At last week’s U.S. Amateur Team West tournament, one game stood out for me as the most important. It was the fourth round of the tournament, and the Kolty Club team went into the round with two wins and a surprise defeat in round three. We were paired against the Caltech team, and from the beginning things went wrong. On first board, Mike Splane lost his second game in a row. On second board, Juande Perea agreed to a fairly early draw. On third board, my game seesawed; it looked quite good for me for a while, but my opponent handled his time pressure better than I did and by the time control the position looked drawish.

Then there was fourth board, where our player, Marshall Polaris, was playing White against Dillon Holder of Caltech. Marshall lost the exchange in the opening or early middlegame, and his position looked awful. Black’s queen and pieces dominated the board. After a while the time pressure in my game kept me from following his game any more, but I just assumed that Marshall was going down to defeat. That would mean we had only 1/2 point out of 3 games, so I figured that I was only playing for pride because our team had lost the match.

Imagine my surprise, then, when I managed to win a nice rook-and-pawn endgame and Marshall said, “Congratulations! We won the match!” I looked at him in disbelief and asked, “Wait a minute. You won that game?” Thanks to his amazing comeback win, we won the match by 2.5-1.5. That set us up very nicely to win our matches in round five and six, also by 2.5-1.5 margins, and come home with third place in the tournament.

For the last week I’ve been curious how Marshall did it. When I asked him at the time how he won his exchange-down game, he just said, “You’ve got to be tough.” But yesterday I finally had a chance to see the whole game, as we had a team meeting at Paulo Santanna’s house to celebrate, receive our prizes, and look at our best or most interesting games.

I didn’t write down the game and I couldn’t quite reconstruct the whole game this morning, but I do remember the finish, which is the most important part.

White to move.

FEN: 5r1k/pp3p2/2p2Npp/4P3/4R3/5Q1P/q4rPK/8 w – – 0 1

Here’s a short description of how we got to this point. In an exchange Queen’s Gambit, Marshall got careless and played f3 and e4 too quickly, weakening his dark squares. Black’s knight descended on g4 and his bishop on g5, and bad things happened. Black not only won the exchange but was completely in control.

But then a funny thing happened. Black decided the best way to capitalize on his advantage was not to improve the position of his pieces and attack White’s king, but to gobble all of White’s queenside pawns. This gave Marshall time to mobilize his remaining pieces and get some semblance of counterplay. In the above position we see the result: White’s entire queenside has been wiped out and Black has three pawns plus the exchange, but Marshall has gotten his knight to f6 and there is a definite whiff of cheapo in the air.

Black has just played … Rf2, attacking White’s queen, and here Marshall plays a move that says, “I’m not going to be bossed around any more.” He threw down the queen sacrifice, 1. Rh4!

This came as a total shock to his opponent. Years ago, when Jerry Hanken wrote his “Parting with the Lady” column in Chess Life, he said that queen sacrifices — even if they are fairly simple ones like this one — have a huge psychological effect on your opponent. They lead to shock and then panic. That is definitely what happened in this position.

Of course Black cannot take the queen, because 1. … Rxf3?? 2. Rxh6+ Kg7 3. Rh7+ is mate. In fact, Black’s move is basically forced (but also good): 1. … Kg7. White would love to sacrifice his rook here with 2. Rxh6, but Black can simply take it and White’s follow-up is too slow. So Marshall first had to play 2. Qg3, getting his queen in position to “threaten” 3. Rxh6 Kxh6 4. Qh4+.

I put “threaten” in quotes because the whole thing is completely a mirage. White has no real threats. Black can simply play 2. … Rd8, and if 3. Rxh6 then the zwischenzug 3. … Rxg2+! puts the kibosh on White’s attack. 4. Qxg2 Qxg2+ 5. Kxg2 Kxh6 recovers the material for Black and he wins the endgame easily. Or if 4. Kh1 then Black can simply take the queen, because the move 2. … Rd8 created an escape route for his king.

But let’s talk about the curious psychology of lost games. Marshall doesn’t need to worry about whether his moves are objectively winning. There aren’t any winning moves; his position is lost. Instead, he’s trying to get inside his opponent’s head. That’s how you win a lost position. The queen sac and the mate threats have shaken his opponent to his core, and therefore he plays essentially the worst possible move:

2. … Rxf6??

Meaning no disrespect to his opponent, Marshall said that he could tell this move was coming. In Black’s panic, he wants to get rid of what looks like White’s most threatening piece. After all, he’s going to be up four pawns in the endgame, right? What could go wrong?

3. ef+ Kxf6

By the way, there was no chance for Black to bail out with 3. … Kh7, because now the rook sacrifice 4. Rxh6+! is winning.

4. Qc3+ Ke6

At this point, the whole strategy of the game changes. All of a sudden, White’s task is not to construct an ingenious deception out of smoke and mirrors, but to bring the win home with some very precise moves. In fact, it’s a little trickier than you might think, and Marshall spent 15 minutes (!) on his next move.

There is actually only one way to win, and Marshall found it. As he told us yesterday, “The next two moves were my only good moves of the game.”

5. Re4+ …

This is what I would have played too, but you’ve got to follow it up right.

5. … Kd7 6. Qb4! …

Only this move brings home the full point. Black cannot defend both of the invasion squares at e7 and b7, and oh, by the way, his rook is hanging too. Black resigns. (Actually, he may have played on for a couple moves, but this is where Marshall stopped showing us the game.)

A very nice swindle, made even more piquant by the fact that Black really swindled himself, by playing the completely unnecessary move 2. … Rxf6?? To me the psychology of lost games is very interesting, because the goal changes from finding the best moves to finding the most confusing moves, the moves that will cause your opponent to doubt himself and eventually to blunder. An inspiring, tournament-saving game by Marshall!