Database snapshot

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Ferdy
Posts: 4070
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Database snapshot

Post by Ferdy » Wed Jul 31, 2019 12:40 am

Redownload the exe, it now handles the missing winrate, it will use 0.0 and show it in result table.

Code: Select all

enter fen or epd or q to quit? r1bqkb1r/pppp1ppp/2n2n2/4p1N1/2B1P3/8/PPPP1PPP/RNBQK2R b KQkq -
(1) Move Stats:
Move    Score  Winrate
d5      -5     49.62
Bc5     -175   37.04
Nxe4    -248   32.05
Qe7     -290   29.34
Nd5     -365   24.86
b5      -375   24.30
d6      -418   21.98
g6      -453   20.22
Na5     -461   19.83
Ba3     -473   19.26
a5      -487   18.61
Bb4     -509   17.62
Bd6     -512   17.49
a6      -526   16.88
b6      -589   14.37
Rb8     -599   14.00
Nd4     -612   13.53
Nb4     -612   13.53
Nb8     -636   12.71
Rg8     -651   12.21
Ng8     -721   10.11
Ke7     -725   10.00
Be7     -730   9.87
h6      -780   8.60
Ng4     -796   8.23
h5      -805   8.02
Nh5     -885   6.41
Ne7     -29998  0.00

noobpwnftw
Posts: 338
Joined: Sun Nov 08, 2015 10:10 pm

Re: Database snapshot

Post by noobpwnftw » Wed Jul 31, 2019 12:43 am

Ferdy wrote:
Tue Jul 30, 2019 2:45 pm
See there is no 'winrate'. Its winrate should have been 0.0 assuming winrate has a range [0, 100], %.
There is no winrate field for mate scores or exact TB scores(not yet available). It is by design.

User avatar
Ovyron
Posts: 2488
Joined: Tue Jul 03, 2007 2:30 am

Re: Database snapshot

Post by Ovyron » Wed Jul 31, 2019 1:06 am

Ferdy wrote:
Wed Jul 31, 2019 12:40 am
Redownload the exe, it now handles the missing winrate, it will use 0.0 and show it in result table.
Thanks, and noobpwnftw for the db. Apparently my problem was that Chess Openings Wizard's Copy FEN actually produces an EPD :shock: but at least a bug was squashed :)
Great spirits have always encountered violent opposition from mediocre minds.

Ferdy
Posts: 4070
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Database snapshot

Post by Ferdy » Wed Jul 31, 2019 1:39 am

noobpwnftw wrote:
Wed Jul 31, 2019 12:43 am
Ferdy wrote:
Tue Jul 30, 2019 2:45 pm
See there is no 'winrate'. Its winrate should have been 0.0 assuming winrate has a range [0, 100], %.
There is no winrate field for mate scores or exact TB scores(not yet available). It is by design.
All right then I would suggest you add {... "winrate": "not available"} to be consistent with the query output.

Later in my script, I will just calculate a winrate based on the score using logistic function.

User avatar
Ovyron
Posts: 2488
Joined: Tue Jul 03, 2007 2:30 am

Re: Database snapshot

Post by Ovyron » Wed Jul 31, 2019 2:13 am

Okay, so I found a problem with the database at the conceptual level. The point of having a back-propagated score and ranking moves based on those scores, is that, no matter what, if you follow the top ranked move you get an out-of-book position for the datababase that is guaranteed to hold a database score that is better than anything else (otherwise, that would have been ranked even higher.)

I was curious about how the database plays against the Italian. For white I used highest frequency of moves played instead of optimal scoring for moves (lest 4.d3!! is optimal and a completely different line is tested.) Black plays top ranked db move.

This is a critical position (database starts after 3.Bc4):

1. e4 e5 2. Nf3 Nc6 3. Bc4!? Nf6 4. Ng5!? d5 5. exd5 Na5 6. Bb5+ c6 7. dxc6 bxc6 8. Be2 h6 9. Nh3 Bc5 10. d3



Here the query reads:

Code: Select all

Move    Score  Winrate
Qc7     6      50.45
O-O     5      50.38
Nb7     2      50.15
Bf5     -3     49.77
Equivalent to saying "Qc7 is best with a score of -0.06, playing other moves leads to a position that is -0.05 or worse". Out of database position is expected to be good for black.

Moves continue (black still playing top ranked db moves):

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Na5 6. Bb5+ c6 7. dxc6 bxc6 8. Be2 h6 9. Nh3 Bc5 10. d3 Qc7 11. O-O O-O 12. Nc3



Query:

Code: Select all

Move    Score  Winrate
Nb7     3      50.23
This is equivalent to a -0.03 score, which is worse than the score of 10...O-O (-0.05) of the critical position. Basically, the database now regrets playing 10...Qc7. Ideally, the backsolved score for Qc7 should be lower than the score of 10...O-O, because the database already knows white can play this, making 10...O-O more desirable than 10...Qc7.

Another move:

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Na5 6. Bb5+ c6 7. dxc6 bxc6 8. Be2 h6 9. Nh3 Bc5 10. d3 Qc7 11. O-O O-O 12. Nc3 Nb7 13. Kh1



Query:

Code: Select all

Move    Score  Winrate
Rd8     0      50.00
Bxh3    -1     49.92
This is equivalent to a score of 0.00. 10.Qc7 would now have a backsolved score of 0, making both 10...O-O and 10...Nb7 (-0.02) more appealing that the top ranked Qc7 (which would now be ranked third on the critical position.)

---------------- END OF POST ----------------

That's all I wanted to say, what follow is my findings on this line for the curious. It ends here:

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Na5 6. Bb5+ c6 7. dxc6 bxc6 8. Be2 h6 9. Nh3 Bc5 10. d3 Qc7 11. O-O O-O 12. Nc3 Nb7 13. Kh1

r1b2rk1/pnq2pp1/2p2n1p/2b1p3/8/2NP3N/PPP1BPPP/R1BQ1R1K b - -

Full query:

Code: Select all

Move    Score  Winrate
Rd8     0      50.00
Bxh3    -1     49.92

(2) PV info of top move:
0/3 pv Rd8 Ng1 Bf5
I have 14.Qe1 as... "busting" this black line. I find it curious that the db expects Ng1, since the engine drops it at depth 17:

Depth 31 MultiPV=4

16/23 0:01 +0.05 14.Ng1 Bf5 15.Bd2 Rab8 16.Bf3 Bb6 17.Nge2 Nc5 18.Ng3 Bg6 19.h4 Ne6 20.h5 Bh7 21.Rb1 Nd4 22.Nge4 Nxe4 23.Bxe4 (5.238.962) 2943
16/19 0:01 +0.02 14.Qe1 Bd4 (5.238.962) 2943
16/21 0:01 0.00 14.Bd2 Bd6 15.Ng1 Rb8 16.Qe1 Nc5 17.b3 Ne6 18.Ne4 Nxe4 19.dxe4 Qe7 20.Nf3 Nc5 21.Ba5 Rd7 22.Bd3 Nxd3 23.cxd3 Qf6 (5.238.962) 2943
16/21 0:01 -0.26 14.Bf3 Nd6 15.Qe1 Re8 16.Ng1 Bf5 17.Nge2 e4 18.dxe4 Ndxe4 19.Bf4 Qb7 20.Na4 Bxf2 21.Rxf2 Nxf2+ 22.Qxf2 Bxc2 23.Nc5 Qxb2 24.Rg1 Bf5 (5.238.962) 2943
-
17/24 0:03 +0.12 14.Qe1 Bd6 15.f4 Bxh3 16.gxh3 exf4 17.Qh4 Nd5 18.Nxd5 cxd5 19.Bxf4 Qxc2 20.Bh5 Rd7 21.Qg3 Bxf4 22.Qxf4 g6 23.Rab1 Qxd3 24.Qxh6 (9.165.423) 2973
17/25 0:03 -0.03 14.Ng1 e4 15.Bd2 Bf5 16.dxe4 Nxe4 17.Nxe4 Bxe4 18.Bf3 Bg6 19.Qc1 Bd6 20.Ne2 Nc5 21.Ng3 Be5 22.Re1 Rab8 (9.165.423) 2973
17/23 0:03 -0.03 14.Bd2 Bd6 15.Bf3 Nc5 16.b4 Ne6 17.Ne4 Nxe4 18.Bxe4 Nd4 19.c3 f5 20.f4 Bxb4 21.cxd4 Bxd2 22.Bxc6 Qxc6 23.Qxd2 exf4 24.Qxf4 (9.165.423) 2973
17/27 0:03 -0.18 14.f4 Bd4 15.Qe1 Nd6 16.Bf3 Re8 17.Ne2 Bxh3 18.gxh3 Nf5 19.c3 Be3 20.fxe5 Rxe5 21.Ng3 Bxc1 22.Qxc1 Nxg3+ 23.hxg3 (9.165.423) 2973
-
18/32 0:04 +0.25 14.Qe1 Bf5 15.f4 e4 16.Nxe4 Nxe4 17.dxe4 Bxe4 18.b3 Bf5 19.Bb2 Bd4 20.c3 Bf6 21.Ng1 Be4 22.Nf3 Nc5 23.Nd4 Bd3 24.h3 Bxe2 (13.351.876) 2946
18/26 0:04 +0.15 14.Bd2 Rb8 15.f4 Bd4 16.Ng1 Nc5 17.fxe5 Bxe5 18.Nf3 Bd6 19.b3 Ne6 20.Ne4 Nxe4 21.dxe4 Bb4 22.Qe1 Bxd2 23.Nxd2 (13.351.876) 2946
18/22 0:04 +0.08 14.Ng1 Bf5 15.Bd2 Bd6 16.Bf3 Nc5 17.Qe1 Rdb8 18.b3 a5 19.Nge2 e4 20.dxe4 Bxe4 21.Nd4 Be5 22.Nxe4 Ncxe4 23.Bxe4 (13.351.876) 2946
18/22 0:04 -0.08 14.f4 Bd4 15.Ng1 e4 16.Qe1 exd3 17.cxd3 Nc5 18.Nf3 Bf5 19.Ne4 Ncxe4 20.dxe4 Nxe4 21.Nxd4 Rxd4 22.Bf3 Nf6 23.Be3 Rdd8 (13.351.876) 2946
-
19/30 0:06 +0.29 14.Qe1 Bd4 15.Ng1 Nd6 16.Bf3 Rb8 17.Nge2 Nf5 18.Nxd4 Nxd4 19.Bd1 Re8 20.b3 Bf5 21.Ne4 Bxe4 22.dxe4 Ne6 23.Bb2 (19.874.429) 2927
19/28 0:06 +0.17 14.f4 e4 15.Qe1 exd3 16.Bxd3 Bd4 17.Ng1 Nc5 18.Be2 Bf5 19.Nf3 Ne6 20.Bd3 Bxc3 21.Qxc3 Bxd3 22.cxd3 Nd5 (19.874.429) 2927
19/26 0:06 +0.15 14.Bd2 Bf5 15.f4 e4 16.Nxe4 Nxe4 17.dxe4 Bxe4 18.Bd3 Bxd3 19.cxd3 Rxd3 20.Qc2 Rd5 21.Bc3 Rad8 22.f5 Bd4 23.Bxd4 Rxd4 24.f6 gxf6 25.Rxf6 (19.874.429) 2927
19/26 0:06 -0.04 14.Ng1 Bf5 15.Bd2 a5 16.Bf3 Be7 17.Nge2 Nc5 18.Ng3 Bg6 19.h3 Rab8 20.b3 Ne6 21.Nce4 Nxe4 22.Bxe4 Bxe4 23.Nxe4 (19.874.429) 2927
-
20/27 0:12 +0.37 14.f4 Bxh3 15.gxh3 Bd4 16.Qe1 Nc5 17.f5 e4 18.Nxe4 Nfxe4 19.dxe4 Nxe4 20.Bd3 Nf6 21.Qe2 Be5 22.Rb1 Rd4 23.Be3 Rh4 24.Qg2 (37.740.194) 2929
20/27 0:12 +0.15 14.Bd2 Nd6 15.Ng1 Bb6 16.Qe1 Bf5 17.a4 e4 18.Bf4 Qb7 19.Be5 Qe7 20.f4 a5 21.Qh4 Bh7 22.Bxf6 Qxf6 23.Qxf6 gxf6 24.dxe4 Nxe4 (37.740.194) 2929
20/29 0:12 +0.06 14.Qe1 Nd6 15.Na4 Bd4 16.c3 Bb6 17.Ng1 Re8 18.b4 Qe7 19.Nxb6 axb6 20.a4 c5 21.Be3 Nf5 22.a5 Bb7 23.Nf3 Nxe3 24.fxe3 bxa5 25.Rxa5 cxb4 (37.740.194) 2929
20/30 0:12 -0.05 14.Ng1 Bf5 15.Bd2 Rab8 16.Qc1 Re8 17.Bf3 Bg6 18.h3 Bd6 19.Nge2 Nc5 20.Ng3 e4 21.Ngxe4 Nfxe4 22.Nxe4 Bxe4 23.dxe4 Nxe4 (37.740.194) 2929
-
21/30 0:16 +0.30 14.f4 e4 15.Nf2 Nd6 16.Qe1 Bf5 17.Ncxe4 Ndxe4 18.Nxe4 Nxe4 19.dxe4 Bxe4 20.Bd2 Qd6 21.Bc3 Qg6 22.Bf3 Bxc2 23.Rc1 Bd3 24.Be5 Bxf1 25.Qxf1 Bd4 (48.700.495) 2927
21/28 0:16 +0.17 14.Bd2 Nd6 15.Ng1 Bb6 16.Qc1 Bb7 17.Bf3 Re8 18.Na4 e4 19.Nxb6 axb6 20.dxe4 Ndxe4 21.Bf4 Qe7 22.c4 c5 23.Ne2 Bc6 24.b3 Rad8 (48.700.495) 2927
21/29 0:16 +0.09 14.Qe1 Nd6 15.Na4 Bd4 16.c3 Bb6 17.Ng1 Re8 18.b4 Qe7 19.Nxb6 axb6 20.a4 Qc7 21.Bb2 c5 22.c4 Nf5 23.Bf3 Bb7 24.a5 bxa5 25.Rxa5 Rxa5 26.bxa5 Nd4 (48.700.495) 2927
21/30 0:16 -0.06 14.Ng1 Bf5 15.Bd2 Bd6 16.Qc1 Nc5 17.f4 e4 18.dxe4 Ncxe4 19.Nxe4 Bxe4 20.Re1 Nd5 21.Bc4 Re8 22.Ne2 Rab8 23.b3 Bg6 (48.700.495) 2927
-
22/30 0:21 +0.30 14.Bd2 Rb8 15.Rb1 Bd4 16.Ng1 Bf5 17.Nf3 Nc5 18.b4 Ne6 19.Nxd4 Rxd4 20.h3 Qd6 21.a4 Rbxb4 22.Rxb4 Rxb4 23.Ne4 Bxe4 24.Bxb4 Qxb4 25.dxe4 Nxe4 (63.232.784) 2933
22/31 0:21 +0.21 14.Qe1 Nd6 15.Na4 Bd4 16.c3 Bb6 17.Ng1 Re8 18.b4 Qe7 19.Nxb6 axb6 20.a4 c5 21.Be3 Nf5 22.a5 Bb7 23.Bf3 cxb4 24.Bxb6 Bxf3 (63.232.784) 2933
22/31 0:21 +0.20 14.f4 e4 15.Nf2 Nd6 16.Qe1 exd3 17.Nxd3 Bd4 18.Bf3 Re8 19.Ne2 Nd5 20.Qd1 Bb6 21.a4 Nc4 22.Bxd5 cxd5 23.Ng3 Bd7 24.f5 Ne3 25.Bxe3 Bxe3 26.f6 gxf6 27.Rxf6 (63.232.784) 2933
22/28 0:21 0.00 14.Ng1 Bf5 15.Qe1 Re8 16.Be3 Bg6 17.Rd1 Bd4 18.Nf3 Rad8 19.Nd2 Nd6 20.Nb3 Bb6 21.Nd2 (63.232.784) 2933
-
23/32 0:33 +0.20 14.Bd2 Nd6 (99.429.981) 2934
23/33 0:33 +0.15 14.Qe1 Nd6 15.Na4 Bd4 16.c3 Bb6 17.Ng1 Nf5 18.Nxb6 axb6 19.f4 Nd5 20.Qf2 exf4 21.Bxf4 Qxf4 22.Qxf4 Nxf4 23.Rxf4 Be6 24.a4 g5 25.Rf2 Nd6 (99.429.981) 2934
23/36 0:33 +0.08 14.Ng1 Bf5 15.Qe1 Bb6 16.Nf3 Re8 17.Nd2 Nd6 18.Bf3 Bg6 19.b3 Rad8 20.Bb2 Bd4 21.Qc1 e4 22.Ncxe4 Ndxe4 23.Nxe4 Bxe4 24.dxe4 Nxe4 25.Bxd4 Rxd4 (99.429.981) 2934
23/31 0:33 -0.01 14.f4 Bd4 15.Ng1 e4 16.Qe1 exd3 17.cxd3 Bf5 18.Nf3 Nc5 19.Ne4 Ncxe4 20.dxe4 Nxe4 21.Nxd4 Rxd4 22.Be3 Rd6 23.Rd1 Rxd1 24.Qxd1 Rd8 25.Qa4 Re8 26.Qxa7 Qxa7 27.Bxa7 (99.429.981) 2934
-
24/32 0:52 +0.01 14.Qe1 Nd6 (153.039.785) 2941
24/32 0:52 0.00 14.Ng1 Bf5 15.Bd2 Rab8 16.Qc1 Re8 17.Rb1 Bd6 18.f4 e4 19.dxe4 Nxe4 20.Nxe4 Bxe4 21.Bf3 a5 22.Re1 Bd5 23.b3 Nc5 24.c4 Bxf3 25.Nxf3 Nd3 26.Rxe8+ Rxe8 27.Qc3 (153.039.785) 2941
24/32 0:52 -0.12 14.Bd2 Nd6 (153.039.785) 2941
24/34 0:52 -0.16 14.Bf3 Nd6 15.Na4 Bd4 16.c3 Bb6 17.d4 exd4 18.Nxb6 axb6 19.Qxd4 Bxh3 20.gxh3 Nde4 21.Qe3 Re8 22.Bg2 Rad8 23.Qf4 Qxf4 24.Bxf4 Nd5 25.Bg3 Nxg3+ 26.hxg3 Re2 27.c4 (153.039.785) 2941
-
25/40 1:06 +0.38 14.f4 e4 15.Qe1 exd3 16.Bxd3 Bxh3 17.gxh3 Bd4 18.Ne2 Nc5 19.Bc4 Nce4 20.Nxd4 Rxd4 21.Bd3 Re8 22.Be3 Rxd3 23.cxd3 Nd6 24.Qc1 Nf5 (196.387.287) 2952
25/34 1:06 +0.21 14.Qe1 Bf5 15.f4 e4 16.Nxe4 Nxe4 17.dxe4 Bxe4 18.b3 Bd4 19.c3 Bf6 20.Nf2 Bh4 21.Be3 Nd6 22.Qc1 Bxf2 23.Bxf2 Re8 24.Bg4 Rad8 25.Rd1 Nb5 26.c4 Rxd1+ 27.Bxd1 (196.387.287) 2952
25/32 1:06 +0.19 14.Ng1 Bf5 15.Qe1 Re8 16.Bf3 Bg6 17.Nge2 Rad8 18.Ng3 Bd4 19.Bd2 Rb8 20.Rb1 Nc5 21.b3 Nfd7 22.Nce2 Ne6 23.Be4 Bxe4 24.dxe4 Bb6 25.Bc3 Rbd8 26.Rd1 (196.387.287) 2952
25/30 1:06 +0.06 14.Bd2 Nd6 15.Na4 Bb6 16.a3 Bxh3 17.gxh3 Nf5 18.Nxb6 Qxb6 19.Bc3 Re8 20.Rb1 a5 21.Bd2 Nd5 22.Bg4 Nd4 23.Bh5 Nf4 24.Bxf4 exf4 25.Qg4 (196.387.287) 2952
-
26/34 1:41 +0.41 14.Qe1 Re8 15.Ne4 Nxe4 16.dxe4 Bxh3 17.gxh3 Nd6 18.Bd3 Qd7 19.Qe2 Qxh3 20.f4 Rad8 21.f5 Qh4 22.Bd2 Kh7 23.c3 Bb6 24.a4 Nxe4 25.Bxe4 Rxd2 26.f6+ Kg8 27.Qxd2 (300.968.331) 2955
26/39 1:41 +0.37 14.Bd2 Bxh3 15.gxh3 Bd4 16.Rb1 Nc5 17.Bf3 Rab8 18.b4 Bxc3 19.Bxc3 Ne6 20.Re1 Nd4 21.Bg2 c5 22.b5 c4 23.Qd2 cxd3 24.cxd3 Rdc8 25.Ba5 Qd6 26.Bb4 Qb6 27.Rxe5 (300.968.331) 2955
26/37 1:41 +0.28 14.Ng1 Bf5 15.Qe1 Re8 16.Bf3 Nd6 17.Ne4 Nfxe4 18.Bxe4 Bg4 19.Be3 Bb6 20.Qc3 Bd7 21.Ne2 Rab8 22.Bxb6 Qxb6 23.Rab1 Qa6 24.Bf3 Qxa2 25.b3 e4 26.dxe4 Nxe4 (300.968.331) 2955
26/40 1:41 +0.23 14.f4 e4 15.Qe1 Bf5 16.Nf2 Bxf2 17.Qxf2 exd3 18.cxd3 Bxd3 19.Be3 c5 20.Rae1 Bxe2 21.Rxe2 Qc6 22.f5 Re8 23.Rfe1 Rad8 24.Bf4 Nd6 25.Rxe8+ Rxe8 26.Rxe8+ Ndxe8 27.Bg3 (300.968.331) 2955
-
27/40 2:40 +0.38 14.Bd2 Bxh3 15.gxh3 Nd6 16.Bf3 Rab8 17.Rb1 Bd4 18.Bg2 Nf5 19.Ne4 Nxe4 20.Bxe4 Nd6 21.Bg2 Rxb2 22.Rxb2 Bxb2 23.Qe2 Re8 24.Rb1 Bd4 25.c3 Bc5 26.Rg1 Kh7 27.h4 (475.379.422) 2960
27/37 2:40 +0.28 14.Qe1 Bf5 15.Bf3 Be7 16.Ne4 a5 17.Ng3 Be6 18.Qe2 Nc5 19.a4 Ncd7 20.b3 Bb4 21.Bb2 Nd5 22.Bg4 Bc3 23.Bxc3 Nxc3 24.Qf3 Nf8 25.Ne4 Nxe4 26.dxe4 Bxg4 27.Qxg4 (475.379.422) 2960
27/38 2:40 +0.17 14.Ng1 e4 15.Bd2 Bf5 16.dxe4 Nxe4 17.Nxe4 Bxe4 18.Nf3 Rd5 19.Bd3 Bxd3 20.cxd3 Rad8 21.Qa4 Bb6 22.Bf4 Qc8 23.b4 c5 24.Be3 cxb4 25.Bxb6 axb6 26.d4 Nd6 27.Rac1 (475.379.422) 2960
27/42 2:40 +0.13 14.f4 e4 15.Qe1 Bxh3 16.gxh3 exd3 17.Bxd3 Bd4 18.Qg3 Nc5 19.Bc4 Rab8 20.Qg2 Qe7 21.Bb3 Nd5 22.Nxd5 cxd5 23.Bxd5 Bxb2 24.Bxb2 Rxb2 25.Rg1 Qf6 26.Rad1 Re8 27.Rde1 (475.379.422) 2960
-
28/40 4:22 +0.40 14.Qe1 Bf5 15.Bf3 a5 16.Ng1 Re8 17.Ne4 Nd5 18.Ng3 Bd7 19.Bd2 Nd6 20.N1e2 Qb6 21.Ne4 Nxe4 22.dxe4 Nf6 23.Bc3 Qc7 24.a3 Be6 25.h3 Bb6 26.Rd1 a4 27.Bd2 (780.403.420) 2967
28/41 4:22 +0.24 14.Bd2 Nd6 15.Ng1 a5 16.Na4 Ba7 17.b3 Bb8 18.Be3 Nd5 19.Bc5 e4 20.g3 Ba7 21.Bxa7 Qxa7 22.d4 Be6 23.c3 Qc7 24.Qc1 Bf5 25.c4 Nf6 26.Qc3 Ng4 27.h3 (780.403.420) 2967
28/41 4:22 +0.24 14.Ng1 e4 15.Bd2 Bf5 16.dxe4 Nxe4 17.Nxe4 Bxe4 18.Nf3 Rd5 19.Bd3 Bxd3 20.cxd3 Rxd3 21.Qc2 Rd5 22.Bc3 Bb6 23.b4 Nd6 24.Rfe1 a5 25.Rad1 axb4 26.Bxb4 Rd8 27.Bc3 (780.403.420) 2967
28/38 4:22 +0.15 14.f4 e4 15.Qe1 exd3 16.Bxd3 Bd4 17.Ne2 Bb6 18.Nf2 Re8 19.b3 Nc5 20.Bb2 Nd5 21.Bc4 Bf5 22.Be5 Qd7 23.Qc1 Rad8 24.Ng3 Bh7 25.c3 Bc7 26.Re1 Bxe5 27.fxe5 (780.403.420) 2967
-
29/42 5:45 +0.46 14.Qe1 Bf5 15.Bf3 Rab8 16.Ng1 Bg6 17.Be4 Nxe4 18.Nxe4 Bd4 19.f4 Nd6 20.Nxd6 Qxd6 21.Rb1 Qe6 22.Ne2 Qxa2 23.Bd2 Bxb2 24.f5 Bh7 25.f6 Qa4 26.fxg7 Qxc2 27.Bxh6 (1.025.381.035) 2971
29/42 5:45 +0.33 14.Ng1 Bb6 15.Qe1 Nd6 16.f4 Nf5 17.fxe5 Qxe5 18.Bf4 Qd4 19.a4 a5 20.Qc1 Re8 21.Nf3 Qc5 22.Ne4 Nxe4 23.dxe4 Rxe4 24.Bd3 Re8 25.Re1 Rxe1+ 26.Qxe1 Ne7 27.Ne5 (1.025.381.035) 2971
29/47 5:45 +0.29 14.f4 e4 15.Qe1 Bf5 16.Nxe4 Nxe4 17.dxe4 Bxe4 18.f5 Re8 19.Bf4 Qe7 20.Bd3 Bxd3 21.cxd3 Qxe1 22.Raxe1 Bd4 23.Bc1 Rxe1 24.Rxe1 Nc5 25.Nf4 Rd8 26.g4 Bf6 27.Rd1 (1.025.381.035) 2971
29/40 5:45 +0.21 14.Bd2 Nd6 15.Ng1 Bb6 16.b3 e4 17.Bf4 Qe7 18.Na4 Bf5 19.Nxb6 axb6 20.Re1 Nd5 21.Bg3 Qf6 22.Qd2 Bg6 23.a4 Nf5 24.dxe4 Nc3 25.e5 Rxd2 26.exf6 Nxg3+ 27.fxg3 (1.025.381.035) 2971
-
30/40 8:05 +0.38 14.Qe1 Bf5 15.f4 e4 16.Nxe4 Nxe4 17.dxe4 Bxe4 18.f5 Re8 19.Bf4 Qe7 20.Bd3 Bxd3 21.cxd3 Qxe1 22.Raxe1 Bd4 23.Bc1 Nc5 24.Nf4 a5 25.g4 Red8 26.Kg2 a4 27.Be3 (1.440.695.697) 2970
30/45 8:05 +0.25 14.f4 (1.440.695.697) 2970
30/35 8:05 +0.12 14.Bd2 Nd6 (1.440.695.697) 2970
30/41 8:05 0.00 14.Ng1 e4 15.Bd2 Bf5 16.dxe4 Nxe4 17.Nxe4 Bxe4 18.Qc1 Bd6 19.Bf3 Bg6 20.Be3 Nc5 21.Rd1 Rab8 22.Ne2 Ne4 23.Ng3 Ng5 24.Bg4 Ne4 25.Bf3 (1.440.695.697) 2970
-
31/44 11:11 +0.39 14.Qe1 Bf5 15.Bf3 Re8 16.Ne4 Nxe4 17.Bxe4 Nd6 18.Ng1 Bd7 19.Ne2 a5 20.h3 a4 21.Ng3 Nxe4 22.dxe4 Be6 23.Be3 Qb6 24.b3 Red8 25.Qc3 Bxe3 26.fxe3 Qc7 27.Ne2 (2.003.435.795) 2981
31/45 11:11 +0.25 14.f4 e4 15.Qe1 Bxh3 16.gxh3 exd3 17.Bxd3 Bd4 18.Qe2 Nc5 19.Bc4 Rab8 20.f5 Qe5 21.Qg2 Bxc3 22.bxc3 Qe4 23.Qxe4 Ncxe4 24.Rg1 Kf8 25.Be3 Rb7 26.Rae1 Re7 27.Bd3 (2.003.435.795) 2981
31/48 11:11 +0.22 14.Ng1 e4 15.Bd2 Bf5 16.dxe4 Nxe4 17.Nxe4 Bxe4 18.Qc1 a5 19.Bf4 Bd6 20.Bxd6 Nxd6 21.Nf3 Qb6 22.Bd3 Bxf3 23.gxf3 Qd4 24.Rg1 Kh8 25.Qe3 Qxb2 26.f4 Qf6 27.Rab1 (2.003.435.795) 2981
31/39 11:11 +0.18 14.Bd2 Nd6 15.Na4 Bb6 16.Nxb6 axb6 17.a4 c5 18.b3 Bxh3 19.gxh3 Nf5 20.Bf3 e4 21.Bg2 Nh4 22.Rg1 exd3 23.Bxh6 Rac8 24.Bg5 Nxg2 25.Rxg2 dxc2 26.Qxc2 Qc6 27.Rag1 (2.003.435.795) 2981
Great spirits have always encountered violent opposition from mediocre minds.

Ferdy
Posts: 4070
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Database snapshot

Post by Ferdy » Wed Jul 31, 2019 2:44 am

Ferdy wrote:
Wed Jul 31, 2019 1:39 am
noobpwnftw wrote:
Wed Jul 31, 2019 12:43 am
Ferdy wrote:
Tue Jul 30, 2019 2:45 pm
See there is no 'winrate'. Its winrate should have been 0.0 assuming winrate has a range [0, 100], %.
There is no winrate field for mate scores or exact TB scores(not yet available). It is by design.
All right then I would suggest you add {... "winrate": "not available"} to be consistent with the query output.

Later in my script, I will just calculate a winrate based on the score using logistic function.
Output with logistic.

Code: Select all

enter fen or epd or q to quit? rnbqkbnr/pp2pppp/2p5/3p4/3PP3/8/PPP2PPP/RNBQKBNR w KQkq d6 0 3
(1) Move Stats:
Move    Score   Winrate  logis
e5      19      51.44    51.42
Nd2     16      51.21    51.20
Nc3     16      51.21    51.20
exd5    14      51.06    51.05
f3      2       50.15    50.15
Bd3     0       50.00    50.00
Qe2     0       50.00    50.00
Qd3     -1      49.92    49.93
Qf3     -1      49.92    49.93
c3      -16     48.79    48.80
Ne2     -22     48.33    48.35
Be3     -56     45.77    45.81
c4      -58     45.62    45.66
g3      -63     45.24    45.29
a4      -75     44.34    44.40
h3      -75     44.34    44.40
Nh3     -76     44.27    44.32
Nf3     -77     44.19    44.25
Bg5     -81     43.89    43.95
g4      -85     43.60    43.66
b3      -91     43.15    43.22
Bf4     -92     43.08    43.14
h4      -96     42.78    42.85
Be2     -98     42.63    42.70
a3      -100    42.48    42.56
Bd2     -107    41.96    42.04
Qd2     -107    41.96    42.04
Na3     -108    41.89    41.97
b4      -149    38.90    39.01
Qh5     -150    38.83    38.94
f4      -159    38.18    38.30
Ke2     -193    35.78    35.92
Kd2     -212    34.47    34.62
Bc4     -375    24.30    24.51
Bb5     -440    20.86    21.08
Ba6     -482    18.84    19.06
Bh6     -506    17.75    17.98
Qg4     -1343   1.68     1.75

It seems close when using K=1.2.

Code: Select all

def logis(cp):
    K = 1.2    
    return 100 * (1 / (1 + math.exp(-K*cp/400)))

noobpwnftw
Posts: 338
Joined: Sun Nov 08, 2015 10:10 pm

Re: Database snapshot

Post by noobpwnftw » Wed Jul 31, 2019 3:10 am

My scores are evaluations of moves at current position, in relation to those of other moves of the same position, they have no meaning to any subsequent positions their corresponding moves may lead to(the back-propagation algorithm came up with the scores in the first place, one shouldn't try to put chickens back into eggs).

A "new" move may be discovered later to be better than what is known from the evaluations by the DB using a specific version of engine at the time, which is designed to have a considerable correctness rate. After that, there are mechanisms to refine those results by various means, one simple way is to probe the DB of the position after a move, it would automatically queue unexplored moves directly connected to a previously known DB position(and probed recently) for analysis, another is to manually queue the position or move via API, finally there exists a "Discover" program people can run, which loads any UCI-compatible engine and play games, contributing moves encountered during the process.

Ferdy
Posts: 4070
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Database snapshot

Post by Ferdy » Wed Jul 31, 2019 3:55 am

noobpwnftw wrote:
Wed Jul 31, 2019 3:10 am
finally there exists a "Discover" program people can run, which loads any UCI-compatible engine and play games, contributing moves encountered during the process.
I tried to compile ChessDBDiscover and it seems to work. Created an engine dir and put engine.exe, a renamed stockfish dev engine, then run ChessDBDiscover.exe with default config values.

Image

So what does the ChessDBDiscover.exe is doing? Does it play engine vs engine matches based on StartFEN in config file? If so how many games? Does it send the results to the cloud?

User avatar
Ovyron
Posts: 2488
Joined: Tue Jul 03, 2007 2:30 am

Re: Database snapshot

Post by Ovyron » Wed Jul 31, 2019 4:11 am

noobpwnftw wrote:
Wed Jul 31, 2019 3:10 am
the back-propagation algorithm came up with the scores in the first place, one shouldn't try to put chickens back into eggs
But if 10...Qc7 eventually leads to a score of 0.00 while 10...O-O leads to a score of -0.05 and 10...Nb7 leads to a score of -0.02, against white's best play, surely Qc7 should be ranked third best? The more blunders in the database played by white, the better 10...Qc7 looks even though there's a line that leads to 0.00 and nothing better for black?

(I'm just trying to demonstrate that the back-propagation algorithm is flawed)
Great spirits have always encountered violent opposition from mediocre minds.

noobpwnftw
Posts: 338
Joined: Sun Nov 08, 2015 10:10 pm

Re: Database snapshot

Post by noobpwnftw » Wed Jul 31, 2019 7:45 am

Ovyron wrote:
Wed Jul 31, 2019 4:11 am
noobpwnftw wrote:
Wed Jul 31, 2019 3:10 am
the back-propagation algorithm came up with the scores in the first place, one shouldn't try to put chickens back into eggs
But if 10...Qc7 eventually leads to a score of 0.00 while 10...O-O leads to a score of -0.05 and 10...Nb7 leads to a score of -0.02, against white's best play, surely Qc7 should be ranked third best? The more blunders in the database played by white, the better 10...Qc7 looks even though there's a line that leads to 0.00 and nothing better for black?

(I'm just trying to demonstrate that the back-propagation algorithm is flawed)
Within a small margin, the more blunders in the database for white, the better it looks for black due to weighted averaging because otherwise it is no different than using an engine or building a minmax tree.

Post Reply