Three squares on the diagonal, but one draws and one loses.

It must have been incredibly difficult to play since the move the tablebase says is a mistake 69 Qf1+ loses in 83 moves with best play.

click for larger view

... is a theoretical draw per the 7-piece Lomonosov tablebases after 47.b4. What tablebases did you use?

Sorry <IMlday>. But you should be proud to have put up such a stubborn defense against Bisguier.

And I now always say "theoretical" because in a game between human players, particularly in a long game like this one where exhaustion is a factor, and even between the best chess engines, the two players don't necessarily always play the best moves.

47.Nb3 Draw
47.Nb7+ Draw
47.Ke3 Draw
47.Kf5 Draw
47.Kg4 Draw
47.Kg4 Draw
47.b4 Draw
47.Kf3 Black wins in 40
47.Kg3 Black wins in 32
47.b3 Black wins in 31
47.Nc6 Black wins in 11

So <IMlday>'s 47.Kf5 should have drawn, the mistake would have had to be made later. But I don't know if I have the motivation to run multiple FinalGen analyses in order to determine where the losing move was made!

However, I realized that I didn't have to run FinalGen in other positions. I could use the Lomonosov tablebases and do a binary-like search between 47.Kf5 which both the Lomonosov tablebases and FinalGen indicated was a Draw after the played 47...a6 and the game's final move 126...Kg7. I technically should run FinalGen again to verify that White has no escape (although it should be obvious) but there are too many pieces on the board (3 queens, and only 1 piece exclusive of kings and pawns are allowed for each side). But, a queen up, it should be obvious that the position is a win for Black.

But, just to make sure, Stockfish indicates that Black mates in 7 after 127.Qb7 and all other White moves result in mate in a lesser number of moves. And it reached d=120 while displaying the top 10 moves in less than 2 minutes. So, match that with your high powered system <RandomVisitor>! :-)

At any rate I did a binary-like search to try to bracket the moves that resulted in a Draw and a win for Black. It wasn't that straightforward because some moves resulted in a win for Black but they might have been drawn due to the 50-move draw rule, but I was just too lazy to verify that no pawns had been moved or no pieces captured within 50 moves. So make those Black wins conditional on the 50-move draw rule being satisfied. And remember that all wins and draws are "theoretical" since they assume best play by both sides at all times, something that's not likely to happen in a long game between two human players like this one.

In case you just want to know the result, I think that White's losing move was either:

(a) <69.Qf1+>. After 68...Kf4 White could have drawn after 69.Qb2 but after the move actually played, 69.Qf1+, Black can mate in 82 moves. But after the move actually played by Black, 69...Qf3, White had a draw after 70.Qg1. So a case of double inaccuracies by each side.

(b) <71.Kg7> Black mates in 52 moves after the played 71...f4. Period.

Here are the results in ascending move order with the second column highlighting the order in which the positions were evaluated and the third column showing the tablebase result after the move was played. It shows how the game transitions from being a Draw to a win by Black in a successively decreasing number of moves.

47.Kf5 <1> Draw after the played 47…a6.

57.b7 <5> Draw after 57…f5. Also draw after the played 57…a1Q

62.Qh4+ <7> Draw after 62…Qe4. Also draw after the played 62…Kd5.

64...Ke5 <9> Draw after 64.Qd6+. Also draw after the played 64…Kd5.

67.Qe1+ <4> Draw after the played 67…Qe4.

68.Qa1 <16> Draw after 68…Qd4. After the played 68…Kf4, also Draw.

68...Kf4 <11> Draw after 69.Qb2. But after the played 68.Qf1, Black mates in 82.

69.<Qf1+> <12> Black mates in 83 after 69…Kg5. But after the played 69…Qf3, it's a Draw.

69...Qf3 <10> Draw after 70.Qg1. But after the played 70.Qc1+, Black mates in 52.

70.<Qc1+> <13> Black mates in 52 after the played 70…Kg4.

70.Kg4 <14> Black mates in 51 after 71.Qb1. After the played 71.Kg7, Black mates in 49.

71.Kg7 <15> Black mates in 49 after the played 71…f4.

72.Kf6 <8> Black mates in 48 after 72…Kh3. After the played 72…Qe3, Black also mates in 48.

77.Kg6 <6> Black mates in 44 after 77…Qd5. After the played 77…Qe8+ Black mates in 45.

87.Kc6 <3> Black mates in 28 after 87.Qe6+. After the played 87.Qc3+, Black mates in 34.

And here is the search sequence I used to find when the position changed from a Draw to a win for Black. The first column highlights the order in which the positions were evaluated and it shows the binary-like sequence of the evaluations. The second column is the move played and the third column is the result after that move is played.

<1> 47.Kf5 Draw after the played 47…a6.

<2> 126...Kg7 Black mates in 7 after 127.Qb7.

<3> 87. Kc6 Black mates in 28 after 87.Qe6+. After the played 87.Qc3+, Black mates in 34.

<4> 67.Qe1+ Draw after the played 67…Qe4.

<5> 57.b7 Draw after 57…f5. Also draw after the played 57…a1Q

<6> 77.Kg6 Black mates in 44 after 77…Qd5. After the played 77…Qe8+ Black mates in 45.

<7> 62.Qh4+ Draw after 62…Qe4. Also draw after the played 62…Kd5 . <8> 72. Kf6 Black mates in 48 after 72…Kh3. After the played 72…Qe3, Black also mates in 48.

<9> 64...Ke5 Draw after 64.Qd6+. Also draw after the played 64…Kd5.

<10> 69...Qf3 Draw after 70.Qg1. But after the played 70,Qc1+, Black mates in 52.

<11> 68...Kf4 Draw after 69.Qb2. But after the played 68.Qf1, Black mates in 82.

<12> 69.<Qf1+> Black mates in 83 after 69…Kg5. But after the played 69…Qf3, it's a Draw.

<13> 70.<Qc1+ > Black mates in 52 after the played 70…Kg4.

<14> 70...Kg4 Black mates in 51 after 71.Qb1. After the played 71.Kg7, Black mates in 49.

<15> 71.Kg7 Black mates in 49 after the played 71…f4.