Is white winning with a ~1.5 score, not growing?

[Pafifi]

[pgn]
[FEN "1rb2r1k/3p1pp1/p3p2p/1p2P3/4q3/1N2B1R1/PPP1N1P1/2KR1B2 w - - 0 1"]

1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5
[/pgn]
This puzzle is from an EPD I found in our forum (so long ago that I forgot the URL). The best move given is Nf4 and I probed a few moves with stockfish. However, the score for white is always around 1.5 and I can't find a clearly winning path. I have SF fight itself with a 3 min time control and it draws.

[Vinvin]

The position is in my latest testset.
1rb2r1k/3p1pp1/p3p2p/1p2P3/4q3/1N2B1R1/PPP1N1P1/2KR1B2 w - - bm Nf4; id "Hard-Talkchess-2020.026" ;

The source I found is :
ICCF 2011 : http://immortalchess.net/forum/archive/ ... 2-p-6.html

http://www.talkchess.com/forum3/viewtop ... 17#p662017
https://github.com/mcostalba/PoorFish/b ... 120sec.pgn

[Hai]

[pgn]
[FEN "1rb2r1k/3p1pp1/p3p2p/1p2P3/4q3/1N2B1R1/PPP1N1P1/2KR1B2 w - - 0 1"]

1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5
[/pgn]
This puzzle is from an EPD I found in our forum (so long ago that I forgot the URL). The best move given is Nf4 and I probed a few moves with stockfish. However, the score for white is always around 1.5 and I can't find a clearly winning path. I have SF fight itself with a 3 min time control and it draws.

LC0 40x512 and 30x384 have completely different opinions compared to the moves you have displayed as best.

After 7.Rxc5 you can play 8.a3 +1.30 or 8.b4 +1.40.

At the beginning with 0 seconds, LC0 40x512 said 1.Bc5 is +1.55, but later:

1.Bc5 Rg8 2.Re3 Qh4 3.g3 Qh2 4.Nf4 Bb7 5.Bd3 Rgc8 6.Rf1 Kg8 7.Kb1 Rc7 8.Bb4 Rbc8 9.Nd2 Rc5 10.a3 R5c7 11.Kc1 Rc6 12.Be4 R6c7 13.Bd3 Ba8 14.c3 Bd5 15.Nxd5 exd5 16.Nb3 Qh5 17.Kb1 Rc4 18.Ka2
+/- (1.33) depth: 18/41 00:05:58 1702kN
But it's still +0.10 better evaluation than Stockfish from live book with Nf4.

[zullil]

[pgn]
[FEN "1rb2r1k/3p1pp1/p3p2p/1p2P3/4q3/1N2B1R1/PPP1N1P1/2KR1B2 w - - 0 1"]

1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5
[/pgn]
This puzzle is from an EPD I found in our forum (so long ago that I forgot the URL). The best move given is Nf4 and I probed a few moves with stockfish. However, the score for white is always around 1.5 and I can't find a clearly winning path. I have SF fight itself with a 3 min time control and it draws.

Stockfish-dev currently has:

+1.58 1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5 8. b4 Rc7 9. Kb2 Ba8 10. a3 Bb7 11. c3 Rgc8 12. Bc2 Rd8 13. Nd3 Bd5 14. Nc5 d6 15. Nb3 Rf8 16. Nd4 Rb8 17. Bd3 Rbc8 18. Bb1 Rf8 19. Rhh3 Rd8 20. Bd3 Bb7 21. Rh1 Rb8 22. Rh5 Rbc8 23. Bc2 Kg8 24. Rh1 Kf8 25. Bb3 Re8 26. Rf1 Kg8 27. Re1 Rb8 28. Rh1 Kh7 29. Bc2+ Kh8 30. Rh2 Rf8 31. Re3 Rd8 32. a4 bxa4 33. g4 a3+ 34. Kxa3 Kg8 35. Kb2 e5 36. Nf5 f6 37. g5 (depth 63, 1:55:35)

Lc0 (network 384x30-t40-swa-2036000) currently has:

Nodes: 38.0M, N/s: 13.0k, Time: 48m 51s

68.7% Bc5 Rg8 Re3 Qh4 g3 Qd8 a3 Bb7 Bd6 Rc8 Nc5 Qb6 b4 a5 Kb2 axb4 axb4 Ra8 Nc3 Bc6 Rd4 f5 exf6 gxf6 Be7 Rg6 Bd3 Qa7 Nb3 Qa3+ Kb1 f5 Bd6 Rg5 Be2 Kg8 Bf4 Rg7 Bxb5 Bxb5 Nxb5 Qa2+ Kc1 h5 Nc3 Qa7 Red3 d5 Be5 Rg4 b5 Rc8 Kb2 Rc4 Rd1 Qb6 Rxc4 Rxc4 (N: 36.3M, WDL: 536 303 161, P: 19.06%)

70.9% a3 Rd8 Nd2 Qf5 Rf3 Qg4 Rf4 Qg6 Ne4 Bb7 Nd6 Kg8 Nc3 Qg5 Nce4 Bxe4 Rxe4 Qg3 Bf4 Qg6 Bd3 Qxg2 Re3 Qg4 Rf1 b4 a4 b3 c3 Qh4 Rg3 Rdc8 Nxc8 Rxc8 Rgg1 Rc5 Rh1 Qd8 (N: 1.1M, WDL: 570 279 151, P: 1.56%)

65.6% Nc5 Qf5 Bd4 Bb7 Nxb7 Rxb7 Kb1 Rc7 Nc3 Qf4 Re3 Rc6 g3 Qg4 Be2 Qg6 Bd3 Qh5 Be2 Qg6 a3 Kg8 Bd3 Qg4 Re4 Qg5 Rf4 d5 Ne2 f6 Rdf1 (N: 221k, WDL: 492 327 181, P: 12.92%)

Both searches are continuing ...

[zullil]

[pgn]
[FEN "1rb2r1k/3p1pp1/p3p2p/1p2P3/4q3/1N2B1R1/PPP1N1P1/2KR1B2 w - - 0 1"]

1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5
[/pgn]
This puzzle is from an EPD I found in our forum (so long ago that I forgot the URL). The best move given is Nf4 and I probed a few moves with stockfish. However, the score for white is always around 1.5 and I can't find a clearly winning path. I have SF fight itself with a 3 min time control and it draws.

Stockfish-dev currently has:

+1.58 1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5 8. b4 Rc7 9. Kb2 Ba8 10. a3 Bb7 11. c3 Rgc8 12. Bc2 Rd8 13. Nd3 Bd5 14. Nc5 d6 15. Nb3 Rf8 16. Nd4 Rb8 17. Bd3 Rbc8 18. Bb1 Rf8 19. Rhh3 Rd8 20. Bd3 Bb7 21. Rh1 Rb8 22. Rh5 Rbc8 23. Bc2 Kg8 24. Rh1 Kf8 25. Bb3 Re8 26. Rf1 Kg8 27. Re1 Rb8 28. Rh1 Kh7 29. Bc2+ Kh8 30. Rh2 Rf8 31. Re3 Rd8 32. a4 bxa4 33. g4 a3+ 34. Kxa3 Kg8 35. Kb2 e5 36. Nf5 f6 37. g5 (depth 63, 1:55:35)

Lc0 (network 384x30-t40-swa-2036000) currently has:

Nodes: 38.0M, N/s: 13.0k, Time: 48m 51s

68.7% Bc5 Rg8 Re3 Qh4 g3 Qd8 a3 Bb7 Bd6 Rc8 Nc5 Qb6 b4 a5 Kb2 axb4 axb4 Ra8 Nc3 Bc6 Rd4 f5 exf6 gxf6 Be7 Rg6 Bd3 Qa7 Nb3 Qa3+ Kb1 f5 Bd6 Rg5 Be2 Kg8 Bf4 Rg7 Bxb5 Bxb5 Nxb5 Qa2+ Kc1 h5 Nc3 Qa7 Red3 d5 Be5 Rg4 b5 Rc8 Kb2 Rc4 Rd1 Qb6 Rxc4 Rxc4 (N: 36.3M, WDL: 536 303 161, P: 19.06%)

70.9% a3 Rd8 Nd2 Qf5 Rf3 Qg4 Rf4 Qg6 Ne4 Bb7 Nd6 Kg8 Nc3 Qg5 Nce4 Bxe4 Rxe4 Qg3 Bf4 Qg6 Bd3 Qxg2 Re3 Qg4 Rf1 b4 a4 b3 c3 Qh4 Rg3 Rdc8 Nxc8 Rxc8 Rgg1 Rc5 Rh1 Qd8 (N: 1.1M, WDL: 570 279 151, P: 1.56%)

65.6% Nc5 Qf5 Bd4 Bb7 Nxb7 Rxb7 Kb1 Rc7 Nc3 Qf4 Re3 Rc6 g3 Qg4 Be2 Qg6 Bd3 Qh5 Be2 Qg6 a3 Kg8 Bd3 Qg4 Re4 Qg5 Rf4 d5 Ne2 f6 Rdf1 (N: 221k, WDL: 492 327 181, P: 12.92%)

Both searches are continuing ...

Stockfish-dev:

+1.57 1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5 8. b4 Rc7 9. Kb2 Ba8 10. a3 Bb7 11. c3 Rgc8 12. Bc2 Rd8 13. Nd3 Bd5 14. Ne5 d6 15. Nf3 Bc6 16. Nd4 Bd7 17. Be4 Rb8 18. Nb3 Rb6 19. Bb1 Bc6 20. Nd4 Be8 21. Bc2 Bd7 22. Bd3 d5 23. Nb3 Rd6 24. Rf3 Kg8 25. Re3 Kf8 26. Be2 Bc6 27. Rd1 Be8 28. Na5 Kg8 29. Rd2 Kf8 30. Bd1 Kg8 31. Nb3 Kf8 32. Re1 Rdc6 33. Rd3 Rd6 34. Bf3 Ra7 35. Rde3 Bc6 36. Nd4 Bd7 37. g3 Rc7 38. Nb3 Bc6 39. Be2 Be8 40. Nd4 Bd7 41. Rh1 f6 42. Rf1 Ra7 (depth 71, 6:36:22)

Lc0 (network 384x30-t40-swa-2036000):

Nodes: 100M, N/s: 13.6k, Time: 2h 2m (limit met)

71.2% a3 a5 Bc5 Rg8 Nxa5 b4 Re3 Qh4 axb4 Rxb4 Bxb4 Qxb4 Nb3 Bb7 Ned4 Rb8 c4 Qe7 Nc2 h5 Rd4 Bc6 Nd2 g6 g3 f6 exf6 Qxf6 Be2 d5 Nf3 Ba4 c5 Bxc2 Kxc2 g5 Re5 Qg6+ Rd3 Qf6 Rde3 g4 Nd4 h4 Rh5+ Kg8 (N: 63.1M, WDL: 556 312 132, P: 1.56%)

68.7% Bc5 Rg8 Re3 Qh4 g3 Qd8 a3 Bb7 Bd6 Rc8 Nc5 Qb6 b4 a5 Kb2 axb4 axb4 Ra8 Nc3 Bc6 Rd4 f5 exf6 gxf6 Be7 Rg6 Bd3 Qa7 Nb3 Qa3+ Kb1 f5 Bd6 Rg5 Be2 Kg8 Bf4 Rg7 Bxb5 Bxb5 Nxb5 Qa2+ Kc1 h5 Nc3 Qa7 Red3 d5 Be5 Rg4 b5 Rc8 Kb2 Rc4 Rd1 Qb6 Rxc4 Rxc4 (N: 36.3M, WDL: 536 303 161, P: 19.06%)

65.6% Nc5 Qf5 Bd4 Bb7 Nxb7 Rxb7 Kb1 Rc7 Nc3 Qf4 Re3 Rc6 g3 Qg4 Be2 Qg6 Bd3 Qh5 Be2 Qg6 a3 Kg8 Bd3 Qg4 Re4 Qg5 Rf4 d5 Ne2 f6 Rdf1 (N: 221k, WDL: 492 327 181, P: 12.92%)

Stopping here.

[peter]

The source I found is :
ICCF 2011 : http://immortalchess.net/forum/archive/ ... 2-p-6.html

This position arose in 5 games stored at Herbert Bellman's corr.CD between 2011 und 2018, each time White won with Nf4, so I guess it's not a bad move at least.

BTW Walter Eigenmann has it in his "Nightmare2"- collection too,
https://glarean-magazin.ch/2016/08/25/c ... eigenmann/

[Jouni]

But end position of SF long think is drawish! Definitely not good test position.

[Pafifi]

[pgn]
[FEN "1rb2r1k/3p1pp1/p3p2p/1p2P3/4q3/1N2B1R1/PPP1N1P1/2KR1B2 w - - 0 1"]

1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5
[/pgn]
This puzzle is from an EPD I found in our forum (so long ago that I forgot the URL). The best move given is Nf4 and I probed a few moves with stockfish. However, the score for white is always around 1.5 and I can't find a clearly winning path. I have SF fight itself with a 3 min time control and it draws.

LC0 40x512 and 30x384 have completely different opinions compared to the moves you have displayed as best.

After 7.Rxc5 you can play 8.a3 +1.30 or 8.b4 +1.40.

At the beginning with 0 seconds, LC0 40x512 said 1.Bc5 is +1.55, but later:

1.Bc5 Rg8 2.Re3 Qh4 3.g3 Qh2 4.Nf4 Bb7 5.Bd3 Rgc8 6.Rf1 Kg8 7.Kb1 Rc7 8.Bb4 Rbc8 9.Nd2 Rc5 10.a3 R5c7 11.Kc1 Rc6 12.Be4 R6c7 13.Bd3 Ba8 14.c3 Bd5 15.Nxd5 exd5 16.Nb3 Qh5 17.Kb1 Rc4 18.Ka2
+/- (1.33) depth: 18/41 00:05:58 1702kN
But it's still +0.10 better evaluation than Stockfish from live book with Nf4.

1.Bc5 Rg8 2.Re3 Bb7! SF thinks that black is better now.

[drewdrew]

Stockfish-dev:

+1.57 1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5 8. b4 Rc7 9. Kb2 Ba8 10. a3 Bb7 11. c3 Rgc8 12. Bc2 Rd8 13. Nd3 Bd5 14. Ne5 d6 15. Nf3 Bc6 16. Nd4 Bd7 17. Be4 Rb8 18. Nb3 Rb6 19. Bb1 Bc6 20. Nd4 Be8 21. Bc2 Bd7 22. Bd3 d5 23. Nb3 Rd6 24. Rf3 Kg8 25. Re3 Kf8 26. Be2 Bc6 27. Rd1 Be8 28. Na5 Kg8 29. Rd2 Kf8 30. Bd1 Kg8 31. Nb3 Kf8 32. Re1 Rdc6 33. Rd3 Rd6 34. Bf3 Ra7 35. Rde3 Bc6 36. Nd4 Bd7 37. g3 Rc7 38. Nb3 Bc6 39. Be2 Be8 40. Nd4 Bd7 41. Rh1 f6 42. Rf1 Ra7 (depth 71, 6:36:22)

At higher depth, Stockfish here - default contempt - changed its mind to Nc5 with an increasing score.
Nf4 still played one move later though:

info depth 77 seldepth 134 multipv 1 score cp 274 nodes 21413489726436 nps 356142409 hashfull 1000 tbhits 29589812328 time 60126200 pv b3c5 e4f5 e2f4 f5e5 f1d3 h8g8 d1h1 f7f5 d3e2 f8f7 c5d3 e5e4 h1h6 e4e3 g3e3 g7h6 d3e5 f7g7 e3d3 b5b4 d3d4 g8h7 d4d1 a6a5 g2g4 f5g4 e5g4 g7g5 g4f6 h7h8 f6e4 g5g7 e4d6 c8b7 b2b3 b7a8 d1f1 b8b6 d6c4 b6b8 e2d3 a8g2 f1f2 g2c6 f4g6 h8g8 c4e5 c6b5 d3b5 g7g6 e5g6 b8b5 c1d2 b5d5 d2e3 g8g7 g6f4 d5e5 e3d4 e5f5 d4e4 f5c5 f4d3 c5h5 c2c4 b4c3 e4d4 c3c2 f2c2 h5d5 d4c4 g7f7 c2h2 h6h5 d3f4 d5g5 f4h5 f7e7 h2h4 e7d6 c4c3 g5c5 c3d3 c5f5 h5f4 d6c6 f4e2 c6d6 h4h7 f5d5 d3c2 d5g5 h7h4 g5c5 e2c3 c5e5 c2d3 e5f5 d3d4 d6c6 h4h7 f5f2 h7h8 f2f4 d4e3 f4f5 e3d3 f5f3 d3d2 c6c5 c3e2 d7d5 h8a8 c5b4 a8a7 e6e5 a7e7 b4a3 e7e5 a3a2 e2d4 f3g3 e5d5

[zullil]

Stockfish-dev:

+1.57 1. Nf4 Qxe5 2. Bd3 Bb7 3. Rh1 Qd6 4. Nh5 Rg8 5. Nc5 Rbc8 6. Bf4 Qxf4+ 7. Nxf4 Rxc5 8. b4 Rc7 9. Kb2 Ba8 10. a3 Bb7 11. c3 Rgc8 12. Bc2 Rd8 13. Nd3 Bd5 14. Ne5 d6 15. Nf3 Bc6 16. Nd4 Bd7 17. Be4 Rb8 18. Nb3 Rb6 19. Bb1 Bc6 20. Nd4 Be8 21. Bc2 Bd7 22. Bd3 d5 23. Nb3 Rd6 24. Rf3 Kg8 25. Re3 Kf8 26. Be2 Bc6 27. Rd1 Be8 28. Na5 Kg8 29. Rd2 Kf8 30. Bd1 Kg8 31. Nb3 Kf8 32. Re1 Rdc6 33. Rd3 Rd6 34. Bf3 Ra7 35. Rde3 Bc6 36. Nd4 Bd7 37. g3 Rc7 38. Nb3 Bc6 39. Be2 Be8 40. Nd4 Bd7 41. Rh1 f6 42. Rf1 Ra7 (depth 71, 6:36:22)

At higher depth, Stockfish here - default contempt - changed its mind to Nc5 with an increasing score.
Nf4 still played one move later though:

info depth 77 seldepth 134 multipv 1 score cp 274 nodes 21413489726436 nps 356142409 hashfull 1000 tbhits 29589812328 time 60126200 pv b3c5 e4f5 e2f4 f5e5 f1d3 h8g8 d1h1 f7f5 d3e2 f8f7 c5d3 e5e4 h1h6 e4e3 g3e3 g7h6 d3e5 f7g7 e3d3 b5b4 d3d4 g8h7 d4d1 a6a5 g2g4 f5g4 e5g4 g7g5 g4f6 h7h8 f6e4 g5g7 e4d6 c8b7 b2b3 b7a8 d1f1 b8b6 d6c4 b6b8 e2d3 a8g2 f1f2 g2c6 f4g6 h8g8 c4e5 c6b5 d3b5 g7g6 e5g6 b8b5 c1d2 b5d5 d2e3 g8g7 g6f4 d5e5 e3d4 e5f5 d4e4 f5c5 f4d3 c5h5 c2c4 b4c3 e4d4 c3c2 f2c2 h5d5 d4c4 g7f7 c2h2 h6h5 d3f4 d5g5 f4h5 f7e7 h2h4 e7d6 c4c3 g5c5 c3d3 c5f5 h5f4 d6c6 f4e2 c6d6 h4h7 f5d5 d3c2 d5g5 h7h4 g5c5 e2c3 c5e5 c2d3 e5f5 d3d4 d6c6 h4h7 f5f2 h7h8 f2f4 d4e3 f4f5 e3d3 f5f3 d3d2 c6c5 c3e2 d7d5 h8a8 c5b4 a8a7 e6e5 a7e7 b4a3 e7e5 a3a2 e2d4 f3g3 e5d5

Were you known as drudru elsewhere? I think I recognize your massive hardware.