Underpromotions are interesting and unusual. Despite being a weaker piece, on rare occasions promoting to a knight is better than promoting to a queen (can’t recall when this has happened in a tournament game of mine, except perhaps occasionally in a certain variation that could’ve arisen in a game). Of course, the queen has the combined powers of a rook and bishop. Therefore only in very rare cases promoting to a rook is better than promoting to a queen (and in EXTREMELY rare cases a bishop is the best choice). The reason in such cases will almost always be, to avoid giving stalemate in a winning endgame position. Technically there are positions where you have to underpromote to a bishop or rook to draw a game (by stalemating yourself), but I don’t know if such a thing ever happened in a tournament game.
However, every now and then a given underpromotion can be considered EQUIVALENT to promoting to a queen. Usually, this is when the opponent’s clearly best reply is to capture the promoted piece. So in some cases you can even argue that promoting to a bishop is as good as choosing a queen. I believe that if an under-promotion like this is made, it is to show a certain sense of humor. For example, I think there have been a few famous cases of humorous underpromotions by Ivanchuk (Ivanchuk-Topalov, 1996 comes to mind). Ivanchuk is my idol in certain ways, which I’ll go into another time. On various occasions in serious tournament games, out of similar intent I’ve underpromoted, sometimes to a bishop. For instance, my game against GM Mikhalevski from Foxwoods 2014 came to mind: on move 46 I promoted my g-pawn to a bishop (though 5 moves later I was smart enough to promote my a-pawn to a queen, and win!). However, in a recent game, against a GM I took this too far. Let me show you what I mean.
I recently played in the Third Saturday round robin tournament in Djenovici, Montenegro. It was my first time in that Balkan country and the place was rather beautiful. Overall, I did pretty well, after a rough start. In the penultimate round 8 I was white against a veteran Serbian GM who’s very solid. For instance, in this event going into our game he had 4/7 with 1 win and 6 draws. I also had 4/7. Let’s see how the game went:
IM Justin Sarkar (ELO 2339) – GM Bosko Abramovic (ELO 2362) [D51] Third Saturday Djenovici (8), 22.03.2019
1.d4 Nf6 2.c4 e6 3.Nc3 d5 No Nimzo today. I actually spent more time considering what to play against the Nimzo-Indian, in my pregame prep. 4.Bg5 Nbd7 5.e3 c6 6.Nf3 h6 This less common move keeps the game in D51, which was a new ECO code for my tournament practice. More another time, on my knowledge of ECO codes and their classifications. Most common is 6…Qa5, which is the Cambridge Springs Variation and ECO code D52. 7.Bh4 g5!? 8.Bg3 Nh5
Virtually unknown before 3 years ago but has become kind of trendy since. I actually had a feeling Abramovic might go for this, in case he chose the QGD/Semi-Slav, as I saw a couple of prior games of his in it (without seeing any other 6th moves he had played, such as the Cambridge Springs, in his big game collection). Again though, I was more focused on the Nimzo-Indian, while trying to briefly consider various other things too. While this was my first time actually facing the 7.Bh4 g5 8.Bg3 Nh5 sideline, I’d looked into it in summer 2017, focusing on an interesting sharp plan for white involving Qc2 and 0-0-0. 9.Qc2 Incidentally the other published Abramovic games featured different 9th moves by white, though I had to assume he was aware of this move/plan. Nxg3 10.hg3 Bg7 11.0-0-0
Interestingly enough this was first reached (via a weird opening move order transposition) in Igor Ivanov-Taimanov, 1975, which white won (black played the dubious 11…Qa5, which has not been repeated) Nb6 This was played in one prior published game, by GM Dreev, sometime in fall 2017. Other moves are 11…a6 and 11…Qe7. 12.Ne5 Bxe5 Black can try playing the immediate 12…Qe7, when I suppose 13.f4 is an idea for white. 13.de5 Qe7
This position was reached in Koziak-Dreev, Bastia 2017. Actually I became aware of this last summer, when preparing for a game as white against GM Petr Haba at the 2018 Pardubice Open. This sideline was one of many things Haba could’ve played (he actually played something sharp and unexpected in another Slav line, yet I won what was one of my better games). While reviewing this line then I noticed the new 11…Nb6 move and win by Dreev. When engine checking for ways white can improve, I stumbled upon something very interesting on this move. Koziak-Dreev, 2017 continued 14.Kb1 Bd7 15.f4 Qc5 and Black eventually won. Perhaps 16.fg5 Qxe3 17.g6 favors white, though it’s not so clear.
Not the king, but the knight! A totally non-human move, which only a computer would play. The shock value in itself makes it worth playing, especially if it might be objectively strong 🙂 It probably deserves a double exclam. I never looked into the particulars to appreciate it, but figured that if Black tried to copy Dreev’s play, this was the move to play. I reviewed my notes just long enough for the Abramovic game, to recall this position (and the Nb1 move!). This move befuddled my opponent. I guess the idea is to play 15.c5 or 15.Nd2. For instance on 14…Bd7 15.c5 is very strong as the knight is forced to c8. How else does black develop? On 14…Qc5 15.Nd2 with a strong initiative. c5? But this runs into another problem 15.Nc3!
Knight back! Now the dual threat 16.cd5 and 16.Nb5 is a real problem. Bd7 On 15…dc4, 16.Ne4 or Nb5 is a major issue. 16.cd5 ed5 17.Nxd5 Nxd5 Maybe when playing 14…c5 he thought he had 17…Qxe5? here, just like I did for a moment or two before making my 15th move. But of course, attacking the queen with 18.f4 pretty much wins on the spot. 18.Rxd5 b6 19.f4 0-0-0 20.Rd6 Be6 21.Ba6+ Kb8 22.Rhd1 Rxd6 23.ed6 Qf6 Black is unlikely to survive for long, but this should just hasten the end. 24.Qe4 Bc8
Just about anything should win here. The cleanest is 25.Qe7! Qxe7 26.de7 Kc7 (I saw this, then decided this wasn’t “simple enough”, conveniently filtering out the move 27.Bb5, winning on the spot): 27.Bb5 1-0 25.d7 Bxa6 26.Qe8+ Kc7 Little did he realize, he just set a trap. On 26…Qd8 27.Qxh8 Qxh8 I’d probably promote to a rook, as here black OBVIOUSLY has little choice but to play 28…Qxd8 anyway 🙂 Then 29.Rxd8+ Kc7 30.Rf8 or Rh8 should win easily.
At first the game score on the chess-results site when it initially got posted that night had ended here as 1-0. This would be a logical ending point, as white wins easily after 27.d8/Q+ Qxd8 and now either 28.Rxd8 Rxe8 29.Rxe8 or 27.Qxf7+ followed by taking the queen. So, you can say I won the first game. But unfortunately, the game score on the site later got updated and the truth had to be told: a second game is just about to begin!
27.d8/B+?? What the ?? Actually the delay in updating the game was caused by the delay the person inputting the game had in figuring out how to make a bishop promotion! Maybe it was set on “Always Queen” or something. Seriously, I just assumed black had to take on d8 and surely the king moves lost quickly. However, my eyes played tricks on me, in a nasty way. This is the “second best” promotion choice here (others would lose, due to not giving check while the e8 queen is hanging), yet it throws away the win. After like a minute’s pause, my opponent played … Kb8! Oops!! He had that one! Now my queen is hanging, so 28.Bxf6+ Rxe8 is just a queen trade, while the tempting “double check” with 28.Bc7+?? Kxc7 29.Rd7+ Kc6 just… loses! That leaves 28.Qd7 when uh-oh black has 28…Qe6 and there’s no knockout blow for me. But in a few minutes I figured I just had to play it anyhow as there’s nothing better. 28.Qd7 Qe6
Only move, but good enough to stay in the game. I had to adjust to the shock that I blew the win and probably had to go into an opposite-color bishop ending where I have at most a tiny edge with a draw being very likely. This underpromotion hallucination was quite a moment. Totally unheard of, just like for instance GM David Navara’s game against GM Moiseenko from the 2011 World Cup. To quote Wikipedia: “After a long struggle, Navara offered a draw in a winning position. Earlier in the game, Navara accidentally touched a piece, but Moiseenko did not insist on the “touch-move” rule that would have lost him the game. Moiseenko was subsequently outplayed by the Czech GM, who with a forced mate on the board offered him a draw”. While of course this situation is different and unrelated, it is an unheard of occurrence of comparable magnitude. Or here, unheard of stupidity by me. Well maybe you can joke that my promotion to a bishop was a desire to when on the brink of victory, give my opponent a draw! Let’s see what happened, and who if either of us adjusted better to the shock: 29.Bc7+ I figured this was relatively best, though took awhile to come to terms with it Ka8 30.Qxe6 Similar story here. 30.Be5? is just asking for trouble after Qc4+ when White’s king is at least as weak and only Black can be better. fe6
White has many moves, but the opposite-color bishop endgame is pretty equal. I spent awhile trying to decide whether or not to even still try to win, but just couldn’t quite get myself to give up on such hopes yet. 31.e4 a slightly desperate attempt to keep the game going, by creating a passed pawn with f5 next. Bb7 32.f5 Bxe4 Grabbing a pawn with 32…ef5 33.ef5 Bxg2 is also fully playable. White’s compensation for the pawn should be enough only for a draw. I’ll leave it up to you to verify this by testing a few lines if you wish. 33.f6 33.fe6 Re8 forces white to trade rooks to save the e-pawn, when black has the “better side of a draw” in a pure ocb. Whereas, 33.Be5 followed by 34.f6 is similar to the game with 33.f6 and slightly more accurate, forcing the rook to immediately commit to either h7 or f8. Rf8 One of a few plausible moves. Probably 33…Rh7 is a bit more accurate, guarding the 7th rank. 34.Be5
Bd5? Already a big mistake. Rather than try to block the d-file the bishop should go backwards… on the other diagonal with 34…Bg6! when 35.Rd6 Kb7, 35.Rd7 Rf7, or 35.Rh1 h5 should hold the balance without undue trouble. 35.Rh1 Now the rook penetrates on the h-file, while gobbling a key pawn. Bxg2 Black ought to try 35…Bxa2, with 2 extra queenside pawns, even though after 36.Rxh6 followed by soon picking up the g-pawn white’s pawns should be much stronger. 36.Rxh6 Be4 37.g4 I suppose I played this first to not allow 37.Rh5 g4, though 38.Rg5 picks up the pawn as an attempt to defend it with the bishop loses quickly to 39.Rg7 followed by 40.f7. Kb7 38.Rh5 Kc6 39.Rxg5
Rf7 39…Bf3 with idea to play …Kd5 was a better try. 40.Kd2 b5 41.Ke3 Bb1
42.Kf4 Felt no need to move the a-pawn, as 42…Bxa2 43.Rg7 Rf8 44.g5 would win very quickly. Rh7 43.Rg8 Rh1 44.a3 Yet for some reason here I chose to move it, just in case. 44.Kg5 is more direct and to the point. Rf1+
45.Kg3 I talked myself into playing this for the wrong reasons, whereas on the natural and stronger 45.Kg5 Kd5 just about any bishop move should win easily. Kd5 46.Rg5?! a bit fancy/clumsy, although still winning. I didn’t want to have to calculate 46.Bf4 e5 47.f7 Rxf4 48.f8/Q Rxf8 49.Rxf8 even though I knew it should win, perhaps rather easily. I felt there was room to miscalculate, as black tries to create counterplay with the active king and queenside majority without white immediately being able to keep pushing the g-pawn. I recalled a lucky save I had last summer, with black in a similar type of exchange down ending. Maybe that had a psychological impact and talked me out of playing this line. Ke4? This was also the main thing I saw, when deciding on 46.Rg5, including the remaining moves played in this game. However, shortly after playing my move I realized black had a better try in 46…Be4! Already I was slightly regretting Kg3, followed by Rg5, and it was even bothering me that I hadn’t yet worked out a clear win, though 47.Bc3+ followed by probably 48.Rg8 should still be winning. 47.Bc3 b4 48.ab4 cb4 49.Re5+ Kd3 50.Re1! Black resigns 1-0
And now, game 2 has been won. The first game ended on move 26 and a second game began on move 27. This was a hard lesson for me, regarding being overly cute or fancy, maybe slightly alleviated by the final outcome. Hope you enjoyed seeing how it’s possible to win two games in one – against a GM. Just don’t promote to a bishop, like me!