Loading field map
Castle theory is the branch of shogi study concerned with the defensive formations that players build around their king. In shogi, the king begins the game on the back rank, exposed to the enemy's pieces. A castle (gyoku-gakoi, literally "king enclosure") is a coordinated arrangement of several pieces—typically gold generals, silver generals, and sometimes a rook, bishop, or pawn—that shelters the king, gives it escape routes, and connects it to the rest of the army. Castle theory asks what makes such formations sound, how they should be built, when they should be abandoned, and how they interact with the attacking plans they are meant to support.
Shogi differs from Western chess in ways that make castling especially important. The board is nine by nine, and the enemy's pieces start only three ranks away. There is no queen, but pieces can be dropped anywhere on the board, including directly in front of the king, and captured pieces are reused. A king left in its starting position is vulnerable to a sudden drop of a silver or gold in front of it, often leading to a forced mate. The king must therefore move, usually to the side, and be wrapped in defensive pieces before the middle game begins.
A castle is not merely a wall. It must satisfy several requirements at once. The king needs a safe square or two to flee to if the wall is breached. The pieces used in the castle must not be so tied down that they cannot participate in the attack. The formation must be buildable in a reasonable number of moves, because every move spent defending is a move not spent attacking. And the castle must be appropriate to the overall strategy: a slow, patient game allows a heavy castle; a sharp, fast attack demands a light one. Castle theory is the study of these trade-offs.
The fundamental defensive piece in shogi is the gold general, which moves one square in any direction except diagonally backward. Golds are slow but sturdy, and they form the backbone of nearly every castle. The silver general, which moves one square diagonally or straight forward, is faster and more flexible but has weak retreating moves; silvers are often placed in front of the king where they can advance to attack, while golds guard the rear and flanks. Pawns, though weak, are crucial for preventing enemy drops directly in front of the king and for creating "holes" that channel the enemy's attack into predictable paths.
The king itself usually ends up on one of the three files on the left or right side of the board, most often the second or third file from the edge. The choice of side depends on which flank the player intends to attack from: in shogi, the king is placed on the side opposite the main attack, so that the two armies do not get in each other's way. This is a central principle: the castle is not an isolated defensive structure but half of a paired plan, with the attack on the other wing.
Castles are conventionally divided into two broad families: static-rook (ibisha) castles, used when the rook stays on its original file, and ranging-rook (furibisha) castles, used when the rook moves to the left side of the board. The distinction matters because the two styles have different rhythms. Static-rook games tend to be slower, with both sides building heavy castles before attacking. Ranging-rook games are often faster, with the ranging side building a light castle and attacking while the static-rook side is still developing.
Within the static-rook family, the most important castle is the Yagura (also called "Fortress"). It is built with the king on the left side, protected by a silver and two golds arranged in a distinctive stepped pattern, with the rook remaining on the right. The Yagura is considered the most solid castle in shogi, but it takes many moves to build and leaves the king somewhat far from the center. It is used when both sides are willing to play a long game. The Mino Castle is the most common ranging-rook castle, built with the king on the left, a gold on the right of the king, a silver on the left, and another gold above the silver. It is quick to build, flexible, and strong against frontal attacks, but weaker against attacks from the side. The Anaguma ("Bear in the Hole") is an extreme castle in which the king is moved all the way to the corner of the board and buried behind three golds and a silver. It is nearly impregnable from the front but takes a very long time to build and leaves the king with few escape routes if the wall is breached.
These three—Yagura, Mino, and Anaguma—are the archetypes, but there are many others. The Nakazumi is a light castle used in some ranging-rook openings. The Kakugawari ("Bishop Exchange") castles are built after the bishops have been traded early, and tend to be simpler because the reduced material makes attacks less dangerous. The Honeycomb and Silver Crown are older or less common formations. What matters for understanding the field is not the full catalogue but the logic: each castle trades off solidity, build time, flexibility, and suitability to a particular opening.
The earliest shogi games, from the medieval and early modern periods, did not use castles in the modern sense. The king often stayed near the center, and games were decided by direct clashes. The development of castles is tied to the slow formalization of shogi strategy in the Edo period (1603–1868), when professional players began to analyze openings systematically. By the late 19th and early 20th centuries, the major castle forms had been named and codified. The Yagura, in particular, became the subject of intense study, and for much of the 20th century it was considered the standard test of a player's positional understanding.
The mid-20th century saw a major shift with the rise of ranging-rook strategies. Players such as Yamada Takuji and later professionals developed the Mino and Anaguma castles into sophisticated systems, and the rivalry between static-rook and ranging-rook styles became the central axis of opening theory. In the late 20th century, computer analysis began to influence castle theory. Engines showed that some traditional castles were less sound than believed, and that certain attacking plans could break them more easily than human players had thought. This has led to a partial reevaluation: some castles that were once considered solid are now seen as too slow, while some light, flexible formations have been rehabilitated.
It is important not to overstate the role of computers. They have sharpened the analysis of specific positions, but they have not replaced the underlying concepts. Castle theory remains a human framework for organizing the infinite variety of positions. What computers have done is to make the trade-offs more precise, and to show that some traditional judgments were too rigid.
Three tensions run through all of castle theory. The first is solidity versus speed. A heavy castle like the Anaguma is nearly unbreakable, but it takes so many moves that the opponent may attack first. A light castle like the Nakazumi can be built in a few moves, but it offers less protection. The choice depends on the opening and the opponent's plan. The second tension is defense versus offense. The pieces in a castle are not available for the attack. A player who castles heavily may find that the opponent's attack arrives first; a player who castles lightly may find that the king is in danger when the attack stalls. The third tension is central versus edge. A king in the center is well placed to defend both flanks, but it is also exposed to attacks from both sides. A king on the edge is safer from one side but has fewer escape routes. The best castle is one that matches the overall strategy, not one that is maximally solid in the abstract.
These tensions are not resolved by a single rule. They are managed by experience and by the specific demands of the position. Castle theory is therefore not a set of formulas but a way of thinking about the king's safety in relation to the whole game.
In practice, players evaluate a castle by asking several questions. Is the king safe from immediate threats, such as a drop of a silver or gold in front of it? Are there escape squares if the wall is breached? Can the pieces in the castle move out to join the attack without leaving the king exposed? How many moves did the castle cost, and what did the opponent do with those moves? Is the castle appropriate to the material balance and the stage of the game?
A key concept is the "hole" (ana), a square adjacent to the king that is not defended by any piece. Holes are dangerous because an enemy piece can be dropped into them, often leading to mate. A well-built castle has no holes near the king, or at least none that the opponent can exploit. Another concept is the "wall" (kabe), a piece that blocks the king's own escape route. A silver placed in front of the king may be good for attacking but bad for defense if it prevents the king from moving. Castle theory is largely the art of avoiding holes and walls while still using the pieces effectively.
In contemporary shogi, castle theory is not a settled body of doctrine but a living part of opening study. Professional players and computer engines continue to refine the evaluation of specific formations. Some traditional castles, such as the Yagura, remain in use but are no longer considered automatically best; they are chosen when the position suits them. Others, such as the Anaguma, are still played but are now met with specific counter-plans that were less well known in the past. The rise of computer-assisted analysis has also led to the creation of new, hybrid formations that do not fit neatly into the old categories.
For the learner, the most important lesson of castle theory is not the memorization of specific formations but the understanding of the principles behind them. A castle is a solution to a problem, and the problem changes with every move. The player who understands why the Mino is strong against frontal attacks but weak from the side, or why the Yagura takes too long against a fast ranging-rook attack, can adapt to unfamiliar positions. The player who only memorizes the shapes will be lost when the opponent deviates.
Castle theory is thus best understood as a branch of strategic thinking, not a collection of recipes. It teaches the player to see the king not as a piece to be hidden but as a piece to be placed, protected, and eventually used. The best castles are not the most solid ones but the ones that fit the game being played.