[Event "Superbet Chess Classic Romania"] [Site "Bucharest, Romania"] [Date "2023.05.07"] [Round "2.1"] [White "Rapport, Richard"] [Black "Duda, Jan-Krzysztof"] [Result "1-0"] [WhiteElo "2745"] [BlackElo "2724"] [Annotator "Lang, JJ"] [Variant "From Position"] [ECO "?"] [Opening "?"] [FEN "8/5p2/1p2k3/5p2/1PPbPP1p/1K5B/7P/8 w - - 0 45"] [SetUp "1"] [Source "https://lichess.org/study/gLTaYxXi/WldVAaCL"] [Orientation "white"] { [#] } 45. Bxf5+ { At first, I was disgusted when I saw the engine claim Black has only one move that holds here. I thought it would be one of those highly complicated computer variations that make no sense. But, on reflection, Black must play 45. ... Ke7 because he cannot allow White to play e4-e5 with check, as then the pawn reaches the e6-square much faster. Once this happens, White's f-pawn becomes passed, allowing White to play for the "principle of two weaknesses" by tying Black down to guarding the f-pawn and going after the queenside. } 45... Kd6?? (45... Kf6?? 46. Ka4 Be3 47. Kb5 Bd4 (47... Bxf4 48. Kxb6 $18) 48. Kc6 Bg1 49. h3 Be3 50. Bg4 Bd4 51. e5+! Ke7 52. c5 bxc5 53. bxc5 Kd8 54. Kd6 $18) (45... Ke7! 46. e5 Be3! 47. Ka4 { Note that, if White had the one extra tempo that would come from e4-e5 coming with check, White could now play Ka4-b5 and stop Black from capturing on f4. } 47... Bxf4 48. Kb5 Be3 49. Kc6 Bg1) 46. e5+ Kc7 (46... Ke7 47. Ka4 { see notes to 45. ... Kf6??. }) 47. e6 fxe6 48. Bxe6 Kd6 49. Bg4 h3 50. Bxh3 Bg1 51. f5 Bxh2 52. f6 { Black's king must stay on d6, as to meet f6-f7 with ...Kd6-e7. But White's king will come around to b5, advancing the c-pawn, and Black's king will be overloaded. Black resigned. } 1-0