Guess v1.2.1 rating

[tissatussa]

Here is the ZIPped .pgn of 33 games 5m+3s against other engines, mostly played with White, all from starting position : woodpusher-v1.2.1-33-games-5m+3s.zip

I guess v1.2.1 has about rating 1600.
Today i may let the newer v1.3.0 play against those same engines again, to see how Woodpusher improved. I will post the results here.

[sictransit]

Very interesting! I'll take a look as soon as I find some time.

I've only used Stockfish as a sparring partner. When I clamp the ELO to 2000 they play about equal, but that's on my machine. The major improvement I've made is with respect to move ordering. That greatly improved how deep it is able to go within the allotted time. Previously it has never been able to finish the "mate in 10" test, but now it succeeds in about 15 secs.

I'm also using CuteChess as a GUI, but on Windows. The illegal move issue turned out to be bad timing on my part, but since I made a couple of fixes I've never seen it again. Maybe it behaves differently on Linux, but there was a bug in the code that needed fixing regarding of OS.

Thank you so much for the interest you invest in my little project.

[sictransit]

Right now I'm generating a more interesting opening book, with more options for each game state. As it is now, with the current version, two instances of WP will always play the same game. Additionally, WP will always open with E4.

I'll introduce a bit of randomness.

[tissatussa]

In fact a good opening book is a nice idea, but overall i decided not to use an opening book while playing engine-vs-engine games. An idea might be to create an UCI option to set the max ply of the opening book to reach positions for the engine to start its own evaluations. Such UCI option can be an extended version of the current 'OwnBook' option : 0 is no-book, 5 means 5 ply etc.

The randomness can also be an UCI option : 0 is none.
I have nothing against Woodpusher always playing a same variation (randomness 0), as a chess player it can help me to find better moves to "solve a puzzle" - being that position. I didn't yet find the time to play a game against v1.3.0, testing your "negative time adjustment" .. i will now and post it here later.

[tissatussa]

..and post it here later

[ replay game ]

[Event "engine-vs-engine"]
[Site "Holland @ https://lichess.org/DJcWNDxK "]
[Date "2024.12.01"]
[Round "?"]
[White "Woodpusher v1.3.0"]
[Black "Roelof Berkepeis"]
[Result "1-0"]
[ECO "A04"]
[GameDuration "00:22:43"]
[Opening "Reti Opening"]
[PlyCount "65"]
[TimeControl "1200+3"]

1. Nf3 {+0.28/8 31s} b6 {11s} 2. e3 {+0.62/8 31s} Bb7 {5.3s}
3. d4 {+0.68/7 5.9s} e6 {4.9s} 4. Bd3 {+0.62/7 19s} Nf6 {2.0s}
5. O-O {+0.49/7 19s} Be7 {5.3s} 6. c4 {+0.31/7 30s} Na6 {8.1s}
7. a3 {+0.55/7 8.6s} c5 {3.7s} 8. Nc3 {+0.69/8 34s} Nc7 {13s}
9. e4 {+0.71/8 34s} Ng4 {57s} 10. dxc5 {+1.24/7 8.6s} d6 {63s}
11. Qa4+ {+1.81/8 35s} Qd7 {13s} 12. Qxd7+ {+1.68/8 35s} Kxd7 {3.4s}
13. cxd6 {+1.74/8 35s} Bxd6 {2.5s} 14. h3 {+1.66/7 12s} Ne5 {2.6s}
15. Nxe5+ {+1.44/8 36s} Bxe5 {1.8s} 16. Be3 {+1.39/7 16s} Bxc3 {20s}
17. bxc3 {+1.64/7 10s} Na6 {5.0s} 18. a4 {+1.53/7 13s} Rac8 {5.6s}
19. a5 {+1.52/8 18s} Nc5 {4.1s} 20. Bxc5 {+1.48/9 32s} bxc5 {16s}
21. a6 {+1.79/9 21s} Ba8 {21s} 22. e5 {+1.72/8 33s} h5 {11s}
23. Ra5 {+1.76/7 7.0s} h4 {26s} 24. Rb1 {+1.72/8 20s} Rh5 {20s}
25. f4 {+1.92/10 49s} f6 {2.4s} 26. exf6 {+2.13/10 49s} gxf6 {3.2s}
27. Bg6 {+2.02/9 49s} Rh6 {5.8s} 28. f5 {+2.02/10 48s} e5 {14s}
29. Bf7 {+2.03/9 21s} Ke7 {19s} 30. Be6 {+2.19/10 17s} Rc7 {29s}
31. Rab5 {+2.24/10 48s} Rh8 {35s} 32. Rb8 {+2.23/10 34s} Rd8 {12s}
33. Rxd8 {+5.07/12 53s, Black resigns} 1-0