Stefan Pohl Computer Chess

private website for chessengine-tests

Anti Draw Openings - The future of Computerchess



Download of the Anti Draw Openings V1.5 collection: here


New in V1.5

Drawkiller V5 was built complete new from scratch. Whats new?

1) Drawkiller QR-switch (queen and rook on the center-lines exchange their places, which doubles the number of possible DK openings)
2) Drawkiller human: Drawkiller (classic and QR-switch) positions (as FEN-code) followed by 4,5,6 human opening  moves out of the Megabase 2021 (both players 2300+ Elo). Because the human openings are played out of the normal
chess starting position, not out of one of the 4 Drawkiller positions, a lot of move-lines were illegal and had to be sorted out.
3) No longer balanced Drawkiller openings, but now not one but 7 different unbalanced Drawkiller opening-sets,  with increasing advantage for white (=same concept as in Unbalanced Human Openings and WOMP openings).
Read more in the ReadMe-File in the Drawkiller V5 folder.



This Anti Draw openings-sets collection contains 5 different openings concepts:
Drawkiller V5, No Black Short Castling V2, No Black Castling V2, White One More Pawn V1 and Unbalanced Human Openings V3. Additionally, the download contains the tools, which are needed for the Armageddon- and my Advanced Armageddon-rescoring of results.

NBSC and NBC were remastered, which means, all endpositions were re-evaluated with KomodoDragon 1.0  (15 seconds/ position on a Quadcore PC) and the openings-sets were complete rebuilt. 

Additionally, Drawkiller V5 was built complete new from scratch. 

Read more in the ReadMe-File in the Drawkiller V5 folder.

For UHO (Unbalanced Human Openings) V3 all endpositions of the raw-data of UHO V2 with an eval of at least +0.60 for white were re-evaluated on a 12core AMD Ryzen PC with 30 seconds per endposition, which means, that the eval of the UHO V3 endpositions is based on (around) 10x more calculated nodes than in UHO V2. That gives a search depth around 28-30 plies in UHO V3, which is around 4-5 plies deeper than in UHO V2 (!!!). Additionally, the UHO V3 opening-sets now contain not only one, but 7 UHO-openings sets with increasing advantage for white (from +0.90 up to +1.50). Read more in the ReadMe-file in the UHO-folder.

WOMP = "W"hite "O"ne "M"ore "P"awn Openings

Means, that an opening-line was created, in which the black pawn on a7 is deleted:
1. Na3 a5 2. Nb5 a4 3. Nc3 Nf6 4. Nxa4 Ng8 5. Nc3 Nf6 6. Nb1 Ng8 
Behind these 6 moves, human opening moves (4, 5 and 6 moves deep) out of the Megabase 2021 were added and the endpositions were checked by KomodoDragon 1.0 avx with 10 seconds/move on a AMD Ryzen 12 core machine. There are 7 folders with different WOMP V1 openings-sets, with increasing advantage for white (from +0.90 up to +1.50). Read more in the ReadMe-file in the WOMP-folder:

I believe, this opening collection with 5 different opening concepts is giving the perfect solution for all purposes in computerchess: Unbalanced Human Openings are 100% human chess moves - nothing else. Drawkiller is 100% constructed (classic) and 50% constructed (human), for the best statistical results (take a look at the Drawkiller test results - just mind-blowing!). 
NBSC/NBC openings and WOMP openings are in between: Half constructed and half human (a move-line, which forbids black castlings or deletes the black pawn on a7, is added to human played openings). 
All these openings are classical chess, no chess variant (like Chess960), because all lines start from the classical chesss starting position (except the Drawkiller human lines, they are starting from a Drawkiller postion as FEN-code). So, all these openings can be used by all chessengines and chessGUIs on the planet, because no special, new chess-rules are needed like the castling-rules in Chess960 (Fischer Random Chess).

Here the testing results with long thinking-time and comparsion with 6 classical openings-sets. I think, the results speak  for themselves...But take a look at the results of my new Drawkiller unbalanced openings: Just mind-blowing! 

10'+5'' singlethread, (22 Threads of 24 used). No ponder, no bases (3-5 Syzygy for GUI)
1000 games each testrun. Cutechess-cli. AMD Ryzen 12core CPU.
Stockfish 12 avx2 vs. KomodoDragon 1.0 avx2. Contempt=0.

One run took about 20 hours. The Elo-number is calculated out of the score of Stockfish 12 in the head-to-head vs. KomodoDragon 1.0 by ORDO.


Classical openings-sets (for comparsion):

Balsa 2724:                             1000 (+ 67,=913,- 20), 52.4% +16 Elo, Draws 91.3%
GM_6mvs_balanced:                       1000 (+ 87,=896,- 17), 53.5% +24 Elo, Draws 89.6%
Chad 6mvs:                              1000 (+108,=859,- 33), 53.8% +26 Elo, Draws 85.9%
Hert500:                                1000 (+112,=857,- 31), 54.0% +28 Elo, Draws 85.7%
Feobos c3:                              1000 (+114,=852,- 34), 54.0% +28 Elo, Draws 85.2%
Stockfish 8mvs v3:                      1000 (+137,=817,- 46), 54.5% +32 Elo, Draws 81.7%

Average result of classical openings: 
Score (SF 12 point of view):                                   53.7% +26 Elo, Draws 86.6%

Pohl AntiDraw Openings:                 Average result:        58.6% +60 Elo, Draws 48.1%

UHO_V3_6mvs_+90_+99:                    1000 (+290,=546,-164), 56.3% +44 Elo, Draws 54.6%
UHO_V3_8mvs_+90_+99:                    1000 (+286,=583,-131), 57.8% +54 Elo, Draws 58.3%
UHO_V3_8mvs_big_+80_+109:               1000 (+295,=585,-120), 58.8% +62 Elo, Draws 58.5%


NBSC_Noomen_3mvs_250:                    500 (+154,=242,-104), 55.0% +36 Elo, Draws 48.4%
NBSC_V2_4mvs:                           1000 (+302,=541,-157), 57.3% +52 Elo, Draws 54.1%
NBSC_V2_6mvs:                           1000 (+275,=564,-161), 55.7% +40 Elo, Draws 56.4%

NBC_Noomen_3mvs_250:                     500 (+170,=198,-132), 53.8% +26 Elo, Draws 39.6%
NBC_V2_4mvs:                            1000 (+313,=496,-191), 56.1% +42 Elo, Draws 49.6%
NBC_V2_6mvs:                            1000 (+323,=468,-209), 55.7% +40 Elo, Draws 46.8%


WOMP_4mvs_+090_+099:                    1000 (+352,=490,-158), 59.7% +68 Elo, Draws 49.0%
WOMP_5mvs_+090_+099:                    1000 (+327,=487,-186), 57.0% +50 Elo, Draws 48.7%
WOMP_6mvs_+090_+099:                    1000 (+304,=523,-173), 56.5% +46 Elo, Draws 52.3%
WOMP_6mvs_big_+080_+109:                1000 (+306,=518,-176), 56.5% +46 Elo, Draws 51.8% 


Drawkiller_V5_classic_+0.90_+0.99_2mvs: 1000 (+440,=370,-190), 62.5% +90 Elo, Draws 37.0%
Drawkiller_V5_classic_+0.90_+0.99_3mvs: 1000 (+437,=392,-171), 63.3% +96 Elo, Draws 39.2%
Drawkiller_V5_classic_+0.90_+0.99_4mvs: 1000 (+434,=393,-173), 63.0% +94 Elo, Draws 39.3%
Drawkiller_V5_classic_+0.80_+1.09_4mvs: 1000 (+401,=442,-157), 62.2% +88 Elo, Draws 44.2%

Drawkiller_V5_human_+0.90_+0.99_4mvs:   1000 (+392,=441,-167), 61.3% +80 Elo, Draws 44.1%
Drawkiller_V5_human_+0.90_+0.99_5mvs:   1000 (+400,=426,-174), 61.3% +80 Elo, Draws 42.6%
Drawkiller_V5_human_+0.90_+0.99_6mvs:   1000 (+368,=472,-160), 60.4% +74 Elo, Draws 47.2%
Drawkiller_V5_human_+0.80_+1.09_6mvs:   1000 (+358,=491,-151), 60.4% +74 Elo, Draws 49.1%


Andreas Strangmueller (FGRL) did 3 interesting testruns on a very powerful machine,
using my Drawkiller V4 openings and a 4 moves classical openings-set for comparsion:


CPU:          Dual Intel Xeon E5-2673 v4, 40 cores (80 threads)
OS:           Windows 10 64-Bit
Tool:         Cutechess-Cli, no draw and resign rules
Engines:      Stockfish 13 x64 BMI2, default settings
Hash-Table:   2048 MB
Time control: 60 seconds + 0.6 seconds
Tablebases:   None
Ponder:       Off
Each opening: Repeated with changing colors

Stockfish 13 x64 BMI2 32 threads vs. Stockfish 13 x64 BMI2 16 threads, 1000 games

Score always Stockfish 13 with 32 Threads point of view


4 moves classical openings   : 1000 (+67/=923/-10),   52.9% +20 Elo, Draws 92.3%
Drawkiller_balanced_V4_2mvs  : 1000 (+125/=861/-14),  55.6% +39 Elo, Draws 86.1%
Drawkiller_unbalanced_V4_2mvs: 1000 (+333/=517/-150), 59.2% +64 Elo, Draws 51.7%


Advanced Armageddon rescoring of the Drawkiller unbalanced result:
Drawkiller_unbalanced_V4_2mvs: 1007 (+683,=  0,-324), 67.8% +131 Elo, Draws 0%


Conclusions on this testruns, done by an independent tester:
Mention, that the testsetup of playing SF 13 vs. another SF 13 with less speed on a pretty strong hardware is a very, very drawish setup: Using the classical openings with 4 moves,  the draw-rate is enormous: 92.3%. My Drawkiller balanced openings gave a really nice Elo-spreading: nearly doubled (+39 Elo) compared to the classical openings (+20 Elo), but the draw-rate was lower but still pretty high (86.1%).
But, in the final testrun, using my brand new concept of unbalanced Drawkiller openings, the result was just mind-blowing: A more than 3x bigger Elo-spreading (+64 Elo) compared to the classical openings (+20) Elo and a draw-rate of only 51.7% (!!!). Incredible! And much lower (=better), than I expected. This is really a breakthrough in a new dimension and will prevent the draw-death of high-end computerchess at least for the next decade.

Results published by Andreas Strangmueller on talkchess, look here


All ideas of the Pohl AntiDraw Openings by Stefan Pohl (except the NBSC concept: Idea by Larry Kaufman (Komodo team)).
All work of the Pohl AntiDraw Openings done by Stefan Pohl (except the generation of the pawn moves added to the 
Drawkiller lines: Done by Hauke Lutz)


(C) 2021 Stefan Pohl (SPCC)(