Written by GM Mauricio Flores RiosGrandmaster and author of Chess Structures: A Grandmaster Guide

Endgame

The endgame is the final phase of a chess game, reached when the queens have been traded or so few pieces remain that the kings can leave shelter and join the fight. Material is reduced, direct mating attacks recede, and the game becomes a battle of pawn promotion, king activity, and precise calculation. As a grandmaster, I can tell you where the rating points really live: club players study openings for hundreds of hours and endgames for almost none, and it shows in every tournament hall — half-points are handed away in positions that are objectively drawn or won.

What Is the Endgame?

There is no bell that rings between the middlegame and the endgame, but two practical tests catch nearly every case. First, the material test: queens are off, or only a handful of pieces remain beyond kings and pawns. Second — and this is the test I actually use — the king safety test: the endgame has begun when the kings can walk into the center without being mated. The moment your king stops being a liability and becomes a fighting piece worth roughly four points of activity, you are playing an endgame, whatever the material count says.

This definition matters because it changes what a "good move" is. Queenless middlegames with all four rooks still reward attacking ideas; a pure king-and-pawn ending rewards counting and geometry, nothing else. The closer you get to the pure end of that spectrum, the more the game stops being about general judgment and starts being about exact knowledge: in the opening, five reasonable moves might all keep the balance, while in many endgames one move draws and everything else loses — a difference that is invisible unless you have studied the position type before.

The endgame is also where evaluations become absolute. Middlegame advantages are probabilistic; endgame evaluations are frequently binary — this rook ending is a draw, this pawn ending is lost, full stop. What you learn here is not an opinion but the truth.

Why Endgame Skill Pays the Highest Rating Dividends

Every extra endgame you convert or save is a half-point, and half-points are the currency of rating. Think about how club games are actually won: someone is up a pawn after mutual middlegame errors, pieces come off, and the result is decided by whoever understands the resulting ending better. Over a season, knowing which pawn-down rook endings hold and which extra-pawn positions win is worth dozens of rating points — from knowledge that never goes out of date.

Capablanca's advice — that a student should study the endgame before everything else, because the endgame can be studied on its own while openings and middlegames must be judged by the endings they lead to — remains the best single piece of chess training advice ever given. He was right for a structural reason, not a sentimental one. Endgame knowledge is upstream of everything else: you cannot evaluate a middlegame trade without knowing whether the resulting ending is won or drawn, and you cannot choose an opening plan intelligently without knowing which pawn structures produce good endings. When I wrote about pawn structures in Chess Structures, the evaluation of nearly every plan ultimately rested on an endgame question — which side's pawn structure survives the simplification.

There is also a brutal practical point: the endgame is played when both clocks are lowest and both players are tired. Knowledge that works on autopilot is worth the most exactly there — an opening novelty helps you once; a defensive method saves you for the rest of your life.

The Phase Change: The Rules That Flip

The endgame is not "the middlegame with fewer pieces." Several evaluation rules literally invert, and failing to flip them is the number-one endgame error I see in students' games.

King activity is everything. The piece you spent forty moves hiding becomes your most important attacker. In king-and-pawn endings, king position usually outweighs an extra pawn; in rook endings, the king's ability to reach the action often decides the game. My rule for students: the moment the queens leave the board, ask on every move whether your king can take a step toward the center. Centralizing the king a move too late is one of the most common ways to turn a win into a draw.

Passed pawns rule the board. In the middlegame a passed pawn is a nice asset; in the endgame it is the whole plot. With few pieces left to blockade it, a passed pawn ties down enemy forces or simply runs through — which is why endgame strategy so often reduces to "create a passed pawn, escort it, promote it." Outside passed pawns are especially decisive: they drag the defending king away from everything else.

Pawn weaknesses finally present the bill. An isolated pawn or a crippled majority can be irrelevant in a sharp middlegame, where activity papers over structural sins. In the endgame there is no attack to compensate, and every weak pawn becomes a permanent address where your pieces must stand guard. Many endgames are won by nothing more than attacking a fixed weakness until the defense runs out of pieces to defend it with.

Activity beats material — especially for rooks. The Tarrasch rule — rooks belong behind passed pawns, yours or the opponent's — is the emblem of a broader truth: in rook endings, an active rook is routinely worth a pawn, sometimes two. Grabbing a pawn at the cost of a passive rook is the classic way to lose a drawn rook ending. When in doubt, choose activity.

The Theoretical Endings Roadmap

"Endgame theory" sounds infinite; it is not. A surprisingly small set of exact positions covers the vast majority of what a club player will ever need, and there is a correct order to learn them, because each layer is the foundation of the next.

Stage 1 — the basic mates. King and queen versus king, king and rook versus king, and the two-bishop mate. These basic checkmates are non-negotiable — every other winning plan in chess ends in one of them — and they teach the box-shrinking, waiting-move technique that reappears everywhere.

Stage 2 — king and pawn versus king. The atom of endgame theory. Three tools decide it: the opposition (the king face-off that determines who must give way), key squares (the squares whose occupation by the attacking king guarantees promotion), and the square rule (the instant test of whether a king can catch a running pawn at all). Master these and every pawn ending becomes counting rather than guessing — and, crucially, you finally know which piece trades into a pawn ending are safe.

Stage 3 — rook endings. Rook endings are the most common endings in practical chess, because rooks are the pieces that survive longest. Three positions form the core: the Lucena position, the fundamental win with an extra pawn (the "building a bridge" technique); the Philidor position, the fundamental draw a pawn down; and the Vancura position, the drawing method against a rook pawn. Knowing which target position you are steering for transforms rook endings from a fog into a map.

Stage 4 — minor piece endings. Bishop versus knight, same-colored and opposite-colored bishops, and the famous wrong bishop — the rook-pawn whose promotion square your bishop does not control, turning an extra piece into a draw. This stage is where endgame knowledge starts feeding back into your middlegame: you begin steering trades toward the minor piece ending that favors you.

Stage 5 — the concepts layer. On top of exact positions sit the universal weapons: zugzwang, the situation where every legal move loses and the obligation to move becomes fatal; triangulation, the king maneuver that hands the opponent the move so the zugzwang bites them; and the fortress, the defender's answer — a setup with endless safe waiting moves that no amount of material can break. These three ideas are the grammar in which all precise endgame play is written.

Practical Endgames: Technique When Theory Runs Out

Most real endgames are not theoretical positions — they are messy, multi-pawn middles-of-nowhere that no book covers exactly. This is where technique takes over, and technique is a set of habits, not a table of positions.

Do not rush. In a better ending, time is on your side. Improve your king, improve your worst piece, fix the opponent's pawns on vulnerable squares, and only then start concrete action. Pawn pushes are irreversible — each one spends a tempo and creates a potential weakness — and hasty "winning attempts" are how most superior endings get thrown away. The winning side benefits from the long game; it is the defender who should seek a crisis.

The principle of two weaknesses. One weakness is almost never enough to win, because the defense concentrates everything on it. The method is to fix a second target on the other wing and stretch the defense between the two, attacking each in turn until the defender — who must travel further to cover both — arrives one tempo late. Every clean grandmaster conversion of a "slightly better" ending shows this pendulum at work.

Deny counterplay first. Before collecting your winnings, ask what the opponent's dream is — the passed pawn they want to run, the activity they want to generate — and take it away. A won ending with enemy counterplay is worth less in practice than a merely better ending with none, because counterplay creates exactly the crises where wins slip. Prophylaxis is not passivity; it is the fastest road to a clean conversion.

Tablebases: Where Chess Is Already Solved

Here is a humbling fact: for every position with seven or fewer pieces on the board — kings included — chess is completely solved. Endgame tablebases contain the exact game-theoretic result of every such position with perfect play by both sides, no evaluation or opinion involved. Some of the discovered wins take several hundred precise moves, sailing far past the 50-move rule and defying every human principle along the way.

For practical players the meaning is twofold. Tablebases are the ultimate teacher — checking your ending against one means being corrected by the truth itself, not an engine's estimate. And they have audited classical endgame theory, which has held up remarkably well: the human roadmap above is confirmed knowledge. The endgame is the one region of chess where perfection exists and can be studied.

Common Endgame Mistakes

The same handful of errors costs club players the most half-points, and every one of them is avoidable.

Trading into a lost pawn ending. Pawn endings are the most unforgiving in chess — there is no piece play to muddy the water, so a lost pawn ending is simply lost. Never allow the last pieces off the board until you have counted the pawn ending to the end. "Simplifying when ahead" is good advice that becomes a blunder the moment the resulting king-and-pawn position is not actually winning.

Passive rooks. Tucking the rook in front of a passed pawn or chaining it to a weak pawn's defense is the standard way to lose a holdable rook ending. Give up a pawn for activity rather than defend everything passively — the active rook earns the pawn back with interest.

Lazy kings. Leaving the king on g1 "where it's safe" while the opponent's monarch marches to the center concedes the equivalent of a piece. King activity is not optional in the endgame; it is the main event.

Stalemate accidents. With a huge material edge, careless "obvious" moves can leave the starving enemy king with no legal move — and stalemate turns your whole game's work into a half-point. When the opponent is down to a bare king, check their available replies before each of your moves; as the defender, hunt for stalemate tricks deliberately — it is a legitimate saving resource.

How to Train Endgames at ChessMind AI

Endgame skill is built by playing critical positions against precise resistance, not by nodding along to diagrams. Our endgame trainer is the heart of the program: it hands you the essential theoretical positions — opposition duels, Lucena and Philidor, minor-piece technique — and makes you prove them move by move against a defense that punishes every inaccuracy. Start there, in the roadmap order above. Pair it with the basic checkmates trainer until the fundamental mates are pure reflex. And before deciding how much training time endgames deserve, take the GM diagnostic — it measures your endgame play against the rest of your game and shows where the cheapest rating points are hiding. For most club players, the answer is: here.

Frequently Asked Questions

When does the endgame begin?

There is no official boundary, but the practical test is king safety: the endgame has begun when the kings can safely leave shelter and fight in the center. Queens traded plus reduced material is the typical trigger, though some queenless positions still behave like middlegames when many pieces remain.

Which endgames should I learn first?

In order: the basic mates, then king and pawn versus king (opposition, key squares, square rule), then the core rook endings (Lucena, Philidor, Vancura), then minor-piece endings such as the wrong bishop. Each layer is the foundation for the next, and pawn endings especially are the hidden bottom line of every piece trade.

Are most rook endings really drawn?

The old joke that "all rook endings are drawn" contains real truth: rook endings have enormous drawing resources, and a pawn-down defender with an active rook very often holds. But they are drawn only for players who know the methods — in practice, more half-points change hands in rook endings than anywhere else, precisely because they are so common and so technique-dependent.

Why did Capablanca say to study endgames first?

Because endgames can be studied on their own terms, while openings and middlegames can only be judged by the endings they produce. Endgame knowledge is upstream of everything: it tells you which trades to allow, which structures to aim for, and which advantages are real. It is also permanent — theory barely changes, so nothing you learn expires.

Is the endgame just memorization?

No. The exact theoretical positions worth memorizing number a few dozen for a club player. Everything else is technique — king activity, the principle of two weaknesses, restraining counterplay, patience — plus calculation. The memorized positions serve as destinations; technique is how you steer toward them.

Do I really need endgames if my games end in the middlegame?

Yes. Your games end in the middlegame partly because neither player trusts their endgame — you decline favorable simplifications you cannot evaluate. And as opponents improve, fewer games end in early attacks; the share decided in the endgame rises at every level.

Related chess terms

Openings featuring this idea