# Is this a fortress?

I have just played a thoroughly enjoyable game, where in the middlegame I was forced to give up the queen for two minor pieces and defend. After 30 moves and some trades, the situation got simplified to following position:

``````[fen "8/4k3/2p5/3p4/1P1q4/3N4/2P5/2KN4 w - - 0 0"]

1. Kd2 Kd6 2. N1b2 Qh4 3. Na4 Qh6+ 4. Kc3 Qf6+ 5. Kb3 Qf7 6. Nac5 d4+ 7. Kb2 Qc4 8. Ne4+ Kc7 9. Kc1 Qd5 10. Nec5 Qg5+ 11. Kb2 Kb6 12. Kb3 Kb5 13. Kb2 Qd5 14. Nb3 Qc4 15. Nbc5 Qc3+ 16. Kc1 Kc4 17. Kb1 Qa3 18. Ne4 Qa8 19. Nd2+ Kc3 20. Ne4+ Kc4 21. Nd2+ Kd5 22. Nb3 Qa3 23. Nbc5 Kc4 24. Ne4 Qa8 1/2-1/2
``````

As you can see, the game went further on for some 25 moves; I managed to hold the position and the game ended in a threefold repetition.

As far as I can tell, with correct play by white, black should not be able to make any progress. Is that really so? (EDIT: Just to clarify, I seek some conclusive analysis, not merely a yes/no statement)

• Downvoter - would you care to elaborate as to why this is not a good question? Mar 13, 2016 at 18:16
• I'm just wondering, what answer do you expect? Mar 14, 2016 at 14:54
• @Saibot, GloriaVictis may correct me, but I think the goal is to obtain concrete annotations explaining why this is (or isn't) indeed a fortress. IMO the last part of your answer (re: aiming for c2 with the queen and king) takes the first step in that direction, but the sort of answer being sought is one which explains in detail with variations whether or not that winning approach can be realized or not.
– ETD
Mar 14, 2016 at 23:04
• @ETD I think it's more proper to use reliable tools, such as tablebase and Monte Carlo simulation instead of just talking about ideas. Because we are not endgame specialists, and our analyses will probably be flawed. Anyway, if GloriaVictis wants to hear opinions, here is my opinion. Mar 15, 2016 at 1:11
• @ETD sums up what I am looking for pretty well. If such annotations can further be backed up by tablebase or simulations, so much the better and I may even then provide a bounty. Regarding your suggestion to run the simplified version of the position through a 7-men tablebase - I do not have access to that, but somebody here might. Mar 15, 2016 at 8:26

White defends, but it is really a pain to hold this position. It took me few minutes to formulate correct plan since Stockfish found some strong ideas for Black.

Below is the image of the fortress we are aiming for:

If Black king tries to penetrate through queenside, he must aim for `c3` square in order to create double attack on `c2` pawn (using king and queen). We stop that plan with timely `Na4+`, which pushes the king back.

If Black tries to penetrate to `b1` with the king ( through queenside ), hoping to win the `c2` pawn ( queen + king create double attack on the pawn) White will be in time to reposition `Nc5` to `e2`, blocking black queen's attack along the second rank, thus defending by simply moving the king to `d2` and `d1`.

The last option for Black is to get to `d1` or `d2` with the king, but White manages to reposition `Nc5` to `d2` which creates impenetrable fortress since both knights control so many important squares (`e5`, `e4`, `f4`, `f3`, `f2`, `f1`), effectively cutting off the black king. White can also move the king to `b1` and push out black king with timely `Nb2+`, in case he is late to reposition the knight to `d2`.

The whole point is that we must reposition `Nd1` to `c5`, after which White defends.

Since analysis is very lengthy, I will post an illustrative game I played against Stockfish. Other important ideas will be shown in sublines.

``````[Title "AlwaysLearningNewStuff vs Stockfish"]
[FEN "8/4k3/2p5/3p4/1P1q4/3N4/2P5/2KN4 w - - 0 1"]

1. Kd2 Qg1 (1...Ke6 2.Nc3 Kf5 3.Na4 Ke4 4.Nac5+ Kf3 5.Nb3 Qe3+ 6.Kc3! Ke2 7.Kb2 Kd1 8.Kb1! Qe2 9.Nb2+) 2. Nc3! d4 3. Ne2 Qe3+ 4. Kd1 Ke6 5. Nec1 Qg1+ 6. Kd2 Qh2+ (6...Kf5 7.Nb3 Ke4 8.Nbc5+ Kf3 9.Ne5+ Kf2 10.Ned3+ Kf1 11.Kc1! Ke2+ 12.Kb2 Kd2 13.Nb3+ Kd1 14.Kb1! Qg2 15.Nb2+!? Ke2) 7. Kd1 Kd5 8. Nb3! Kc4 (8...Ke4 9.Nd2+ Ke3 ?? 10.Nf1+ Ke4 11.Nxh2+-) 9. Nbc5! \$8 Qg1+ 10. Kd2 Qe3+ 11. Kd1 Kc3 12. Na4+ \$8 Kc4 13. Nac5 Qg5! 14. Na4 \$7 Qg1+ 15. Kd2 Qg5+ 16. Kd1 Kb5 17. Nac5 \$8 Qg1+ 18. Kd2 Kc4 19. Na4 \$8 Qe3+ 20. Kd1 Kb5 21. Nac5 Kc4 22. Na4 Qg5 23. Nac5 Kb5 24. Nb3 Ka4 25. Nbc5+ Kb5 (25...Ka3 26.Nb3! Qg1+! 27.Ke2! Ka2 28.Nd2! Qe3+ 29.Kd1 Qe6! 30.Kc1! Qh6 31.Kd1 Qh1+ 32.Ke2) 26. Nb3 Qe3 27. Nbc5 Qg1+ 28. Kd2 Qg5+ 29. Kd1 Qh6 30. Nb3 Kc4 31. Nbc5 Qh1+ 32. Kd2 Qf1 33. Na4 Qg1 34. Nac5 Kd5 35. Ke2 Qg2+ 36. Kd1 Qh1+ 37. Kd2 Kc4 38. Na4 Qh6+ 39. Kd1 Kb5 40. Nac5 Kc4 41. Na4 Qg5 42. Nac5 Qh4 43. Kd2 Qh6+ 44. Kd1 Qh1+ 45. Kd2 Qa1 46. Ne4! Qa7 47. Nec5 Qa1 48. Ne4 Qf1 49. Nec5 Kd5 50. Na4 c5 51. bxc5! Qg2+ 52. Kc1! Qg5+ 53. Kb2 Ke4 54. Nb6! Qd2 55. Kb1 Qg5 56. Nc4! Kd5 57. Nd6! Qg1+ 58. Kb2 Qg4 59. Kb1 Qd1+ 60. Kb2 Qg1 61. Ka2 Kc6 62. Kb2 Kd5 63. Ka2 Qh1 64. Kb2 Qd1 65. Nb5 Qg1 66. Nd6 Qd1 67. Nb5 Qf1 68. Nd6 Kc6 69. Nc4 Qd1 70. Nd6 Qd2 71. Kb1 Kd5 72. Kb2 Qh6 73. Kb1 Kc6 74. Kb2 Kd5 75. Kb1 Kc6 76. Kb2 Qh1 77. Ka2 Qg1 78. Kb2 Qg8 79. Kb1 Kd5 80. Kb2 Qh7 81. Kb1 Qh6 82. Kb2 Kc6 83. Kb1 Qg7 84. Kb2 Qh6 85. Kb1 Qh7 86. Kb2 Qh1 87. Ka2 Qh4 88. Kb2 Kd5 89. Ka2 Qh2 90. Kb2 Qh1 91. Ka2 Qd1 92. Kb2 Qf1 93. Ka2 Qf3 94. Kb2 Qh1 95. Ka2 Ke6 96. Kb2 Qd1 97. Nc4 Kd7 98. Nd6 Kc6 99. Nc4 Kc7 100. Nd6 Kb8 101. Nc4 1/2-1/2
``````
• Excellent answer, thanks. It is impressive how many strong plans the machine manages to generate! Apr 27, 2016 at 14:22
• It is impressive how many strong plans the machine manages to generate! I agree, Stockfish gave me so much headache before I finally managed to find the correct plan. Thanks for accepting, if you need anything just leave a comment, but as long as you understand that `Nc5` is used to push back black king you will be fine. Best regards! Apr 27, 2016 at 15:03
• Thanks for the bounty, you really didn't have to do that... I am glad that I have helped, and to be honest, the position was instructive to analyse, which bolstered my endgame knowledge. Again, thank you for the bounty, best regards until next time! Apr 28, 2016 at 16:12

Pawn on b4 and c6 does not contribute to position significantly.

I suggest, remove these pawns and check the result of 7-men position from Lomonosov tablebase.

Also, you can conduct a Monte Carlo simulation, in which you use same engine with same hardware, playing against each other starting from this position. 30 seconds time control is enough, enable tablebases if you can, let the simulation play 100 games. If all of them draws, the position is almost certainly draw.

Also from my perspective, in order to progress Black King should get close to capture the c2 pawn. Providing this, Black can push and capture the pawn, otherwise Black Queen can't penetrate alone. And as far as I can see, Black King can't get close to pawn because of the White Knights.

• I didn't vote downvote this answer, and I agree that attacking c2 is likely the most promising winning attempt. But I think it's a mistake to suggest that b4 and c6 don't matter; with them still around, e.g. a queen sac for both knights could sometimes end in a won pawn ending. For example, while trying to attack c2, suppose we get to a position with Kb1 Nb3 Nd3 b4 c2 for white, and Ke2 Qc3 c6 d5 for black, with black to move. (I'm not saying this position can be forced.) Winning c2 seems still out of reach, but even so there is a win here by 1...Qxb3+! 2.cxb3 Kxd3, with a won pawn ending.
– ETD
Mar 14, 2016 at 23:16
• I mean, if Black can't penetrate without those pawns, can't penetrate with those pawns too, because those pawns do not contribute. And this is useful to evaluate the position I think. But yeah nice point, queen sac against two knights is a possibility, I didn't notice it. Mar 15, 2016 at 1:03