i7-4770k Quad
ChessGUI
256mb hash each where possible
3-4-5 piece tablebases
Ponder off
LowDraw100.cgb book
40 moves in 16 minutes repeating (adapted for CCRL)
6 cycles 54 rounds

New opening book that is fair, but expected to produce a lowish draw rate.

Standings after Round 36

23.5 - Stockfish 16 64-bit
22.5 - Dragon 3.2 by Komodo 64-bit
18.5 - Ethereal 14.00 64-bit
17.5 - RubiChess 20230410 64-bit
17.5 - Berserk 11.1 64-bit
17.0 - Clover 6.0 64-bit
16.5 - CS Tal 2.00 64-bit
16.5 - Igel 3.5.0 64-bit
16.0 - Revenge 3.0 64-bit
14.5 - Koivisto 9.0 64-bit


Web based link for live viewing (courtesy of Jay - Berserk author).
https://ccrl.live/16092

Alternatively, if you install TLCV (Tom's Live Chess Viewer) on your computer, you can watch the games live move by move. You'll also be able to chat to others following the tournament in the chatroom there.
http://kirill-kryukov.com/chess/discuss ... p?id=42959
Host - GrahamCCRL.dyndns.org Port - 16092

Linux users can use Livius:
https://github.com/kmar/livius

There is also a Livius windows version.
It has live pv boards as a nice addition.
http://www.crabaware.com/livius/

[Graham Banks]

SUPER TOURNAMENT XXXVII

Intel i7-4770k Quad
ChessGUI
256mb hash each where possible
3-4-5 piece tablebases
Ponder off
LowDraw100.cgb book
40 moves in 16 minutes repeating (adapted for CCRL)
6 cycles 54 rounds

New opening book that is fair, but expected to produce a lowish draw rate.

Standings after Round 45

29.5 - Dragon 3.2 by Komodo 64-bit
28.5 - Stockfish 16 64-bit
23.5 - Berserk 11.1 64-bit
23.0 - Ethereal 14.00 64-bit
21.0 - CS Tal 2.00 64-bit
21.0 - Igel 3.5.0 64-bit
21.0 - RubiChess 20230410 64-bit
21.0 - Clover 6.0 64-bit
19.5 - Revenge 3.0 64-bit
17.0 - Koivisto 9.0 64-bit


Web based link for live viewing (courtesy of Jay - Berserk author).
https://ccrl.live/16092

Alternatively, if you install TLCV (Tom's Live Chess Viewer) on your computer, you can watch the games live move by move. You'll also be able to chat to others following the tournament in the chatroom there.
http://kirill-kryukov.com/chess/discuss ... p?id=42959
Host - GrahamCCRL.dyndns.org Port - 16092

Linux users can use Livius:
https://github.com/kmar/livius

There is also a Livius windows version.
It has live pv boards as a nice addition.
http://www.crabaware.com/livius/

[Graham Banks]

Draw rate is currently 79%.

[Graham Banks]

SUPER TOURNAMENT XXXVII

Intel i7-4770k Quad
ChessGUI
256mb hash each where possible
3-4-5 piece tablebases
Ponder off
LowDraw100.cgb book
40 moves in 16 minutes repeating (adapted for CCRL)
6 cycles 54 rounds

New opening book that is fair, but expected to produce a lowish draw rate.

Final Standings

34.5 - Dragon 3.2 by Komodo 64-bit
34.0 - Stockfish 16 64-bit
27.5 - Berserk 11.1 64-bit
27.0 - Ethereal 14.00 64-bit
26.0 - RubiChess 20230410 64-bit
25.5 - CS Tal 2.00 64-bit
25.5 - Igel 3.5.0 64-bit
24.5 - Clover 6.0 64-bit
24.0 - Revenge 3.0 64-bit
21.5 - Koivisto 9.0 64-bit


The complete tournament pgn (zipped) can be downloaded here:
https://kirill-kryukov.com/chess/discus ... p?id=54096

[Graham Banks]

81% draw rate, so we'll try again with a different book.