winning endgame for White

This endgame is from a recent Live game of mine here at chess.com. According to the ~2500-rated computer analysis there are no mistakes, inaccuracies or blunders in the portion of the game shown here. There may be other moves that will win more quickly but I am pleased with this particular endgame not only because I (playing as White) won and because, as I said, there are no mistakes etc., but because, at the beginning of the problem here - according to the computer analysis - Black is better by ~0.8. I'm posting it as a problem although I don't really expect anyone to find all 22 moves. I've added remarks after a few of the moves. You can see them by clicking on "solution" and manually advancing.

On a number of different moves black needed to try Kd7 to cut your rook off.  I didn't like 14.Kxh5 but it's probably not bad.  Playing Rc7 before a black Kd7 is a consideration.

f5 was terrible, checking your king and forcing it into his unprotected pawns was also silly :)

I think I would have gone for 7...Rb3+ and 8...Kf7.  Chosing to give up the g pawn doomed the rest of his kingside pawns

Good job though, you turned a bad situation into a winning one :)

There are actually plenty of mistakes in that game fragment.

• 2...Rd5+ is an error since the rook is awkwardly placed after Ke4. 2...g5 is a better idea, controlling the f4- and h4-squares so White can't push a pawn to either of those squares.
• 4...Rb5 is a strange decision. Why does Black concede the d-file to White? There can only be one reason...
• 6...f5+ is just terrible. Black makes a critical weakness on g6 and gives White a protected passed pawn for no good reason. 6...Rc5 defending the c-pawn and 6...Rb2 at least justifying ...Rb5 by going for an exchange of c- for h-pawn are the most sensible moves.
• 7...Rc5 is the final losing mistake, but by now Black's entire position is grim because of that one errant pawn push. 7...Kf7 preparing ...Rb2 gives Black more chances.

Nicely played endgame. I'm pretty sure black could have won from the original position though, at worst it would draw.... he needs to brush up on his technique!

orangehonda - by capturing with the king, the rook remains free to move laterally. I like it like that.

khpa21 - you raise a point that I've thought about before; namely, if my opponent and I submit the same game for analysis, does the analysis indicate to him and not to me mistakes and inaccuracies he's made? I know the analysis will tell me when my opponent blundered, but I've never known the analysis to tell me when my opponent moved inaccurately or made a mistake (I think I just answered my own question.). Interesting that the ... f5+ is so unanimously disliked by us but not considered a blunder by the computer analysis.

samredway - thanks; since a year ago when I read Silman's Complete Endgame Course and a few other chess book I've improved my endgame quite a bit. Still working on tactics & overall game.

Well, it's because f5 is a "see a check give a check move" and hopes you fall for losing your rook.  Also it simply drops a pawn for no reason.  On my computer I'm getting about -0.60 (edge for black) before f5 and 0.00 (drawn) afterwards.

For move 14 I don't like capturing the pawn at all :)  The king's activity is valuable as always and the h pawn is more or less worthless, capturing is like passing on your turn, I like Kg5 instead.

khpa21 the 2 ... g5 push looks good at first but the white king and rook then will be able to get to Black's f, g and h pawns before Black has time to bring his king and rook to their defense.

I agree that 4 ... Rb5 looks strange. However, if Black is to advance his c-pawn he must first put his rook at some distance from the enemy king.

I think, instead of the 6 ... f5+ push, Black could try what seems a logical follow-up to 4 ... Rb5: 6 ... c5. This gives White more to defend against, brings his rook and pawn into better relation with each other, and Black still has the ... f5 push available.

I'm surprised that the computer analysis didn't indicate a blunder at 7 ... Rc5. Black could instead protect his g-pawn by 7 ... Kf7 and still have the rook free to move to any of the ranks in his own territory. Both those opportunities are lost to him after 7 ... Rc5.

orangehonda I've lost too many games by not capturing a passed pawn when I could. No way I'm going to let that one slide past and by doing so tie my rook to the h-file.

It wasn't technically passed yet, although I can  understand abandoning your own h pawn and black may get a 2nd passer soon is a bit scary especially with a clock running.  I've lost endgames by trying to play like Capa and ignoring material for a winning position... and along those lines 14.Kxh5 is a good practical move, leaving black nothing to hope for.

yes, with enemy rook on c2 it was scary to leave the black pawn on the h-file, even with my rook already on the h-file behind that black pawn and my own h-pawn blocking it.