Another endgame
Earlier in the season I'd been a little lucky to agree a draw in a rook ending, with both players failing to notice that the pawn ending they were about to enter was won for Opponent. In this game I actually went into the losing pawn endgame - and was again lucky enough to get away with it. I strongly suspect that all of the last four moves have been half-point blunders and in this position 33. a4? almost certainly is (I think that 33. axb4 holds). After 33. ... Rxd5 34. Rxd5+ Kxd5 35. h4 ... Opponent played 35 ... gxh4?, missing a win that I'm tempted to describe as "study-like". Correct is 35. ... c4 36. bxc4+ Ke6! In the game position with the g-pawns exchanged, 37. ... Ke6? loses easily to 38. Ke3, when white has the b-pawn covered and black cannot stop both the c-pawn and the h-pawn. However with the g-pawns still on the board, the line goes 37. Ke3 b3 38. Kd2 b2 39. Kc2 f4! and black wins.