From pogchamps, specifically semifinals of consolation bracket of pogchamps 3:

[FEN "4k3/7p/5p2/8/P2rr3/5N2/6PP/4R1K1 w - - 0 1"]

enter image description here

Black (Michelle Khare) just blundered a rook, but white (pokimane) failed to capitalise.

White is currently 1 pawn down but is about to be 2 pawns down so it's basically (rook and) knight vs rook (and rook).


  1. How does black win this? I tried running in an engine and even though it gives black an evaluation of around -1, the evaluation slowly went up to -0.4. Or if it's not really a win for black then how does white draw this?

  2. The pawns are on 1 wing. What, if anything, does this mean? Eg more/less likely black will win/white will draw as opposed to pawns on both wings.

Note 1: If this turns out to be a fortress or something, then please feel free to edit the tags.

Note 2: This might be covered in Karsten Müller's endgame series probably Volumes 9 or 12, but I think it was covered a bit in Volumes 1, 2 or 4. What I do remember in Volumes 1-4 that might be relevant is that all the pawns are (going to be) on 1 wing.

Note 3: In case it isn't obvious, I'm talking about serious higher level play. A 1500+ in pogchamps should of course win this easily. So what I really mean to ask is how black wins the position, if possible and not literally how Michelle Khare beats Pokimane like literally explaining to a 600+ how to beat another 600+.

  • 1
    a pawn falls after the echanges. It is tricky to win a rook vs knigth with two pawns each player all in the king's side. Engines for sure win this but after dozens of moves. Oct 8 '21 at 13:03
  • 1
    That eval bar is being generous. This would only be 1 to 1.5 in blacks favour reviewing the position White just needs to play Rb1 or Rc1. Black can't push the rooks , can't mate. the position is better for black but i would say its very much drawn. Oct 8 '21 at 13:55
  • 1
    @BCLC i would estimate that stockfish 14 needs 50+ moves per side to resolve this to the point of a End Game table base could be used. Most engines are just going to dance around with the rooks for a long time. I could see an engine getting a win out of it...though with the help of end game table bases Oct 8 '21 at 14:18
  • 1
    @BCLC that's it. If you have at least one passed pawn and the rook it is easy to win. With the pawns in the same side very difficult to win Oct 8 '21 at 14:42
  • 1
    @Universal_learner ok thanks! do you have a link/reference/source? if you do then you can post as answer...or well you can post as answer anyway without i guess
    – BCLC
    Oct 8 '21 at 18:58

I can only try to give a very generic and partial answer, since details matter extremely. I shamelessly copy stuff from Müller/Lamprecht "Fundamental Chess Endings".

  • Q2 can be answered fully: as always, pawns on different wings means the knight usually loses (being a slower unit than a rook or even a bishop), especially against a rook. Here, they are on one wing, and Black even has a rook pawn, meaning that resacrificing the exchange to win with the last pawn is even harder.
  • As always, the weaker side should try to exchange pawns.
    • But note that this is not the all-cure, in the game Reddmann-Müller :-) (RPP-NPP same wing, BCE 7.15) White (the R party) had even to sacrifice a P to convert into a clear win.
    • This means the knight should not land on a "bad circuit", but try its best to hold a fortress. E.g. RP-NP, same line, usually can be hold.
  • Zugzwang gives N knightmares. (Sorry.) It is thus the best weapon to break pseudo fortresses.
  • Without a P stronghold the N might get chased and ultimately caught.
  • Black must try to penetrate with the king from below. For example, White's Kf2/Nf1/Pg2/Ph3 can only be busted by running in via d3-e2.
  • Black's pawns are split (disconnected), thus annoying counterattacks are possible if Black tries the former.
  • I played blitz with engine suggestions and it seems that Black wins with exactly the suggestions outlined above: White plays h3 and then "nothing". Black protects the bad Ps with the R if necessary, plays a P to h4 (White can never play g3, it frees the f pawn - this also means if White sets up g3-h2 instead of h3-g2, it is even easier to attack by h4), penetrates with the K and keeps the wK encased, and as soon as the bK appears on e3, even f3 is no longer a stronghold for the wN, as Rxf3 always wins.
  • Overall: A win with precise Black play, at least Carlsen should convert without problems. White always has too much "holes" to construct a fortress.
  • thansk Hauke Reddmann! re Karsten Müller, yeah i actually got the pawns on 1 wing idea from h! I mean, such idea is already somewhere in Volumes 1-4 of the endgame series.
    – BCLC
    Oct 9 '21 at 13:43
  • 'Reddmann-Müller :-) (RPP-NPP same wing, BCE 7.15)' --> haha you have a link to this?
    – BCLC
    Oct 9 '21 at 13:47
  • 1
    @BCLC: We were young and we were stupid, it was fine while it lasted...(Both rated ~2300 at the time, he was only 18) No, this is too ancient and not electronically available until I input my about 5000(?) games. Which will not happen to soon since 5000(?) ;-) Thus, you can only find the end of the game in Karsten's endgame book. Oct 10 '21 at 7:40
  • 1
  • @Glorfindel: I hope this doesn't hold for typos, those annoy me to the max :-) Nov 6 '21 at 13:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.