lichess.org
Donate

Endgame training

Hi, i have an idea that probably has been already suggested in the forum, but i couldn't find it, so here it is. It would be nice if lichess also had endgame training, maybe using local stockfish.

Anyway, this is the best chess site ever!!, keep up with the awesome job!
The endgame training that works for me is using stockfish to analysis my endgames. I'm always doing something wrong so I'm learning a lot.

"this is the best chess site ever!!" We agree.
@Asius

You can always do yourself an endgame training. Just setup the position you want to train, press menu then press continue from here, then choose play with the machine.

Or take some ending study position like here, and there are others.
en.lichess.org/study/ktY9CFCa
Select chapter, press menu, then continue from here and so on.

Or browse your own games and select position you want to continue to train with the machine.
#2 what's definitely wrong with your approach (even though you think it works for you) is that computer analysis does not help you develop endgame techinque, which is all endgame based around really.

There are many nuances in all endgames including even pawn-only endgames, which may make your past experience-through-computer-analysis in seemingly similar position misleading. The only way to get better at endings is: having knowledge of as many different endgames as possible (things you must be looking for, must be aware of, etc...), and an ability to calculate deeply and to accurately estimate the position in end of a variation.

In endgame practice, I think stockfish might have been used for a bit as a vis-a-vis to test your technique, but due to its ability to play for opponents mistake it might turn into rather pathetic excercises on one hand, and very good excercises on another. Qualified human opponent is a lot better (or excercise made by one).

TL;DR
Stockfish isn't very helpful in endgame development. Using it as a vis-a-vis to play endgame positions might be helpful to a degree, but nothing beats a knowledgable human opponent (or excercise made by such).
@BobC, yes, that's what i do too, sometimes. But i feel it would be nice for new people to have that already easily accesible on the site. For example, mate with 2 bishops, bishop and knight, and so forth.

@wasilix, i don't agree on everything. I think using stockfish to help you develop your endgame it's a good idea. In my case, that approach has helped me a lot, it was my weakest part of the game, and since i started using stockfish to understand some endgame positions, i think im a lot better. But is true that because sometimes the suggested move by stockfish isn't the best one (specially local stockfish), and it maybe misleading, nothing beats a human prepared exercise. Anyway, i think local stockfish will do a decent job
@Asius
Speaking of knight + bishop endgame for instance. Stockfish isn't gonna tell student about how one should force opponent king to a board edge at first, and then to an appropriate corner. These things though aren't very hard to understand by oneself. If only endgames were all about mating a naked king though. In a rook vs rook + pawn endgame, stockfish will not help revealing what's position of Philidor or Lucena position... In a lost endgame rook + pawn vs queen stockfish will not help understand that you might've had a draw if only you knew how the draw can be achieved.

I'm not a strong player, but don't let this mislead you. I've been coached by a very strong ukrainian GM for a year in past, and I can assure you endgame it is a super complex topic, and stockfish may only be useful in very specific cases, that are generally simple anyway.
#4 I agree that Stockfish would provide useful evaluations if it had at least 5-man tablebases (so it would provide correct evaluations in addition to correct moves).
I would distinguish between endgame training and endgame learning . For learning is important the human factor , where an instructor verbalize explanations . To test learning by training, I think it might be valid to play against the machine.

Here you have a well organized catalog of positions for endgame training: es.lichess.org/forum/lichess-feedback/endgame-training-3?page=1
#8
It might be valid indeed, but still. From a practical point in any non-trivial endgame there are numerous ways of how losing side (or a side trying to squize a win from a draw especially) can make your job harder and set traps for you. Stockfish won't do this. Also if there are practical chances for sf to squize a win from a draw, it might just mindlessly move pieces around because "it's a draw". Don't let me get into it, I've had some pretty bad experience here unfortunately, engines are often disappointing in such cases. Human on another hand will try his practical chances, set traps for you, which might lead to your faster success, but might as well lead to your demise. That's how one fights odds in endgames especially, and I'm not convinced engines are capable of it.
#9 Stockfish will play challenging (though not "trappy") moves if tablebases are installed, otherwise it plays "mindless" moves as you suggest (although technically the position is drawn anyway).

This topic has been archived and can no longer be replied to.