Please turn on javascript in your browser to play chess.
Only Chess Forum

Only Chess Forum

  1. Standard member thesonofsaul
    King of the Ashes
    30 Mar '09 14:39
    I've reached the point in my game where I wish to take up serious endgame study. Yet, as I sit here going over positions in an endgame book, I feel that I am not getting anywhere. I understand the situations before me, but I do not believe that I would recall them well enough in a pressure real time game situation. What I need is a way to practice endgames so I can fix them in my mind.

    If anyone has any suggestions for endgame study it would be much appreciated. Also, if there is already a thread discussing this someone could just link to that and save everyone time and trouble.
  2. 30 Mar '09 15:05
    If your serious about it you could use a program such as fritz, allows you to set up king and pawn, king and rook etc etc endings. With alot of endgame positions the wrong move at the wrong time can kill it for you. So at least with a pogram you can easily see where you went wrong and it is easy to go back and reset the position than over the board.

    Chessbase also have a 4 DVD set about endgames, loads of information in them however I just felt the guy moved a little too fast at times. But if your serious it may be worth a look.
  3. Standard member wormwood
    If Theres Hell Below
    30 Mar '09 16:34 / 2 edits
    Originally posted by thesonofsaul
    I've reached the point in my game where I wish to take up serious endgame study. Yet, as I sit here going over positions in an endgame book, I feel that I am not getting anywhere. I understand the situations before me, but I do not believe that I would recall them well enough in a pressure real time game situation. What I need is a way to practice end ...[text shortened]... dy a thread discussing this someone could just link to that and save everyone time and trouble.
    yeah, I also agree that the best way is to drill the positions against an engine hundreds of times, until it becomes automatic and requires absolutely no conscious thinking whatsoever. simply learning about the endgame isn't even nearly enough. that's only 1% of the task.

    the remaining 99% is drilling it so many times you can't get it wrong.

    for example, I've drilled KQ vs kr so many times that it just took me about 30s to mate an engine from the following random position:



    1. Ke2 Rb5 2. Qd3 Re5+ 3. Kf3 Kg5 4. Kg3 Re6 5. Qd5+ Kf6 6. Kf4 Ke7 7. Kf5 Rf6+ 8. Ke5 Rh6 9. Qb7+ Kd8 10. Qf7 Rc6 11. Kd5 Rb6 12. Qf4 Kd7 13. Qa4+ Kc7 14. Qa7+ Rb7 15. Qc5+ Kb8 16. Kd6 Rg7 17. Qe5 Rc7 18. Qf4 Kc8 19. Qf5+ Kb8 20. Qe5 Rb7 21. Kc6+ Ka7 22. Qa1+ Kb8 23. Qa5 Rb1 24. Qe5+ Ka7 25. Qd4+ Kb8 26. Qh8+ Ka7 27. Qh7+ Kb8 28. Qxb1+ Kc8 29. Qb7+ Kd8 30. Qd7# 1-0


    I could do it in my sleep, and after 9...Kd8 I moved as fast as I could move the mouse, absolutely no thinking required. that's the way how you remember it in a real game, months or years after reading about it.

    that's the kind of level to which you need to drill every endgame position, and it will give you the ability to see those positions coming a mile away. book knowledge is insufficient for chess.
  4. Standard member wormwood
    If Theres Hell Below
    30 Mar '09 16:42
    Originally posted by Audacious
    Chessbase also have a 4 DVD set about endgames, loads of information in them however I just felt the guy moved a little too fast at times. But if your serious it may be worth a look.
    if you mean the karsten müller dvds, they're excellent. but you need to use 'pause' a lot because he doesn't wait for you. I copied all the starting positions into a database, and loaded them into fritz endgame training module, so that drilling become very easy.
  5. 30 Mar '09 17:39
    A good post Wormwood - I applaud your dedication and agree with
    you about making certian parts of the game as easy as breathing.

    But (there is always a but).
    And this is not a knock. Don't take it the wrong way.
    it's a genuine concern.

    Can you do it OTB in an Allegro finsih in the last round of a
    tough weekend congress with a crowd hovering around your board,
    2 minutes left on your clock v a human who will not make the best move
    (as a box does) but play a move that maybe wrong but 'tricky'.

    Your opponent slaps the clock, the crowd is murmouring, a piece
    gets knock over, if you make an illegal move you lose, your
    clocking is ticking, winner gets £200 loser gets £25, your clock
    is ticking, he plays a move, it's a stalemate trap, your clock is ticking....

    I've seen that ending blown so many times in the last round OTB
    by some very experienced players.

    I'm not kidding about playing a human. They are not computers.
    They cheat.

    In a position very close to this, I saw White win in a blitz finish.



    How?

    1.Ra7+ Kg6
    2.Ra6+ Kg5
    3.Ra4 Kg4 (he was expecting 3.Ra5+)
    4. Rxg4 1-0

    Try it out at your club OTB with 5 minutes on the clock. You will see
    a big difference and doing it this way may just win you £200.

    (ps. I had to do it once in an allegro finsih - I dd it, not following
    any routine, I kept setting Rook winning traps and one finally worked).
  6. Standard member wormwood
    If Theres Hell Below
    30 Mar '09 18:28
    Originally posted by greenpawn34
    A good post Wormwood - I applaud your dedication and agree with
    you about making certian parts of the game as easy as breathing.

    But (there is always a but).
    And this is not a knock. Don't take it the wrong way.
    it's a genuine concern.

    Can you do it [b]OTB
    in an Allegro finsih in the last round of a
    tough weekend congress with a crowd hov ...[text shortened]... t, not following
    any routine, I kept setting Rook winning traps and one finally worked).[/b]
    I don't play OTB, but I'm fairly confident I couldn't fail this particular ending unless extremely low on time. 5min would easily suffice. in your position the rook will drop very soon as it's cut from its queen and can't regain any of the Real defences. the hard part is already done there.

    here's a typical way:



    1. Ra7+ Kg6 2. Ra6+ Kg5 3. Ra4 Qb5 4. Ra3 Kg4 5. Ra1 Qd3 6. Kg2 Qc2+ 7. Kf1 Qd3+ 8. Kf2 Qd4+ 9. Ke2

    playing it against a human is harder though, I know exactly what you mean. that's because you need to actually think a little, as the engine skips all the most ridiculous moves. but you also get much more time as he won't move immediately like the machines will.
  7. 30 Mar '09 18:44
    I'm impressed, if only because I'm not exactly confident about my own KQ vs KR endgame. Here's a favorite of mine:




    Two knights vs Pawn
  8. 30 Mar '09 19:08
    Originally posted by greenpawn34
    A good post Wormwood - I applaud your dedication and agree with
    you about making certian parts of the game as easy as breathing.

    But (there is always a but).
    And this is not a knock. Don't take it the wrong way.
    it's a genuine concern.

    Can you do it [b]OTB
    in an Allegro finsih in the last round of a
    tough weekend congress with a crowd hov ...[text shortened]... t, not following
    any routine, I kept setting Rook winning traps and one finally worked).[/b]
    You're right about playing v's a computor that always makes the best move. I once learned to checkmate with knight and bishop v's king - a mating net of great beauty. But because the computer always played the best moves this actually made it easier as you could learn the sequence.

    Once you've corralled the enemy king into a corner of the opposite colour to the bishop then it all follows a long pattern to check mate.

    It can be much harder playing someone who makes not such good moves with a shorter but unfamiliar pattern to mate.

    Maybe you could swing them back onto your chosen path but with a long sequence of 30 moves or so - if you slip and have to start again then you fall foul of the 50 move rule.
  9. Standard member wormwood
    If Theres Hell Below
    30 Mar '09 19:44
    Originally posted by Mahout
    You're right about playing v's a computor that always makes the best move. I once learned to checkmate with knight and bishop v's king - a mating net of great beauty. But because the computer always played the best moves this actually made it easier as you could learn the sequence.

    Once you've corralled the enemy king into a corner of the opposite colou ...[text shortened]... 30 moves or so - if you slip and have to start again then you fall foul of the 50 move rule.
    I never try the shorter mates even if my opponent grants the possibility. it's just much more practical to force him into a set starting position, after which there's no unfamiliar sidelines (in KNB).

    in KQkr it's a bit more tricky, as there are MANY ways to go. but there are still a lot of basic techniques which you can learn and apply as needed. there's maybe 5 or 6 basic positions which you need to know inside out, and the rest are variations of these themes. you also need to be able to recognize and switch from one type to another on fly, as that's almost invariably what happens when the opponent does a 'weird move'.

    but at some point it all becomes instinctive, and you'll just observe your hand making moves you didn't consciously think about. then you start drilling the next thing.
  10. 30 Mar '09 19:59
    Tbh I don't think it really matters that an engine will always make the right move compared to that of a human when trying to study endgames. In the chessbase DVD's I mentioned before they explain certain principles that arise depending on the position of the pieces.

    Is it a Rook's pawn, has pawn passed the X rank on the board, taking the opposition, understanding the key squares in the position etc etc. To me if you plan to study it properly, it's analyzing the position and understanding the principles of the endgame technique that wins the so called 'winning position'

    So regardless if the opponent is human or engine, if you understand the principles of how to win King & Rook vs King it shouldn't matter if the opponent plays the best or worse moves in the position because the technique used to win the position is the same.
  11. 30 Mar '09 20:15
    Originally posted by Audacious
    So regardless if the opponent is human or engine, if you understand the principles of how to win King & Rook vs King it shouldn't matter if the opponent plays the best or worse moves in the position because the technique used to win the position is the same.
    I don't think it's as simple as that. Often there are multiple ways to try to defend an endgame and an engine may choose defences that a human won't, and vice versa.

    For example, in KQ vs KR, often the KR have to stay together in order to avoid the rook being lost to a series of checks and then a fork. But such sequences are not always obvious to a human even if the engine knows them perfectly. So maybe we have to practice spotting these "human" variations too.
  12. Standard member thesonofsaul
    King of the Ashes
    30 Mar '09 20:19
    I agree that it is the principles that I am after, not how fast I can preform an endgame win. I want to be able to recognise won and drawn positions thereby giving me more options during the game, that is, more goals to strive for other than material advantage and/or checkmate. If I know a position is winning and I know the idea behind it, then I can strive for that position.

    Unfortunately, I'm not in a financial situation in which I can shell out the cash for the software. Is there anything on the internet or not old fashioned study stratagies I could use while I save up my pennies?
  13. 30 Mar '09 20:19
    chesstempo.com has tactics training as well as endgame training.

    You can solve the endgames in "Theory" mode(always best move made)

    Or solve the endgames in "Practice" mode(make any move as long as it still leads to a win)

    But...it doesn't explain the ending or tell you the "why" of the endgame.
  14. 30 Mar '09 20:21
    Originally posted by Varenka
    I don't think it's as simple as that. Often there are multiple ways to try to defend an endgame and an engine may choose defences that a human won't, and vice versa.

    For example, in KQ vs KR, often the KR have to stay together in order to avoid the rook being lost to a series of checks and then a fork. But such sequences are not always obvious to a huma ...[text shortened]... knows them perfectly. So maybe we have to practice spotting these "human" variations too.
    The point is when you get to these 'winning' scenario's, the fact is with correct play from the white/black the game is won regardless of what move your opp makes. It's based on you making the correct move.

    So if you understand the principle off how to win to me it shouldn't matter if the opp makes the best or worst move.
  15. Standard member wormwood
    If Theres Hell Below
    30 Mar '09 20:35
    Originally posted by Audacious
    The point is when you get to these 'winning' scenario's, the fact is with correct play from the white/black the game is won regardless of what move your opp makes. It's based on you making the correct move.

    So if you understand the principle off how to win to me it shouldn't matter if the opp makes the best or worst move.
    the first time I got KNB in blitz, I had 25 seconds which should've been enough. but because the human opponent made weird moves, I fumbled about 30 moves before I finally got the opponent into the 'starting corner' with 3 seconds to go.

    about 15 moves later I timed out 1 move (I think) before the mate.


    that's what easily happens against a human, no matter how well you have prepared against engine.