Realistic test positions for beginners

[Henk]

See end game. Source for creating test position(s). Many misevaluations of my engine.

[pgn]
[Event "Computer Chess Game"]
[Site "HP"]
[Date "2016.05.13"]
[Round "-"]
[White "Fairy-Max 4.8S"]
[Black "SkipperWinb"]
[Result "1/2-1/2"]
[TimeControl "120"]
[Annotator "1. +0.11 1... +0.06"]

1. c3 {+0.11/8} Nf6 {+0.06/17 2.4} 2. f4 {+0.08/8 4} Nc6 {+0.00/18 2.4} 3.
Qa4 {+0.04/8 1.9} d5 {+0.00/17 2.4} 4. Nf3 {+0.10/9 4} Qd7 {-0.04/15 2.3}
5. d4 {+0.02/8 2.9} Ne4 {+0.06/14 2.3} 6. Be3 {+0.09/8 2.3} e5
{-0.09/13 2.3} 7. Nbd2 {+0.10/8 2.2} exf4 {-0.14/12 2.2} 8. Bxf4
{-0.10/9 2.3} Be7 {-0.06/13 2.2} 9. Nxe4 {+0.12/9 1.7} dxe4 {+0.07/13 2.1}
10. Nd2 {-0.12/8 1.8} g5 {-0.06/13 2.0} 11. Be3 {-0.07/9 1.9} f5
{+0.00/15 2.0} 12. g3 {+0.01/9 1.9} O-O {+0.02/14 2.0} 13. Qc4+ {+0.11/8 7}
Kh8 {+0.05/12 2.0} 14. d5 {-0.15/8 2.4} Ne5 {+0.00/15 1.9} 15. Bd4
{-0.02/9 2.2} Bf6 {+0.07/15 1.9} 16. Qc5 {+0.00/9 2.0} Rf7 {+0.21/14 1.8}
17. Nc4 {-0.09/9 2.3} b6 {+0.26/16 1.8} 18. Nxe5 {-0.17/10 1.0} bxc5
{+0.47/16 1.9} 19. Nxd7 {-0.33/12 1.3} Bxd4 {+0.20/18 1.8} 20. cxd4
{-0.29/12 1.3} Rxd7 {+0.11/19 1.7} 21. dxc5 {-0.23/12 2.1} Rxd5
{+0.00/19 2.0} 22. Rc1 {-0.14/11 1.1} Rb8 {+0.04/19 2.0} 23. Rc2
{-0.65/10 1.3} Rb5 {+0.69/19 1.9} 24. c6 {-0.96/11 2.0} Rbc5 {+0.69/19 1.9}
25. Rxc5 {-0.98/13 2.2} Rxc5 {+0.73/19 1.9} 26. Bg2 {-0.96/12 1.8} Rc2
{+0.77/15 1.8} 27. O-O {-0.41/10 1.1} Rxe2 {+1.35/15 1.7} 28. Rd1
{-0.67/10 0.8} Be6 {+1.29/14 1.8} 29. Rd8+ {-0.71/9 1.3} Bg8 {+1.29/14 1.7}
30. Re8 {-0.89/10 2.7} g4 {+1.29/13 1.6} 31. h3 {-0.35/10 1.2} gxh3
{+1.34/15 1.6} 32. Bxh3 {-0.11/11 1.1} e3 {+0.63/14 1.7} 33. Bxf5
{-0.15/11 1.2} Kg7 {+0.03/15 1.6} 34. Be4 {-0.22/11 2.8} Rxb2
{+0.50/13 1.4} 35. Bf3 {-0.19/11 1.0} Bxa2 {+0.51/13 1.4} 36. Rxe3
{-0.27/11 0.6} Rb3 {+1.18/15 1.4} 37. Kf2 {-0.35/14 1.0} Rxe3
{+1.66/17 1.3} 38. Kxe3 {-0.49/13 5} Bb3 {+1.74/24 1.4} 39. Kd4
{-0.44/11 0.6} a5 {+1.70/22 1.3} 40. Kc5 {-0.36/11 0.7} Kg6 {+1.34/20 1.2}
41. Be4+ {-0.29/13 1.0} Kg5 {+0.89/19 0.8} 42. Bf3 {-0.33/12 0.6} Kf5
{+1.33/17 0.8} 43. Be2 {-0.77/12 0.7} Ke4 {+1.33/20 0.8} 44. g4
{-0.39/12 0.5} h6 {+1.34/19 0.8} 45. Kb5 {-0.76/13 0.7} a4 {+0.64/20 0.8}
46. Kb4 {-0.77/13 0.8} Kf4 {+0.78/22 0.8} 47. Bb5 {-0.72/12 0.5} Kxg4
{+0.78/20 0.8} 48. Bd3 {-1.43/12 1.0} h5 {+1.99/20 0.8} 49. Be2+
{-1.65/13 1.0} Kg5 {+2.00/21 0.8} 50. Bb5 {-2.01/12 0.4} h4 {+2.41/19 0.8}
51. Bf1 {-2.61/13 1.1} Kg4 {+2.15/20 0.7} 52. Bg2 {-3.57/13 1.1} Bd1
{+2.32/19 0.7} 53. Bd5 {-3.72/13 0.9} h3 {+4.45/18 0.7} 54. Be6+
{-3.89/15 0.7} Kg3 {+4.14/21 0.7} 55. Bxh3 {-4.47/15 0.6} Kxh3
{+10.36/20 0.8} 56. Ka3 {-4.67/16 0.4} Bf3 {+10.55/25 0.6} 57. Kxa4
{-3.73/16 0.5} Bxc6+ {+10.59/25 0.7} 58. Kb4 {-3.80/16 0.4} Kg3
{+10.72/30 0.6} 59. Kc4 {-3.83/16 0.6} Bh1 {+10.72/28 0.6} 60. Kc3
{-4.08/16 0.5} Kf4 {+20.28/23 0.6} 61. Kd4 {-4.59/16 0.8} Kf5
{+20.29/29 0.6} 62. Kc3 {-5.21/16 0.7} Ke5 {+26.85/21 0.5} 63. Kc4
{-5.25/16 0.5} Be4 {+20.29/15 0.5} 64. Kc3 {-5.26/16 0.5} Kd5
{+79.14/25 0.5} 65. Kd2 {-5.31/16 0.5} c5 {+79.14/27 0.5} 66. Kd1
{-11.28/15 0.4} c4 {+79.14/22 0.5} 67. Kd2 {-11.27/15 0.5} Kd4
{+79.16/26 0.5} 68. Kd1 {-11.34/16 0.5} c3 {+79.16/18 0.5} 69. Ke2
{-11.35/14 0.3} Kd5 {+79.16/15 0.5} 70. Kd1 {-11.35/15 0.5} c2+
{+79.16/17 0.5} 71. Kc1 {-11.36/15 0.3} Kd4 {+79.16/23 0.5} 72. Kd2
{-11.36/15 0.3} c1=Q+ {+15.18/9 0.5} 73. Kxc1 {-3.07/16 0.4}
{Xboard adjudication: Insufficient mating material} 1/2-1/2
[/pgn]

[hgm]

Looks like a move generator bug. It seems it doesn't see that the Queen can be captured, giving a +15 score for d8=Q at depth too low to see the mate. (That is a bit of a high score anyway, for just Q+B and no prospect to get more. How do you value the pieces? Or do you give bonus points for a sure win?)

[Henk]

Yes you mean 72 .. c1Q. To be investigated. Also 56 .. Bf3 bad move. First I have to change BKk or bkK evaluations. Or more general these insufficient material cases.

[hgm]

I don't see what insufficient material has to do with this. It claims to have searched 25 ply on Bf3, so it should have seen that it can just gobble up the white Pawn with the King, without any need to abandon its a-Pawn. But from the score it says it sees a promotion. And it could very well be right; it can never force the a-Pawn to promotion with this shade of Bishop, so perhaps trying to hang on to it just slows down things. But the worry is that the promotion it sees is not a real one, but in fact the stunt it pulls off later in the game.