Most players feel that way about endings and agree with David Janowsky when he wrote, “I detest the endgame. A well-played game should be practically decided in the middlegame,” but who are you going to believe? Capabanca wrote, “In order to improve your game, you must study the endgame before everything else…just forget about the openings and spend all that time on the endings.” Pal Benko said he always urges players to study composed problems and endgames. Luke McShane said it's important to develop a good 'feel' for which endings are likely to pose practical problems for the opponent and the knowledge that you know how to handle a particular endgame is invaluable. For the amateur it’s more fun to study openings and tactics but I think most would agree with Capa, Benko and McShane that studying endings is good for us.
When you do decide to study endings the first step is usually to buy one of the popular books but even then, after reading them many players understand almost nothing. Then there exists a boatload of videos, CD’s available. Also, there is the question of exactly what to study. Beginners just need to know basic checkmates and how to promote a pawn. I think some things can be dispensed with though. I remember one player rated 1400 or so who was trying to learn how to mate with a B and N vs. a K. I advised him he was wasting his time because in over 50 years I have never seen this ending arise. If it does in one of my games, I’ll try to recall the procedure for 50 moves then agree to a draw...oh well! Studying K and P or R and P endings would have been more productive for him, but he thought he knew better...he’s still a 1400, but a 1400 that can mate with a B and N vs. a K if it ever comes up in one of his games.
Endings require knowledge of technical positions. You should start with King and Pawn endings and after going over a particular subject, whether from a book, CD or video, you need to put your knowledge to the test. One good way to do this is to use the Shredder Endgame Database. You can set up any ending with 6 pieces or less and then see the results for every move. You can also play out the ending against an engine, but there will be some situations where engines will give a wrong evaluation or they won’t play perfect, but the practice will still be valuable.
Deep Rybka 4, Fritz 12, Komodo 5, Naum 4.2, Stockfish and Houdini 2 all got the only move that wins in this position almost immediately:
1.Kf4 Ke7 2.Kg5 c5 3.Kf5 Kf7 4.f4 c4 5.Ke4 Ke6 6.Kd4 Kf5 7.Ke3 Kf6 8.Kf3 Kf5 9.Ke3 Kf6 0.00/29
1.d4 c6 2.f4 Ke7 3.f5 Kf6 4.Kf4 c5 5.dxc5 dxc5 6.Ke4 c4 7.Kd4 Kxf5 0.00/29
1.f4 Ke6 2.f5+ Kf6 3.d4 c6 4.Kf4 c5 5.dxc5 dxc5 6.Ke4 c4 7.Kd4 Kxf5 0.00/29
1...Ke7 2.Kg6 d5 3.f4 d4 4.Kf5 Kd6 5.Ke4 c5 6.f5 Ke7 7.Kd5 Kf6 8.Kxc5 d3 9.Kd4 Kxf5 10.Kxd3 Ke5 11.Kc4 Kd6 12.Kd4 Kc6 13.Ke5 Kd7 14.d3 Ke7 15.Kd5 Kd7 16.d4 Ke7 17.Kc5 Kd7 18.Kd5 Ke7 5.57/38