How to study chess endgames: the six positions that decide club games
By Nikolas Burk11 min read
Study endgames by learning six positions first: king-and-pawn opposition, the rule of the square, the Lucena and Philidor rook positions, the Vancura draw and the wrong-bishop draw. Each is one idea, not a move list. Play each one out against an engine or tablebase until you can win or hold it without thinking.
Why endgames are where plateaus hide
There is a shape of loss that hides well, and it took me a long time to see it in my own games around 1600 on Lichess: come out of the middlegame a pawn up, or at least comfortable, trade pieces to "simplify", and lose the ending. Not blunder it in one move. Lose it slowly, from a position the engine called 0.00, against someone who knew one thing I didn't.
That is what "equal" means in an endgame: equal if both sides know what they are doing. Your opponent doesn't need to know more than you. You just need to know less. And because endings with few pieces are exact, the gap between knowing and not knowing is a full point.
The usual advice is to buy an endgame manual. Those run to hundreds of positions: the right book for a 2200, the wrong one for you, because at club level the same handful of endings decide the games. King and pawn against king. Rook and pawn against rook. A bishop and a rook pawn that cannot win. You will reach those six, or their close relatives, far more often than anything else in the book.
So this is not a list of the six most beautiful endgames. It is the six that will cost you, or win you, the most games this year. If you are not sure endgames are your problem at all, find out which kind of plateau you are in first; this post is for the player who keeps losing drawn endings and drawing won ones.
Position 1: the opposition
Every king-and-pawn ending reduces to one question: can the attacking king get in front of its pawn and push the defending king off the queening square? The opposition answers it.
Two kings face each other with one square between them. The side that does not have to move has the opposition, and the other side has to give way. Here White wants d6, e6 or f6. With White to move there is no way in: 1.♔︎d5 ♔︎d7 and 1.♔︎f5 ♔︎f7 just repeat the standoff one file over. With Black to move, any king move surrenders a square: 1...♔︎d7 2.♔︎f6, 1...♔︎f7 2.♔︎d6, 1...♔︎e8 2.♔︎e6, and the pawn walks home behind the king.
Once you can read who has the opposition at a glance, your king-and-pawn endings stop being a coin flip. The opposition guide plays through the win from the start, and key squares is the same idea from the pawn's side: the three squares in front of it that the king has to reach.
Position 2: the rule of the square
The second king-and-pawn idea is about the race: a passed pawn runs, the defending king chases, and you need to know in one look whether it gets there.
Draw a square from the pawn to the queening rank, with the pawn's path as one side. If the defending king is inside it, or can step into it on its move, it catches the pawn; if not, the pawn promotes without help. A pawn on its starting rank moves two squares, so measure from b4: the square reaches f8, and the king on h6 is outside it by a file. White doesn't play a single king move.
This saves time in both directions. When your pawn is outside the square, push it and stop thinking. When your opponent's pawn is outside the square of your king, stop chasing and find another plan. The rule of the square guide has the diagonal drawn out.
How do I learn the Lucena and Philidor positions?
Learn them together, as the two halves of one question: in rook and pawn against rook, when does the pawn promote? Rook endings are the ending you reach most often, and nearly all of them turn on whether the attacking king can get in front of its pawn. Lucena is what you do when it has. Philidor is how you stop it.
Position 3: Lucena, the bridge
The white king has reached b8, in front of its pawn, and now it is in the way. Step aside and Black checks from behind for ever on an open board. The bridge is the way out: 1.♖︎d1+ pushes the black king a file further away, 2.♖︎d4 parks the rook on the fourth rank, and only now does the king leave. Black waits, 2...♖︎a1, then checks from behind as the king walks down the board towards the rook, 3.♔︎c7 ♖︎c1+ 4.♔︎b6 ♖︎b1+ 5.♔︎c6 ♖︎c1+ 6.♔︎b5 ♖︎b1+, until 7.♖︎b4 blocks, because the rook that waited on the fourth rank is exactly one rank below the king. The pawn promotes next.
Why the fourth rank? It is the one rank where the rook is out of the black king's reach and still close enough to drop in front of a check on the b-file. Understand that and Lucena is an idea, not a sequence. The Lucena guide has you play every White move yourself.
Position 4: Philidor, the third-rank defence
Now the other side. The white king wants the sixth rank, because from there it escorts the pawn and you get Lucena. Black's rook sits on its own third rank, the sixth from White's side, and simply refuses: 1...♖︎c6, 2...♖︎b6, any waiting move along the rank, and the king cannot cross. White's only way forward is to push the pawn up to the sixth itself, and that is the trigger: with the pawn on e6 the king has nothing left to hide behind, so the rook goes all the way back, ...♖︎b1, and checks from the first rank until White gives up.
Two things to notice: the rook never leaves the sixth rank until the pawn arrives on it, and the defending king stays on the queening file. Leave the rank a move early and the white king slips through. The Philidor guide plays it through from the defending side, including the mistake everyone makes once.
Position 5: the Vancura draw
Rook pawns are different, and this is the position club players get wrong from both sides: the attacker pushes too early, the defender puts the rook on the wrong file.
White's rook sits in front of its own pawn, and the white king wants to reach b6 to take over the job. Black's defence is counterintuitive: the rook goes away from the pawn, to f6, and stays on the sixth rank. Every time the king comes to b5 or b6 the rook checks from the side, 1...♖︎f5+, and the king retreats. The distance is the point: a rook on the b-file is close enough to be chased, exchanged or deflected; from f6 it cannot be, and the pawn can never leave a6 without being lost.
The Vancura guide runs the full cycle of checks, and the rook pawn draw is the king-and-rook-pawn cousin that comes up even more often.
Position 6: the wrong bishop
The last one is not a technique. It is a fact about the board, and it will save you from resigning drawn positions and from trading into "winning" ones that are not.
A bishop and a rook pawn cannot win if the bishop does not control the queening square and the defending king can reach the corner. Here a8 is light, the bishop lives on dark squares, and so the king on a8 can never be evicted: push the pawn to a7 and it is stalemate, hold it back and the king shuttles between a8 and b8 for the rest of the game. The only way Black loses is to leave the corner.
The lesson for your own games is upstream of this position: when you are a piece up and considering a trade into bishop and pawn against king, check the colour of the queening square first. The wrong bishop guide shows every winning try failing, which is the fastest way to believe it.
How should I study chess endgames?
Not by reading. Each of the six fits in two sentences, and you will forget all twelve by Thursday if the knowledge never passes through your hands. This is the routine I would use: ten minutes a day, one position at a time.
Play it out against something that doesn't make mistakes. With seven pieces or fewer on the board that means a tablebase, which holds the exact result of every such position; Lichess runs a free online one (checked 2026-10-03), and any engine will do for the win-or-draw verdict. Play the winning side of Lucena from the diagram until you promote. Play the defending side of Philidor until the engine gives up and repeats.
Hold it three times. A position is learned when you have won or held it three times in a row from the diagram, without a hint, then once from the other side. The other side matters: playing White against the Vancura shows you why the rook has to go to the f-file in a way no explanation does. Miss once and the count starts over. Three clean conversions is the difference between "I've seen this" and "I can do this with thirty seconds on the clock".
One position a week, then rotate. Six weeks covers the list. Then go back to the first, and you will find it takes two minutes instead of ten, which is the point. After that, revisit the set once a month, because the timing is what turns an idea into a reflex you reach for mid-game.
Then look for them in your own games. After each loss, scroll to where the pieces came off and ask which of the six the position was heading towards. This is where the real change happens: not in winning the Lucena when it appears, but in steering towards it ten moves earlier, and away from the wrong-bishop ending on the other side.
None of this moves your rating by a number I can promise. What it does is remove one specific way of losing, which is the only kind of improvement that stays.
What comes next
Once the six are reflexes, three more positions come up often enough to deserve the same treatment:
- Queen against rook, the hardest "winning" ending club players actually reach, and the one most of them draw. It is a king march and a handful of queen forks, and one careless king move throws it away.
- Rook behind the passed pawn, Tarrasch's rule, which decides the rook endings that never reach Lucena or Philidor.
- Connected passed pawns against a rook, the one case where two pawns beat a rook on their own, as long as they advance together.
The full endgame guide index has the rest, from the basic mates to minor-piece endings, each one a single playable idea rather than a chapter.
If you want this done for you
A disclosure first: LearnChess is my app, and I built this part of it for exactly this problem. Everything above can be done with a board, a free tablebase and some discipline; the app removes the setup.
Each of the six positions is an interactive endgame lesson where you play the key side yourself and are corrected when you go wrong, with every move checked against the Syzygy tablebase before it was published. Then the Academy makes you play it out: the position on the board, Knighty on the other side, and a move budget that is the tablebase's own par for that exact position. Hold the Philidor or don't; slip into a lost position and the play-out stops and offers a retry from the slip. It is the "hold it three times" rule with the scoring done for you.
Frequently asked questions

Nikolas Burk
Founder of LearnChess
I plateaued around 1600 on Lichess and built LearnChess to understand my own games more deeply: engine lines in plain English, questions about the board in natural language, and a structured path of lessons, practice and tests. On the blog I share what I learn along the way.
More from the blog
- Tools & comparisons
Chess.com vs Lichess vs Chessable for improving: which one, for what
What each site is actually good for if you want to improve, in one table: playing, puzzles, analysis, openings, explanations. Prices checked November 2026.
12 min read
- Improver's playbook
How to stop blundering in chess: a check that survives the clock
Blunders are a looking problem. A three-question check you can run on every move, two positions to practice it, and why blitz keeps it from sticking.
10 min read
- Improver's playbook
How to analyze your own chess games in 20 minutes (without letting the engine do the thinking)
A three-pass routine for analyzing your own chess games: notes first, engine as grader, one sentence per mistake. For 1200–1800 players short on time.
10 min read