What is the most difficult winnable endgame to learn?

It has been for a while to facilitate the 2N vs P ending but they retracted it after discovering many new 50+ endgames by tablebase. It sort of fits our fast paced times to abort the sea snakes which endanger tournament schedules. They'd rather choose for clock abortion with its unrestricted blitz endings. Something is to be said for converting the 50-move endgame rule into e.g. a 100-move game rule which guarantees a fixed time per move for the first 100 moves in a standard game followed by some variable form of blitz playout.

By the way, the 100-move rule never made it into any of the official versions of the FIDE laws (that is I can't find it) so it must have been an added directive for tournament arbitration or something. But it was in use for a short while.

bishop and knight vs king, or maybe 2 bishops vs king and knight. Possibly king and queen vs rook and queen, if defended perfectly

Queen vs Rook is hard if you play it against the tablebases. I guess it's not so difficult against a human since no one can defend it perfectly.

K+B+N vs K is not that easy, but not that difficult either. It rarely happens in practice (just like K+B+B vs K or K+N+N vs K+P) because the material advantage is immense and the opponent probably resigned 10 moves before (when there were pawns on the Board).

K+B+B vs K+N is extremely difficult, and I don't think you can learn it.

Here is an example from my old book from the 60s by Andre Cheron.

A difficult endgame to learn imo is R+h+f-pawn vs R

What's funny is I find King + Knight + Bishop endgames easier than King + 2 bishops. Much more subtle manuvres required in the latter. Once you master it maybe it is just as easy. I can usually win knight and bishop vs stockfish within 37-40 moves starting from a random position. KNN vs KP I only know the basic forced mate sequence from the edge. No idea how to do it from random positions, yet Hikaru did it in a blitz game.

King + 2 Bishops + Knight vs King + 1 Rook is a tablebase win but I believe it is extremely hard.

K + 2 Knights + Bishop vs King + 1 Rook is also a win but much harder.

K+2 knights+bishop vs rook is usually a draw since all the defender has to do is sacrifice their rook for the bishop(since two knights vs king is a draw)

https://en.wikipedia.org/wiki/Pawnless_chess_endgame#Rooks_and_minor_pieces 

Well yes because with 2 bishops + knight the rook can't sac for anything as the other 2 will always be able to win. You need at least a knight and bishop or 2 bishops.

..

Please give me an example of a 500 move endgame situation. I think if you cant get it done in under 150 moves it is not possible unless your opponent falls asleep and slips into a coma.

https://rocketcheckmate.blogspot.com/2014/06/mate-in-549.html?m=1

Over 500 moves with perfect play by both sides.

FIDE is ignorant. Hypocrisy in its essence. 50 moves rule should be excluded for every solved position that needs more. Otherwise it is blatantly ignorant.

It is as simple as add an article to allow more moves if table base says so. Arbiter should be allowed to check the table base if a player requests and grant him with AT LEAST table base number of moves (if not more? i.e. TB number of moves x2). It only takes few clicks by the arbiter to check.

SmyslovFan wrote:

Of all the basic mates, K+Q vs K+R seems to be the most difficult. Even World Champion Candidates get this one wrong

............................

LOL. Yah, this one is brutal. Especially against Stockfish. Requires months of study. Hardly practical. I tried to learn it, based on a scattering of sources, but eventually gave up.

Not sure that the title shouldn't be, "What is the most difficult learnable endgame to win".

All positions are endgames, but some have more material than others.

Note that all the posts refer to learnable endgames. Nobody has mentioned, for example, KNNNvKRBP, nor KQRRBBNNPPPPPPPPvKQRRBBNNPPPPPPPP.

Very much doubt that is remotely true.

Not even true for up to 7 men. Longest mate is then 549 moves, but it's not pawnless. (It takes a few moves to turn into a pawnless endgame.)

Depends on what counts as a basic mate.

Fine included KNNvKP. Grandmasters occasionally get KQvKR wrong; they routinely get KNNvKP wrong.

The starting position with the pawn is 549 and in a few moves there's a trivial promotion, might be to a knight actually, and mate in 545 with just the pieces.

Correct.

But it's not a mate in 545 under competition rules; it's a draw.

Endgames don't stop at 7 men just because the tablebases (used to) go that high. Expect evidence that the basic rules mates (and possibly the competitition rules mates) go much deeper as further tablebases are produced.

Eh I agree they should give them the tablebase number of moves rounded up, so like 600 moves for that endgame. It's not like both people are going to play perfectly so one guy might still win on move 75 or 100 anyway, in a 500 move endgame.My first thread on this site was about abolishing the 50 move rule. If they insisted on a fixed number for all endgames no matter what, I would say just double it to 100. It should be 100 full moves (200 total piece movements), not 50 full moves and only 100 "plies".

I would guess that the 50 move rule was first arrived at by looking at KBNvK and adding 50% for good luck.

No current tablebases give you the number of moves required in the rule if there are winning conversions and pawns. The number of moves to mate is not the minimum number of moves required in the rule, which refers to the number of moves between pawn moves or captures.