This is Volume III of The Complete Raumschach Theoretical Series (IRF, 2026). To the best of current knowledge, this is the first systematic opening theory ever published for Raumschach, a game invented in 1907 that has never received a formal theoretical treatment of its opening phase. All analyses are derived from geometric first principles and are subject to revision by computer verification and practical play.
Ferdinand Maack (1861–1930), a German physician, first described Raumschach — literally “Space Chess” — in the Frankfurter Zeitung in 1907. His ambition was explicit: chess was imprisoned on a flat plane, while real conflict moves in three dimensions. He wanted a game where attack could come from above, from below, and from every oblique angle simultaneously. After years of experimentation he settled on a 5×5×5 board as the optimal balance between strategic richness and human tractability. He founded the Hamburg Raumschach Club in 1919, which thrived until the Second World War extinguished it.
The game attracted serious attention. Thomas Rayner Dawson (1889–1951), the English chess problemist, wrote an extensive series on Raumschach in The Chess Amateur (1926–27) and left behind a full manuscript on Normal Form of the game, portions of which were eventually published as Raumschachfunken in 1993 and 1995. Alexey Troitsky, celebrated for his endgame compositions, also investigated its endgame theory. Yet in the 119 years since its invention, no systematic opening theory has ever appeared. Endgame positions were studied; problems were composed; but the opening phase — that crucial first act in which strategic foundations are laid and the character of a game determined — has been left entirely to improvisation.
This paper attempts to fill that gap. After establishing the necessary theoretical foundations (geometry, notation, piece characteristics, and governing principles), it presents a catalog of first moves and eleven named openings with annotated variations. All analysis is derived from geometric first principles — from careful reasoning about what each piece can do, which squares carry the most strategic weight, and how the third dimension changes the calculus of development, center control, and king safety.
A note on method: this analysis has not been verified by computer engines optimized for Raumschach. Evaluations reflect strategic reasoning from first principles rather than exhaustive calculation. This paper should be treated as a framework — a first theoretical map of unmapped territory — that future players and researchers can refine, revise, and extend.
Raumschach is played on a 5×5×5 arrangement of 125 cells, visualized as five standard 5×5 boards (levels) stacked vertically. Levels are labeled A through E from bottom to top, with White occupying the lowest levels and Black the highest. Within each level, ordinary algebraic conventions apply: files labeled a through e left to right (from White’s perspective), ranks numbered 1 through 5 near to far. Coordinates take the form Level + file + rank. The absolute center of the board — equidistant from all six faces of the cube — is Cc3.
Cells alternate in all three dimensions. The 3D diagonal (triagonal) movement of the Unicorn stays permanently within one “color class,” just as a Bishop in flat chess stays on one color. A Unicorn can reach exactly 30 of the 125 cells. In the correct starting position, White’s two Unicorns begin on different complementary color classes — together they cover 60 of the 125 cells, one Unicorn for each half of the board. Black’s two Unicorns likewise occupy complementary color classes covering the same 60 squares. These color-class properties are fundamental to Raumschach strategy and are discussed fully in Section V.
LEVEL E (Black home level — top) Rank 5: [ ♖︎a5 | ♘︎b5 | ♔︎c5 | ♘︎d5 | ♖︎e5 ] ← Back rank pieces Rank 4: [ ♙︎a4 | ♙︎b4 | ♙︎c4 | ♙︎d4 | ♙︎e4 ] ← Pawns Ranks 1–3: (empty) LEVEL D (Black second level) Rank 5: [ 🨢a5 | ♗︎b5 | ♕︎c5 | 🨢d5 | ♗︎e5 ] ← Back rank pieces Rank 4: [ ♙︎a4 | ♙︎b4 | ♙︎c4 | ♙︎d4 | ♙︎e4 ] ← Pawns Ranks 1–3: (empty) LEVEL C (Neutral zone — entirely empty at game start) LEVEL B (White second level) Rank 2: [ ♙︎a2 | ♙︎b2 | ♙︎c2 | ♙︎d2 | ♙︎e2 ] ← Pawns Rank 1: [ ♗︎a1 | 🨢b1 | ♕︎c1 | ♗︎d1 | 🨢e1 ] ← Back rank pieces Ranks 3–5: (empty) LEVEL A (White home level — bottom) Rank 2: [ ♙︎a2 | ♙︎b2 | ♙︎c2 | ♙︎d2 | ♙︎e2 ] ← Pawns Rank 1: [ ♖︎a1 | ♘︎b1 | ♔︎c1 | ♘︎d1 | ♖︎e1 ] ← Back rank pieces Ranks 3–5: (empty)
White’s King is at Ac1; Black’s King is at Ec5.
Three features of this starting position shape the entire theory of the opening:
First, the King’s position. Both kings stand on the central file — White’s at Ac1, Black’s at Ec5. In flat chess the king begins on a wing file; here it begins in the very center of its home rank. The entire c-column (Ac1–Bc1–Cc1–Dc1–Ec1) is an open vertical line connecting both kings through all five levels. This is the most dangerous structural feature on the board, and closing or monitoring it is a mandatory early priority. There is no castling in Raumschach, so neither king can escape this exposure by mechanical means.
Second, the stronger pieces begin on Level B. White’s Queen, two Bishops, and two Unicorns all start on Level B — one level above the King and Rooks, and one level below the neutral territory of Level C. Their natural development flows upward and forward into Level C.
Third, Level C is entirely empty. The vast middle layer of the board begins with nothing on it. The opening is a race to occupy this neutral space, specifically to establish defended presence in the region of Level C, ranks 2–4, files b–d — what this paper calls the “hyperspace center.”
The player who first establishes a stable, defended presence on Level C seizes the initiative and gains a concrete positional advantage. All first-move strategy should be evaluated primarily by this question: does this move contest, occupy, or directly support a piece or pawn on Level C?
This is the 3D equivalent of center control in flat chess, but more decisive. A piece on Level C has geometric access to both sides of the board simultaneously across all three axes, in a way that is structurally impossible from Level A, B, D, or E. The player who controls Level C effectively controls the crossroads of the game.
Raumcapa encodes this exact imperative as its single largest named evaluation term: a central-cube dominance score, weighted at 14 — higher than any other individual heuristic weight in the engine — rewarding the net difference between own and enemy influence projected into the inner 3×3×3 cube (Level C, ranks 2–4, files b–d, the same volume defined in Section IV below). That this one term outweighs every other named heuristic individually is not a coincidence of tuning; it is the engine’s evaluation function agreeing, in numerical form, that nothing else evaluated statically matters as much as who dominates this volume.
In flat chess, pawns are the preferred agents of center control because their loss costs relatively little. In Raumschach, ascending to Level C with an unsupported pawn is typically premature: it becomes a target on an exposed outpost far from its supporting army. The Unicorn and Bishop can reach Level C on the first move and immediately exert influence; an unsupported pawn on Cc2 or Cc4 invites counterplay.
Restated: when advancing vertically, pieces first, then pawns. The pawn on Level C earns its place as a supported outpost, not as a pioneer.
Raumcapa enforces a version of this principle as an explicit admission test rather than a general preference. A pawn move qualifies as opening development only if it does one of two things: opens a line for a friendly Bishop, Unicorn, Rook, or Queen that the pawn was blocking, or claims space inside the central cube. A pawn move serving neither purpose is not treated as development at all — it is set aside in favor of genuine piece development, and only reconsidered as a last resort if nothing better is available. This is a stricter, binary version of “pieces first, then pawns”: the engine does not merely prefer piece development, it does not count a pawn move as development unless that pawn move is doing one of the two specific jobs this principle says a pawn climbing toward Level C should be doing.
With no castling and both Kings on the c-file, vertical pressure along the c-column (Ac1–Bc1–Cc1–Dc1–Ec5) is perpetually dangerous for both sides. A developing move that also closes this column serves double duty and is structurally superior to a move that only develops. Neglecting the c-column for the first five moves is a strategic error difficult to correct later.
Unlike what might naively be assumed, no Bishop can close the c-column on move 1 from Ba1, but the Bishop at Bd1 can reach Cc1 in one step (via the edge-diagonal (+1,−1,0)), simultaneously occupying Level C and closing the c-column. This is the most efficient single first move for the d-file Bishop and forms the basis of the Bishop’s Flank opening. The moves that can place a piece on Cc1 on move 1 are: the Star Jump (Knight to Cc1), the Bishop’s Flank (♗︎Bd1 to Cc1), and at cost of early Queen exposure, the Queen (♕︎Bc1–Cc1).
These three principles generate a coherent opening strategy: contest Level C primarily through pieces (especially Unicorns), while reserving one early developing move to close or monitor the c-column threat against the King.
Flat chess has a 2×2 center (d4, d5, e4, e5). Raumschach’s equivalent is a 3×3×3 sub-cube at the heart of the board: Level C, ranks 2–4, files b–d — 27 squares total, with the absolute center at Cc3. Not all central squares are strategically equal:
| Square | Strategic Significance | Reachable on move 1 by |
|---|---|---|
| Cc3 | The absolute center. A piece here commands every plane simultaneously. Extremely difficult to dislodge once defended. Note: no Unicorn of either color can ever reach Cc3 — it lies outside both Unicorn color classes. It is a Queen’s/Rook’s/Bishop’s/Knight’s square exclusively. | No piece or pawn (requires at least 2 moves). Example: 1. ♕︎Bc1–Cc1, 2. ♕︎Cc1–Cc3 |
| Cc2 | Level C front-center. Key approach square; controls Dd3, Ee4 via the main triagonal. The primary White Unicorn outpost for 🨢Bb1. 🨢Be1 cannot reach Cc2 on move 1 — it reaches Cd2 instead. | Pawn (Bc2→Cc2) or Unicorn (Bb1→Cc2 via (+1,+1,+1)) |
| Cc4 | Level C c-file rank 4. Reachable by Black’s 🨢Dd5→Cc4 on move 1 (via (−1,−1,−1)) — Black’s most direct Level C response. Also reachable by pawn (Dc4 descending) and by White Unicorn from Bb1 after several moves. | Black 🨢Dd5→Cc4 on move 1. White: pawn (Dc4→Cc4) only on move 1. |
| Cc1 | Level C c-file base. Cuts the c-column between the two Kings. High structural and defensive value — the Level C square that most directly addresses king safety. | Knight (Ab1 or Ad1→Cc1), Bishop (Bd1→Cc1 via (+1,−1,0)), or Queen (Bc1→Cc1, though premature) |
| Cd2 | Level C d-file, rank 2. Reachable by 🨢Be1 on move 1 (via (+1,−1,+1)) — the natural opening destination for White’s e-file Unicorn. | Unicorn (Be1→Cd2), Bishop (Bd1→Cd2 via (+1,0,+1)) |
| Bc3 | Level B center. Supports Level B control and prepares for Level C occupation via the upward-forward pawn captures Bc3→Cc4. | Pawn (Bc2→Bc3) |
| Ac3 | Level A center. A Knight here commands Level B squares and can support future Level B/C operations without immediately ascending. | Knight (Ab1 or Ad1→Ac3) |
With White’s Bishops on Ba1 and Bd1, the most important diagonals for opening play are the two flank ascension diagonals:
Ba1–Cb1–Dc1–Ed1: the Bishop at Ba1 ascends through files b, c, d on successive levels, passing through Dc1 — one step above the c-column midpoint. From Cb1, the Bishop covers Cc2 (the key Unicorn outpost) via the edge-diagonal, giving it immediate influence over Level C’s front center even from the b-file. From Dc1 (on move 2), the Bishop presses into Black’s territory and also bears on the c-column.
Bd1–Cc1–Db1–Ea1: the Bishop at Bd1 moves along (−1,+1,0) through Cc1. More importantly, Bd1 can also ascend along (+1,0,+1) to Cd2 — reaching Level C’s d-file on move 1. The most powerful immediate move, however, is ♗︎Bd1–Cc1 via (+1,−1,0): this both occupies Level C and closes the c-column in a single move, satisfying two foundational principles simultaneously. A third path runs along (+1,+1,0) to Ce1.
The key asymmetry between the two White Bishops: Ba1 develops to the b-file flank (Cb1), supporting the Cc2 Unicorn outpost; Bd1 can close the c-column directly (Cc1) or enter Level C’s d-file (Cd2 or Ce1). This gives White great flexibility in how to simultaneously occupy Level C and address the c-column.
The longest space diagonal of the board runs corner-to-corner through the absolute center Cc3. A Unicorn on any square along this line radiates influence through the heart of the board. White’s Bb1 Unicorn can reach Cc2 on move one; from Cc2 it can reach Dd3 on move two and Ee4 on move three — penetrating to Black’s home level in three moves along the most central path available. This triagonal advance is the most forcing plan available in the early middlegame.
| Piece | Start | Directions | Opening Priority & Assessment |
|---|---|---|---|
| Unicorn (U) | Bb1, Be1 | 8 triagonal (space diagonal) directions; slides up to 4 squares in each | ★★★★★ Highest priority. 🨢Bb1 can reach Cc2 on move 1 via (+1,+1,+1). 🨢Be1 reaches Cd2 via (+1,−1,+1) or Ad2 via (−1,−1,+1). The two Unicorns are on complementary color classes, together covering 60 squares. The Unicorn is Raumschach’s defining piece and the most powerful in the opening. |
| Bishop (B) | Ba1, Bd1 | 12 edge-diagonal directions (any two-dimensional plane diagonals) | ★★★★ Very high. ♗︎Ba1 can reach Level C on move 1 (Ba1→Cb1). ♗︎Bd1 can reach Level C on move 1 with multiple options: Cc1 (closing the c-column), Cd2, or Ce1. ♗︎Bd1–Cc1 is uniquely powerful: it satisfies both the Level C Imperative and the c-column principle in a single move. |
| Knight (N) | Ab1, Ad1 | (0,1,2) leaper; 24 possible destinations from board center; leaps over intervening pieces | ★★★★ High. Can leap to Level C (Cc1) on move 1, simultaneously closing the c-column — one of only two pieces that can achieve both goals in one move (the other being ♗︎Bd1). Also excellent developed to Ac3 as the prelude to Percy’s Defense. |
| Queen (Q) | Bc1 | 26 directions (all 6 face-directions + 12 edge-directions + 8 triagonal directions) | ★★★ Powerful but premature development risks early harassment. The Queen is best deployed on moves 3–6, once the position is open enough to benefit from its range. |
| Rook (R) | Aa1, Ae1 | 6 orthogonal directions along any single axis | ★★ Low priority in the opening. Rooks require open files, ranks, or columns. These rarely exist before move 5–7. |
| King (K) | Ac1 | 26 directions, one step each | ★ Protect early. The open c-column represents a structural threat from move 1. The King should not be moved in the opening unless absolutely necessary. |
Each Unicorn is permanently confined to its 30-cell color class. The color class of a square is determined by the three invariant quantities (x+y) mod 2, (y+z) mod 2, and (x+z) mod 2 — where x, y, z denote level, file, and rank respectively (A=1, B=2, etc.; a=1, b=2, etc.). A Unicorn move changes all three coordinates by ±1, and one can verify that all three quantities are preserved by any such move. Each square therefore belongs permanently to one of four color classes, and a Unicorn can only ever visit squares of its own class.
A critical property of the correct starting position: White’s two Unicorns begin on Bb1 = (2,2,1) and Be1 = (2,5,1). Computing the class invariants: for Bb1, (2+2, 2+1, 2+1) mod 2 = (0,1,1); for Be1, (2+5, 5+1, 2+1) mod 2 = (1,0,1). The two White Unicorns are on complementary color classes — their reachable sets are entirely disjoint from each other. Together they cover 60 of the 125 squares, one Unicorn commanding each half of the board.
Black’s two Unicorns on Da5 = (4,1,5) and Dd5 = (4,4,5) have classes (4+1, 1+5, 4+5) mod 2 = (1,0,1) and (4+4, 4+5, 4+5) mod 2 = (0,1,1) respectively — also complementary. Together they cover 60 squares. Remarkably, the White Unicorn domain and the Black Unicorn domain are identical: both sides’ Unicorns together patrol the same 60 squares.
Three strategic consequences follow immediately:
White Unicorn squares on Level C: Among Level C’s 25 squares, 🨢Bb1’s class (0,1,1) contains six: Ca2, Ca4, Cc2, Cc4, Ce2, Ce4. 🨢Be1’s class (1,0,1) contains four more: Cb2, Cb4, Cd2, Cd4. Together, White’s two Unicorns can reach 10 of Level C’s 25 squares. The most central and immediately accessible targets are Cc2 (from Bb1 on move 1) and Cd2 (from Be1 on move 1).
The model configuration for the Dual Unicorn System places one Unicorn from each color class on Level C simultaneously — for example, Cc2 + Cd2: 🨢Bb1→Cc2 on move 1 and 🨢Be1→Cd2 on move 1 (or move 2). This places both Unicorns on the c- and d-files of Level C respectively, covering complementary territory and projecting force along four distinct triagonals toward the opponent’s half of the board. Since the two Unicorns are on different color classes, they cannot occupy the same square and do not compete for squares — they coordinate perfectly.
White has two Knights — Ab1 and Ad1. Both can reach Ac3 (on move 1) or Cc1 (on move 1). With the correct starting position, the “freeing” argument that previously favored one Knight over the other (based on unblocking a specific Bishop diagonal) no longer applies straightforwardly, since neither Bishop is blocked from its first-move destination by the Knights. The choice between ♘︎Ab1 and ♘︎Ad1 in systems that develop the Knight to Ac3 is therefore more a matter of subsequent plans than of structural necessity. That said:
In flat chess, pawns advance along a single axis. In Raumschach, White pawns can move in two directions: forward (increasing rank on the same level) or upward (ascending one level, keeping file and rank constant). This gives each pawn two non-capturing options rather than one, until it reaches Level E or rank 5, where one axis is exhausted. The practical consequence: pawn strategy in Raumschach resembles a combination of standard forward advancement and a vertical “escalation” mechanism.
An important constraint: Level A pawns on rank 2 (Aa2–Ae2) cannot ascend to Level B because Level B rank 2 squares are already occupied by Level B pawns. Therefore, all vertical pawn play in the opening is carried out by Level B pawns ascending to Level C. The Level A pawns can only advance forward (to rank 3) in the opening phase.
From any interior position, a White pawn can capture on up to five squares:
The upward-forward capture is critically important: the pawn at Bc2 can capture at Cc3 — the absolute center — if an enemy piece stands there. The pawn at Bc3 can capture at Cc4. This means that any enemy piece placed on the front ranks of Level C can be challenged by a pawn from one level below, provided the attacker has a pawn on the right square.
The Vertical Chain: pawns on Bc2 and Cc2 form a vertical column on file c. This structure gives real coverage of the squares flanking the c-file but commits material heavily to one file and can be undermined laterally. One clarification is worth stating plainly, consistent with the geometric analysis in Volume I (Monograph I, §Pawn): the two pawns do not defend each other. Neither has a capture direction that reaches straight up the column, so Bc2 cannot recapture on Cc2 or vice versa — “coverage” here means the combined spread of squares the pair threatens to either side (Cb2, Cd2, and Cc3 from the lower pawn; Cb3, Cd3, and Dc3 from the upper one), not mutual support between them. Raumcapa’s evaluation scores this exact configuration as a standing liability (a level-doubled penalty, −20 cp to each pawn) rather than a coverage bonus, for precisely this reason.
The Horizontal Wall: advancing multiple Level B rank-2 pawns forward (Bb3, Bc3, Bd3) creates a wall across Level B rank 3. This is the safest structure and hardest to break, but it cedes Level C initiative to the opponent.
The Diagonal Wedge: a combination of forward and upward advancement creates a pointed wedge aimed at Level C — for example, pawns on Bc3 and Cc2 together. The Bc3 pawn supports the Cc2 pawn (via the upward-forward capture Bc3→Cc4 if needed to threaten enemy pieces on Cc4), while the Cc2 pawn stakes out Level C directly. This is the most dynamic structure and the recommended formation in most opening systems.
Raumschach has no initial two-square pawn advance and consequently no en passant captures. Every pawn advances one step at a time. The strategic consequence is that pawn confrontations develop more slowly, and the opening phase naturally lasts 8–12 moves before direct pawn clashes arise — compared to 2–3 moves in flat chess. This gives both players more time to develop pieces before the position crystallizes into direct conflict.
White has 15 legal pawn moves (5 Level-A pawns with 1 non-capturing option each, plus 5 Level-B pawns with 2 non-capturing options each, per the constraint above) plus numerous piece moves from the starting position. The following evaluates the most significant first moves against the three foundational principles:
| Move | Type | Level C Progress | C-Column | Rating |
|---|---|---|---|---|
| 1. 🨢Bb1–Cc2 | Unicorn | Occupies Cc2 immediately with a piece | Neutral | ★★★★★ Best by principle |
| 1. ♘︎Ab1–Cc1 or 1. ♘︎Ad1–Cc1 | Knight | Occupies Cc1 on Level C | Closes c-column | ★★★★½ Most efficient single move (satisfies two principles) |
| 1. ♗︎Bd1–Cc1 | Bishop | Occupies Cc1 on Level C | Closes c-column | ★★★★½ Equally satisfies two principles; activates the d-file Bishop immediately |
| 1. ♙︎Bc2–Bc3 | Pawn | Supports Cc3/Cc4 approach from Level B | Neutral | ★★★★½ Most principled pawn move |
| 1. ♘︎Ad1–Ac3 | Knight | Indirect; prepares follow-up development | Neutral (prep) | ★★★★ Excellent development |
| 1. ♙︎Bc2–Cc2 (pawn ascending) | Pawn | Occupies Cc2 immediately (unsupported) | Neutral | ★★★★ Dynamic but requires fast follow-up |
| 1. 🨢Be1–Cd2 | Unicorn | Occupies Cd2 on Level C | Neutral | ★★★★ Active development; complements Bb1 Unicorn on c-file |
| 1. ♗︎Ba1–Cb1 | Bishop | Occupies Level C b-file flank | Neutral (flanks only) | ★★★★ Active development; supports Cc2 diagonally |
| 1. ♙︎Ac2–Ac3 | Pawn (off-center) | None | No improvement | ★★★ Solid but slow |
| 1. ♙︎Bb2–Bb3, ♙︎Bd2–Bd3, etc. | Pawn (off-center) | None | No improvement | ★★ Passive; not recommended |
| 1. R–anything | Rook | None useful | Minimal | ★ Poor; rooks are not opening pieces |
The ratings above rest on geometric principle alone. They can be checked, in part, against a corpus of 50,000 Raumcapa self-play games (depth-2 search, analyzed to 12 plies; full methodology and the complete position tree are published separately at raumschach.org). Two results are worth setting against the table.
First, broad confirmation: every move rated ★★★★ or better in the table above remains within a few points of an even score across hundreds of corpus games, and none of the moves rated ★★ or below appear often enough among 50,000 games to register at all — engines simply stop choosing them, which is itself a form of agreement with the table.
Second, a genuine surprise. 1. 🨢Bb1–Ca2 posted the second-highest win rate of any first move in the corpus: 53.9% across 1,040 games, trailing only 1. ♘︎Ab1–Cc1 (54.0%, the Star Jump’s Knight formulation). This outperforms the Unicorn Surge itself (1. 🨢Bb1–Cc2, rated ★★★★★ above; corpus EV 49.2%) by a wide margin. The geometric case for the Surge — immediate, unchallengeable occupation of Level C’s front center — remains entirely sound; the corpus simply suggests that the quieter diagonal retreat to Ca2 generates more practical chances, at least at the engine strength sampled. This is the kind of finding theory should welcome rather than explain away. See Opening 9, below.
A caveat worth holding throughout this section: depth-2 self-play is informative about move popularity and rough tendency, not a tablebase. A gap of 53.9% to 49.2% over roughly a thousand games each is suggestive, not dispositive. Treat the corpus as a second opinion, not a verdict.
The following are the first named openings in the history of Raumschach. Nine systems are presented, each with annotated main lines, key variations, and strategic assessments — together with one dedicated tactical continuation embedded within Opening 6. Move annotations use flat chess punctuation: ! (excellent), !? (interesting), ?! (dubious), ? (mistake).
Opening 1
Named for Ferdinand Maack (1861–1930), inventor of Raumschach. The most natural and historically resonant first move: the central Level-B pawn advances to contest the third rank, directly supporting a future assault on Level C.
1. ♙︎Bc2–Bc3
The pawn at Bc2 advances to Bc3. White claims Level B center, controls the upward-forward capture squares toward Cc4, and prepares to support a piece or pawn on Level C. The pawn at Bc3 also frees the route for the Bc2 pawn to ascend to Cc2 as a supported outpost. This is the philosophical successor to 1. d4 or 1. e4 in flat chess: principled, solid, and flexible.
1. ♙︎Bc2–Bc3 ♙︎Dc4–Dc3
Black’s Dc4 pawn descends to Dc3. Symmetrical. Both sides claim the central third rank of their home-adjacent level. The main continuation:
2. 🨢Bb1–Cc2 🨢Dd5–Cc4
3. ♘︎Ad1–Ac3 ♘︎Ed5–Ec3
4. ♘︎Ab1–Cc1!
White’s Knight leaps to Cc1 on move 4: it develops to Level C AND closes the c-column between the kings. Note that Black’s 🨢Dd5–Cc4 on move 2 is the principled Unicorn response — the Unicorn arrives at Cc4, directly facing White’s Unicorn at Cc2 on the same color class. The position is roughly equal, with White holding a slight structural edge thanks to c-column closure.
1. ♙︎Bc2–Bc3 🨢Dd5–Cc4!
Black’s most aggressive response: seizing Level C directly with the Unicorn to Cc4. The Unicorn at Dd5 = (4,4,5) moves (−1,−1,−1) to Cc4 = (3,3,4). Principled and strong — Black establishes a Unicorn on Level C’s c-file immediately. Both Black and White Unicorn domains overlap, so the two Unicorns at Cc2 and Cc4 (after White’s reply) are on the same color class and can potentially threaten each other through subsequent triagonal moves.
2. 🨢Bb1–Cc2 ♙︎Dc4–Dc3
3. ♘︎Ab1–Cc1
White matches with the Unicorn to Cc2, Black reinforces on Level D, and White closes the c-column with the Knight on move 3. White is slightly better: the c-column is secured and White’s Unicorn at Cc2 is well-centralized.
1. ♙︎Bc2–Bc3 ♙︎Dc4–Cc4
Black contests Level C with a pawn. The pawn at Cc4 is a valid response, but it arrives unsupported — unlike the Unicorn counter above.
2. 🨢Bb1–Cc2!
White immediately places a piece on Level C. Now White has a Unicorn at Cc2 (piece, fully mobile) against Black’s pawn at Cc4 (static, no piece support yet). White has a positional advantage: piece versus pawn on Level C.
Assessment: Principled, solid, and the recommended starting point for all Raumschach theory. The Maack Opening is flexible, handles all Black responses gracefully, and leads naturally to well-understood middlegame structures.
Opening 2
White’s most ambitious pawn first move: the c-pawn ascends directly from Level B to Level C on move one.
1. ♙︎Bc2–Cc2
The pawn at Bc2 rises to Cc2, immediately occupying the front center of Level C. This satisfies the Level C Imperative on move 1 with a pawn rather than a piece — bolder and more forcing than the Maack Opening, but slightly less efficient. The Ascension is most effective when followed immediately by piece development to Level C.
1. ♙︎Bc2–Cc2 🨢Dd5–Cc4!
Black answers with the Unicorn to Cc4 — the most principled reply. Level C now features an asymmetric imbalance: White’s pawn (Cc2) versus Black’s Unicorn (Cc4). The pawn is cheaper to maintain; the Unicorn is more mobile.
2. 🨢Bb1–Ce2 ♙︎Dc4–Dc3
3. ♙︎Bc2–Bc3!
White deploys the Unicorn to Ce2 (same color class (0,1,1) as Bb1 — supporting the Cc2 pawn from within the same Unicorn domain), Black advances on Level D, and White reinforces Level B. The Cc2 pawn is now supported. White’s position is solid.
1. ♙︎Bc2–Cc2 ♙︎Dc4–Cc4
Black mirrors with the pawn. Two pawns on the c-file of Level C, two ranks apart, unable to directly capture each other. The contest now shifts to the b- and d-files and to piece development.
2. ♙︎Bc2–Bc3 ♘︎Ab1–Ac3
3. 🨢Bb1–Ce2 🨢Dd5–Cc4... (Black Unicorn would reinforce the Cc4 pawn)
A complex, roughly balanced position results.
1. ♙︎Bc2–Cc2 ♙︎Dc4–Dc3 (Black ignores Level C)
2. 🨢Bb1–Ce2 ♙︎Dc4–Dc3
3. ♙︎Bc2–Bc3
White supports the Cc2 pawn. Black’s restraint has not prevented White from establishing a supported Level C presence.
Assessment: Dynamic and double-edged. Recommended for players who prefer immediate initiative. The Level C pawn must be supported quickly or it becomes a liability.
Opening 3
The most conservative pawn advance: the Level A c-pawn moves forward first.
1. ♙︎Ac2–Ac3
The pawn at Ac2 advances to Ac3. This occupies the central square of Level A and creates structural solidity, but makes no immediate progress toward Level C. Its weakness is that it concedes the Level C race to Black, and it permanently commits the Ac3 square to a pawn, limiting Knight development options (the Knight from Ab1 or Ad1 can no longer go to Ac3).
1. ♙︎Ac2–Ac3 🨢Dd5–Cc4!
2. 🨢Bb1–Cc2 ♙︎Dc4–Dc3
3. ♙︎Bc2–Bc3 ♘︎Ed5–Ec3
4. ♘︎Ad1–Bc3
Black seizes Level C with the Unicorn on move 1, before White has gotten there. The Knight at Ad1 leaps to Bc3 instead of Ac3 (since the pawn is there). From Bc3 the Knight supports Level C operations and controls multiple squares. White is slightly behind in the Level C race.
Assessment: Solid but slow. Acceptable as a surprise weapon, but generally inferior to the Maack Opening because it concedes the Level C initiative.
Opening 4
Immediate Knight centralization to the Level A central square Ac3 — a square that stays home rather than leaping toward the contested center the way the Star Jump’s Knight does, and is no weaker for it.
1. ♘︎Ad1–Ac3
The Knight at Ad1 leaps to Ac3. From Ac3, the Knight controls multiple squares across Levels B and C, contributing immediately to the Level C contest even while sitting on Level A — the level farthest from the action, watching everything that matters without ever needing to enter it.
Named for Persephone — Percy, as she was always called. An orange cat, soft and beautiful and elegant and graceful in every way, and a little wary of strangers: whenever one entered the house, she would jump, Knight-like, straight to the calmest, most defensible spot in the room, as far from the visitor as the space allowed — and stay every bit as watchful and effective from there. A fuller portrait of Percy appears on the site’s Dedications page.
1. ♘︎Ad1–Ac3 🨢Dd5–Cc4
2. 🨢Bb1–Cc2 ♙︎Dc4–Dc3
3. ♘︎Ab1–Cc1! ♘︎Ed5–Ec3
4. ♙︎Bc2–Bc3
White’s four-move sequence: Knight develops centrally (move 1), Unicorn enters Level C at front-center (move 2), second Knight leaps to Cc1 closing the c-column (move 3), pawn reinforces Level B (move 4). Four moves; two pieces developed; Level C occupied; c-column secured. Black’s 🨢Dd5–Cc4 is the principled Level C response throughout.
1. ♘︎Ad1–Ac3 🨢Dd5–Cc4
2. 🨢Bb1–Cc2 ♕︎Dc5–Dc1?!
Black’s Queen descends to Dc1, attacking White’s Unicorn on Cc2 vertically. This is premature — the Queen on Dc1 is somewhat misplaced and will have to move again.
3. ♘︎Ab1–Cc1 ♘︎Ed5–Ec3
4. ♙︎Bc2–Bc3
White closes the c-column with the Knight, ignoring the Queen threat. White has a comfortable position.
Assessment: Classical and deeply principled. Percy’s Defense is a strong repertoire opening. The ♘︎Ad1–Ac3 / 🨢Bb1–Cc2 / ♘︎Ab1–Cc1 three-move sequence satisfies all three foundational principles efficiently.
Opening 5
White’s most dramatic first move: either Knight leaps immediately to Level C, occupying Cc1 on the very first move.
1. ♘︎Ab1–Cc1 or 1. ♘︎Ad1–Cc1
The Knight — from either Ab1 or Ad1 — executes a (0,1,2) leap to land on Cc1. Verification from Ab1 = (1,2,1): adding (+2,+1,0) = (3,3,1) = Cc1 ✓. From Ad1 = (1,4,1): adding (+2,−1,0) = (3,3,1) = Cc1 ✓. Both Knights target the identical square.
Recommendation: ♘︎Ab1–Cc1, preserving the Ad1 Knight for later deployment via Percy’s Defense (♘︎Ad1–Ac3 on a subsequent move). Note that ♗︎Bd1–Cc1 offers the same square on the same move — the choice between Knight and Bishop on Cc1 is a genuine strategic decision.
The Star Jump simultaneously: (1) occupies Level C on move 1; (2) sits directly on the c-column between the two Kings — closing the most dangerous open line; (3) attacks approximately 12–14 squares across three levels.
1. ♘︎Ab1–Cc1 🨢Dd5–Cc4
2. ♙︎Bc2–Bc3 ♙︎Dc4–Dc3
3. 🨢Bb1–Cc2 ♘︎Ed5–Ec3
4. ♘︎Ad1–Ac3
White now has a Knight at Cc1 and Unicorn at Cc2 — a powerful Level C duo. Black’s principled 🨢Dd5–Cc4 response faces the Knight/Unicorn formation directly. The c-column is closed by the Knight from move 1. White’s second Knight develops to Ac3 on move 4.
1. ♘︎Ab1–Cc1 ♕︎Dc5–Cc4
Black’s Queen descends to Cc4, attacking the Knight at Cc1 diagonally. The Knight at Cc1 is under attack.
2. ♙︎Bc2–Bc3!
White defends the Knight indirectly: the pawn at Bc3 now threatens the upward-forward capture Bc3→Cc4, winning the Black Queen if it stays on Cc4. Black must retreat the Queen. The Star Jump has successfully closed the c-column and White continues developing normally.
Assessment: Ambitious and creative. The Star Jump satisfies two of three foundational principles on move 1 simultaneously. It requires careful defensive awareness of the Queen attack but is a fully valid opening system.
Opening 6
The theoretically strongest first move by geometric principle: the Unicorn immediately occupies Cc2, the front center of Level C.
1. 🨢Bb1–Cc2
The Unicorn at Bb1 = (2,2,1) moves along direction (+1,+1,+1) to Cc2 = (3,3,2). The Unicorn Surge is proposed as the objectively strongest first move for four reasons:
1. 🨢Bb1–Cc2 🨢Dd5–Cc4!
Black’s most principled response: 🨢Dd5 moves along (−1,−1,−1) to Cc4 = (3,3,4). Both Unicorns now occupy the c-file of Level C — White at Cc2, Black at Cc4. They are both in color class (0,1,1), which means they occupy the same color domain. They cannot directly capture each other in one step (four ranks apart), but their presence on the same color class creates lasting tactical tension. White continues:
2. ♙︎Bc2–Bc3 ♙︎Dc4–Dc3
3. ♘︎Ad1–Ac3 ♘︎Ed5–Ec3
4. ♘︎Ab1–Cc1
Model development for White. The Knight at Cc1 closes the c-column. White has the slight positional advantage that Cc2 is fractionally more forward than Cc4.
1. 🨢Bb1–Cc2 Cc4 (pawn Dc4 descends)
Black responds with the pawn rather than the Unicorn. White has a Unicorn (piece, fully mobile) against Black’s pawn at Cc4 (static).
2. ♙︎Bc2–Bc3! 🨢Dd5–Cc4... (if Black plays the Unicorn on move 2 instead)
3. ♘︎Ab1–Cc1
White’s pawn at Bc3 supports the Unicorn at Cc2 and restricts the Cc4 pawn. White closes the c-column on move 3. White is slightly better.
1. 🨢Bb1–Cc2 🨢Da5–Cb4!
Black deploys the a-file Unicorn: Da5 = (4,1,5) moves (−1,+1,−1) to Cb4 = (3,2,4). Black contests Level C on the b-file. This is an asymmetric but principled response — Black’s 🨢Da5 is in class (1,0,1), while White’s 🨢Bb1 is in class (0,1,1): they are on different color classes and cannot directly threaten each other.
2. ♙︎Bc2–Bc3 ♙︎Dc4–Dc3
3. ♘︎Ab1–Cc1 🨢Dd5–Cc4
White continues standard development, closing the c-column. Black’s second Unicorn 🨢Dd5–Cc4 joins the fray. Black now has two Unicorns on Level C (Cb4 and Cc4). Note that Cb4 is class (1,0,1) and Cc4 is class (0,1,1) — Black’s Unicorns are also on complementary classes, covering their respective domains on Level C. A rich positional battle results.
1. 🨢Bb1–Cc2 ♙︎Dc4–Dc3
Black declines to contest Level C.
2. 🨢Cc2–Dd3!
The Unicorn advances from Cc2 to Dd3, penetrating Level D on move 2. From Dd3, the Unicorn threatens Ee4 (attacking Black’s E-level pawns) and controls extensive territory. The Refusal is refuted by 2. 🨢Cc2–Dd3!
Assessment: The Unicorn Surge is the most principled, powerful, and immediate first move in Raumschach. Players who master its variations will hold a significant theoretical advantage.
Independent of how Black replies on move 1, White has a second, sharper idea available from the Unicorn Surge position — one that treats Cc2 less as a square to hold than as a launchpad for an exchange combination on Black’s home level.
1. 🨢Bb1–Cc2 …
2. ♗︎Ba1×Ea4! ♖︎Ea5×Ea4
3. 🨢Cc2×Ea4
On move 2, the Bishop travels the full length of the board to capture the pawn sitting undisturbed on its starting square at Ea4 — the geometric mirror of White’s own Aa2 pawn. Played alone, this capture is unsound: Black’s Rook at Ea5 defends Ea4 from the position’s very first moment, so 1. ♗︎Ba1×Ea4 ♖︎Ea5×Ea4 simply loses a Bishop (560 cp) for a pawn (100 cp) — borne out directly by the corpus, which scores the bare capture at only 44.1% across 153 games. The purpose of 1. 🨢Bb1–Cc2 is to fix that problem in advance: the Unicorn at Cc2 reaches Ea4 in a single triagonal leap, so once Black’s Rook takes the bait on move 2, White simply recaptures on move 3.
The material outcome is exactly even — a Bishop (560 cp) for a pawn and a Rook (100 + 460 = 560 cp) — but the resulting position is not. White trades its least mobile minor piece for Black’s rook, opens a permanent hole in Black’s home level, and leaves the Unicorn standing deep in enemy territory on Ea4 with the initiative.
Named for Joey, in memory (d. 22 May 2026). A fuller note appears on the site’s Dedications page.
The corpus figure for the unprepared capture (44.1%) is itself the evidence that the one-move investment in the Unicorn is what makes the combination sound. Recommended as a follow-up plan to the Unicorn Surge whenever Black has not specifically defended Ea4.
Opening 7
A positional first move: a Bishop develops to Level C on move 1, entering the neutral zone. Two formulations are available with the correct starting position:
1. ♗︎Ba1–Cb1 or 1. ♗︎Bd1–Cc1
♗︎Ba1–Cb1: The Bishop at Ba1 = (2,1,1) moves along (+1,+1,0) to Cb1 = (3,2,1). This reaches Level C’s b-file flank. From Cb1, the Bishop covers Cc2 (the key Unicorn outpost) via the edge-diagonal (0,+1,+1), and commands the long diagonal Cb1–Dc1–Ed1 into Black’s territory. It does not close the c-column — that must be addressed on a subsequent move.
♗︎Bd1–Cc1: The Bishop at Bd1 = (2,4,1) moves along (+1,−1,0) to Cc1 = (3,3,1). This simultaneously occupies Level C and closes the c-column — satisfying two foundational principles in a single move, just as the Star Jump does. From Cc1, the Bishop commands the c-file and multiple diagonals. This is arguably the stronger of the two Bishop’s Flank formulations and deserves to be treated as a primary opening in its own right.
1. ♗︎Ba1–Cb1 🨢Dd5–Cc4!
2. 🨢Bb1–Cc2 🨢Da5–Cb4
3. ♘︎Ab1–Cc1 ♙︎Dc4–Dc3
4. ♙︎Bc2–Bc3
Black responds with active Unicorn development to Level C (both Black Unicorns to Cc4 and Cb4). White deploys the Unicorn to Cc2 on move 2, supported diagonally by the Bishop at Cb1. On move 3, White closes the c-column with the Star Jump Knight. By move 4, White has Bishop, Unicorn, and Knight on Level C with the c-column secured.
1. ♗︎Bd1–Cc1 🨢Dd5–Cc4
2. 🨢Bb1–Cc2 ♙︎Dc4–Dc3
3. ♙︎Bc2–Bc3 ♘︎Ed5–Ec3
4. ♘︎Ad1–Ac3
With the c-column already closed on move 1 by the Bishop, White spends moves 2–4 on pure development: Unicorn to Cc2, pawn to Bc3, Knight to Ac3. Black’s 🨢Dd5–Cc4 stakes out the Level C c-file, creating the familiar tension between opposing Unicorns on Cc2 and Cc4. White has achieved an excellent position with the c-column addressed from the very first move.
1. ♗︎Ba1–Cb1 ♕︎Dc5–Dc1?!
Black’s Queen descends to Dc1, aiming at the Bishop at Cb1 diagonally. This is premature. White responds calmly:
2. 🨢Bb1–Cc2! ♕︎Dc1–Dc5
3. ♘︎Ab1–Cc1 ♙︎Dc4–Dc3
4. ♙︎Bc2–Bc3
Black has wasted a Queen tempo. White continues the model development sequence undisturbed.
Assessment: Positional and principled, with two distinct formulations. ♗︎Ba1–Cb1 develops with long-diagonal pressure; ♗︎Bd1–Cc1 immediately solves the c-column problem. Both are strong secondary options behind the Unicorn Surge and Star Jump.
Opening 8
A forcing combinative opening discovered in the first recorded game between a human player and the Raumfischer engine, 1 April 2026 — eleven days after the death of Pluto, an adventurous and happy cat who was never more in his element than when hunting in the woods near home, often returning with several catches from a single outing. Named for him by Avidius Du Vide, owner of the site: a combination that lands three threats from one motion is exactly the kind of multiplied catch Pluto used to bring proudly home. A fuller portrait of Pluto appears on the site’s Dedications page. The opening itself deploys a Queen-supported Knight maneuver that delivers a three-way fork — check on the King, Rook, and Unicorn simultaneously — within three moves.
1. ♕︎Bc1–Cb2
The Queen at Bc1 = (2,3,1) moves along the triagonal direction (+1,−1,+1) to Cb2 = (3,2,2), entering Level C on the very first move. This appears to violate the classical principle of developing minor pieces before the Queen — but it is purposeful: the Queen is being posted on Level C’s b-file specifically so that it can recapture on Cb5 along the rank axis after the Knight maneuver. Because the Queen moves only to Level C rank 2, it is not exposed to immediate harassment. Its presence at Cb2 is the structural foundation on which the entire combination rests.
2. ♘︎Ab1–Bb3
The Knight at Ab1 = (1,2,1) leaps via the Knight direction (+1,0,+2) to Bb3 = (2,2,3). The Knight ascends to Level B and aligns itself on the b-file — one hop below the outpost it will occupy next move.
3. ♘︎Bb3–Cb5†
The Knight executes the identical leap a second time — direction (+1,0,+2) again — from Bb3 = (2,2,3) to Cb5 = (3,2,5). From this Level C outpost on the enemy’s back rank, the Knight delivers check and simultaneously attacks three Black pieces:
Black cannot easily attack the Knight: although the Black Queen at Dc5 can reach Cb5 via the edge diagonal (−1,−1,0), the Queen at Cb2 defends Cb5 along the rank axis (Cb2→Cb3→Cb4→Cb5), making the exchange highly unfavorable for Black. The most practical defense is to escape with the King, surrendering either the Rook or the Unicorn.
1. ♕︎Bc1–Cb2 ♙︎Dc4–Dc3
2. ♘︎Ab1–Bb3 ♙︎Ec4–Ec3
3. ♘︎Bb3–Cb5† ♔︎Ec5–Dc4
4. ♘︎Cb5×Ea5
Black’s pawn advances (moves 1–2) are natural development that leaves the Rook on Ea5 and the Unicorn on Dd5 untouched. After the fork on move 3, the pawn that vacated Dc4 to Dc3 has opened the only available King refuge: nearly every other square adjacent to Ec5 is occupied by Black’s own pieces. The King retreats to Dc4, and White captures the Rook on Ea5. White has won a Rook (≈500 cp) for nothing.
Alternatively, White can take the Unicorn on move 4 (♘︎Cb5×Dd5, ≈800 cp) if it remains on its starting square — the objectively larger material gain. Raumcapa’s evaluation will prefer the higher-valued piece if available; in practice it depends on what Black played on moves 1–2.
3. ... ♕︎Dc5×Cb5
Black captures the forking Knight with the Queen, relieving the check and eliminating the immediate fork. But the Queen at Cb2 stands by:
4. ♕︎Cb2×Cb5
White recaptures. Material balance: White has traded a Knight (≈500 cp) for the Black Queen (≈1500 cp), netting approximately +1000 cp — the equivalent of more than two extra Rooks for nothing. The Queen Exchange Defense eliminates the fork but concedes more than twice the material of simply fleeing and losing a Rook (≈460 cp). It is the inferior defense.
This variation reveals the true purpose of move 1. ♕︎Bc1–Cb2 is not premature Queen development but a calculated anchor: the Queen is posted at Cb2 to ensure that if Black captures the Knight, White recaptures profitably. Remove the Queen from Cb2 and the combination fails — the Knight at Cb5 would be undefended and the Queen capture by Black would be sound. The Queen move is the combination’s load-bearing piece.
Black can execute the identical combination from the other side of the board. The perfectly symmetric line is:
1. ... ♕︎Dc5–Cd4
2. ... ♘︎Ed5–Dd3
3. ... ♘︎Dd3–Cd1†
The Black Queen moves from Dc5 = (4,3,5) along the triagonal (−1,+1,−1) to Cd4 = (3,4,4). The Knight travels Ed5 = (5,4,5) → Dd3 = (4,4,3) → Cd1 = (3,4,1) via two applications of direction (−1,0,−2). From Cd1 = (3,4,1), the Knight delivers an identical triple fork: it checks the White King at Ac1 via (−2,−1,0), attacks the White Rook at Ae1 via (−2,+1,0), and attacks the White Unicorn at Bb1 via (−1,−2,0). The Queen at Cd4 defends the Knight on Cd1 along the triagonal (+1,−1,+1) — precise mirror of the White construction. This line is named the Cd1 Triple Fork.
Note: Raumfischer, the engine discussed in this assessment, has since been retired in favor of Raumcapa. The analysis below is preserved as a historical record of the opening’s discovery and of Raumfischer’s own reasoning at the time.
Raumfischer found this opening without a book, guided entirely by its seven Fischer-derived evaluation heuristics. How well does the result represent Fischer?
The resemblance is genuine in the most important respects. The Knight maneuver (Ab1→Bb3→Cb5) is patient, purposeful, and step-by-step — the piece is improved one hop at a time toward a dominant outpost, then unleashed with maximum effect. Fischer’s greatest games routinely feature exactly this kind of long-range Knight repositioning: the Nd4→Nf5→Nd6 maneuvers in his games against the Sicilian, the extraordinary centralization in Fischer–Benko (Candidates, 1959), the Knight that walks across the board in Fischer–Petrosian (Buenos Aires, 1971). The combination is also concrete and forcing rather than vague and positional — Black has no good choice, and the advantage is material rather than atmospheric. This is Fischer through and through: “Every position has a truth, and it can be found.”
The “every piece must work” principle (Raumfischer §6.2) is satisfied aggressively. From move 1, the Queen is not idle on its starting square but deployed to a specific coordinate with a specific future task. The Knight executes the same leap twice, arriving exactly where it needs to be. Nothing is wasted.
Where the opening departs from Fischer’s habits in flat chess is the early Queen. Fischer almost never brought the Queen out before his minor pieces were developed. But in Raumschach’s three-dimensional geometry, the Queen has more room to maneuver without harassment, and the three-axis board dilutes the tempo-loss risk of early Queen play. Raumfischer’s Fischer personality layer (§5.4) awarded the center-entry bonus when the Queen stepped to Level C on move 1 — correctly, by Fischer’s spatial dominance principle. The Raumschach analog of Fischer’s Najdorf Poisoned Pawn may be precisely this: bold early Queen insertion, calculated material-first thinking, complications that resolve in White’s favor.
Assessment: Tactically decisive and materially forcing. The Cb5 Triple Fork is the most immediately dangerous three-move plan currently known in Raumschach. Its weakness is that an opponent who recognizes the pattern can frustrate it by moving the targeted Rook or Unicorn on moves 1–2 (reducing it to a double fork), but even then White retains a concrete advantage. Highly recommended as a surprise weapon and a model of how three-dimensional tactics differ qualitatively from their two-dimensional counterparts.
Opening 9
Its strength only became visible once 50,000 self-play games made it possible to check first-move popularity against outcome (Section VII, above). 🨢Bb1–Ca2 finished with the second-highest win rate of any first move in the corpus, 53.9% across 1,040 games, trailing only the Star Jump’s 1. ♘︎Ab1–Cc1 (54.0%). Geometrically the move looks modest — the Unicorn retreats one diagonal step rather than advancing onto Level C — but its purpose only becomes clear on the move that follows.
1. 🨢Bb1–Ca2 …
2. ♕︎Bc1–Ec4†
The Queen descends the full length of the board from Bc1 to Ec4, delivering check. This looks reckless — a Queen four levels deep into enemy territory on move 2 — but it is not: Ec4 is precisely the square the Unicorn just moved to cover from Ca2. Any attempt to win the Queen on Ec4 runs straight into the Unicorn’s defense of that square, which is the entire point of move 1. The opening is named for what it does rather than how it looks: a Queen sent forward only once its retreat square and its destination are both secured.
The same configuration generalizes beyond the opening itself into one of the sharpest tactical resources available in Raumschach: if Black ever commits its defenders of Ec4 without attending to its own King’s safety, White’s pressure on that square can mature into a forced spacemate. Raumcapa, playing both sides in a self-play game recorded in the Studio, reached spacemate by exactly this mechanism in six moves — the engine’s shortest decisive game on record, with something of the flavor, if not the brevity, of a fool’s mate.
Named for Jupiter. A fuller note will appear on the site’s Dedications page.
Assessment: Quietly the best-performing first move in the corpus. Score it provisionally — depth-2 self-play is a strong hint, not a proof — but a 53.9% empirical result paired with a genuine spacemate mechanism is a serious combination of practical and theoretical merit. Recommended as a primary weapon pending deeper verification.
Opening 10
The most-played first move in the entire 50,000-game corpus (1,240 games) — and, until now, one with no formal treatment at all. It is not a new idea so much as a continuation of one already in this volume: it travels the identical diagonal ray as the Bishop’s Flank (♗︎Bd1–Cc1, Opening 7), sliding one square further past the column-closing stop at Cc1 to land on Db1.
1. ♗︎Bd1–Db1!?
Bd1=(2,4,1) to Db1=(4,2,1) via the edge-diagonal direction (+1,−1,0) applied twice — the same direction the Bishop’s Flank uses once. From Db1, a second ray opens that the shorter move never reaches: the diagonal (0,+1,+1), which stays entirely within Level D and runs Db1–Dc2–Dd3–De4, landing directly on Black’s e-file pawn on Level D. No Black piece in the starting position defends De4 under its own movement rules — the Unicorn at Dd5 would need a triagonal step it does not have to that square; the Bishop at De5 would need an orthogonal step it cannot make. The threat is real, and immediate.
The cost is equally real. Unlike the Bishop’s Flank, the c1-column is not closed — the Bishop has slid past Cc1 rather than stopped on it, leaving the column exactly as open as it was before move 1.
▶ Line: 1. ♗︎Bd1–Db1!? ♙︎De4–De3 (or 1…♙︎De4–Ce4; either pawn move steps the e-pawn off the threatened diagonal)
2. 🨢Bb1–Cc2
Black has spent a move on defense rather than development; White continues with the Unicorn Surge as if nothing had happened. The exchange is a tempo for a tempo, with the open c-column as the only lasting cost on White’s side of the ledger.
Evaluation: The corpus’s clear favorite by raw popularity, yet its measured EV (49.5%) is almost exactly even — popularity here tracks the size of the immediate threat, not a verified advantage. The Long Ramp is best understood as an aggressive alternative to the Bishop’s Flank, trading the column-closing benefit every other system in this volume treats as close to obligatory for a one-move head start on a concrete threat. Whether that trade is sound, or merely popular, awaits deeper verification.
Opening 11
The corpus’s second most-played first move overall (1,204 games, EV 50.4%) sends the Queen to the exact square the Unicorn Surge (Opening 6) reaches by an entirely different route.
1. ♕︎Bc1–Cc2!?
Bc1=(2,3,1) to Cc2=(3,3,2) via the edge-diagonal direction (+1,0,+1) — a Queen move, not a triagonal one. The shared destination with the Unicorn Surge is a coincidence of geometry, not a shared mechanism: the Queen arrives by a diagonal step available to her alone among the pieces that can reach Cc2 in one move.
The move sits in direct tension with the development order established in Volume II and the piece-priority table of Section V above, where the Queen is recommended as the seventh piece to enter play — after both Unicorns are coordinated, the c-column is secured, and the King has reached its corner. Raumcapa, the evaluation engine cited throughout this series, enforces a stricter version of the same principle as a hard rule rather than a preference: the Queen may not leave her starting square until at least one own Knight has already moved (Volume I, Monograph VI). Under that rule, Raumcapa itself would never produce this move on move 1.
The corpus games behind this line, however, predate the hard rule described above: the 50,000-game self-play corpus was generated by an earlier build of Raumcapa, before the Queen-development restriction of Volume I, Monograph VI was added to its evaluation function. The disagreement is therefore not a contradiction inside the theory; it is a snapshot of an earlier stage of the engine’s own development, a habit its later, stricter self went on to forbid.
▶ Line: 1. ♕︎Bc1–Cc2 🨢Dd5–Cc4 (the same principled reply that meets the Unicorn Surge — Black contests Level C with a Unicorn rather than reacting to the Queen directly)
2. ♘︎Ab1–Cc1
White closes the c-column; the Queen’s early appearance on Cc2 has created no immediate problem for either side, and the position resembles a transposed Unicorn Surge with the roles of Queen and Unicorn exchanged.
Evaluation: Unprincipled by Raumcapa’s own current standard, yet practically sound in its own earlier self-play corpus at a rate matching the Unicorn Surge itself. Recommended only provisionally, and only for players willing to treat the early Queen’s safety as a calculated risk rather than a settled best practice — exactly the kind of judgment Fischer’s own practical style was known for.
In flat chess, a battery is formed by Rook and Queen on the same file, or Bishop and Queen on the same diagonal. In Raumschach, the most powerful formation is the triagonal battery: Queen and Unicorn aligned along the same space diagonal. A Queen on Bc2 and Unicorn on Cc3 form a triagonal battery aimed at Dd4 and Ee5 — two moves deep into enemy territory. This battery can only be blocked by interposing a piece on an intermediate square. The triagonal battery is the most forcing formation in Raumschach.
Because both Kings begin on the c-column and the column is entirely open at the game’s start, any piece placed on the c-column outside the kings creates an immediate vertical skewer threat. A White Rook ascending to Cc1 (once the path through Bc1 is clear) would threaten to capture the Black King at Ec5 along the column unless multiple pieces intervene. This threat is so potent that both sides should dedicate at least one developing move to blocking the c-column within the first four moves. Neglecting this is the single most common structural error available to opening players.
A piece established on Level C with pawn support creates a Level C outpost analogous to the classical knight outpost on d5 or e5 in flat chess. The piece on Level C projects influence across all 26 adjacent directions simultaneously. Establishing and maintaining Level C outposts is the central positional goal in the Raumschach middlegame.
The three-dimensional board creates a tactical motif with no flat-chess equivalent: the vertical fork, in which a single piece simultaneously attacks two pieces on different levels of the same file or triagonal. For example, a Knight at Cc3 might attack the Black pawn at Dc4 and the Black Unicorn at Ec5 simultaneously. Vertical forks are much harder to see than flat chess forks because the player must visualize across levels. Training to spot vertical forks is one of the most important tactical skills in Raumschach.
When a Level B pawn ascends to Level C and, from its new position, immediately threatens to capture an enemy piece or to continue upward to Level D, this “ascending break” is the Raumschach equivalent of a central pawn break. The ideal ascending break achieves three things simultaneously: (a) occupies a defended Level C square, (b) threatens an enemy piece from below via the upward-forward capture, and (c) opens a line for a friendly piece on Level B behind it.
With White’s two Unicorns on complementary color classes, the strategic goal is to place them on two well-separated Level C squares that collectively cover both color domains. Because 🨢Bb1 is class (0,1,1) and 🨢Be1 is class (1,0,1), the model configuration places one on each class — for example Cc2 + Cd2 (🨢Bb1→Cc2 and 🨢Be1→Cd2, both reachable on move 1). This gives White a Unicorn on the c-file and another on the adjacent d-file of Level C, covering complementary territory and projecting force along four distinct triagonals. Since the Unicorns are on different color classes, they cannot occupy the same square and do not compete for squares — the ideal situation for mutual coordination.
The opening phase of Raumschach typically concludes when both sides have developed their major pieces (especially Unicorns and Bishops) to active squares, contested Level C in some fashion, and secured the c-column against immediate threats. This usually requires 7–10 moves. Four questions determine the character of the resulting middlegame:
Who controls Level C? The player with more active, defended pieces on Level C holds a tangible positional advantage. Lone, unsupported pieces on Level C are targets; coordinated piece-and-pawn structures on Level C are fortresses. The player who achieves the Diagonal Wedge pawn structure (Level B pawn + Level C pawn on adjacent files) while maintaining a piece on Level C has achieved the ideal opening configuration.
Is the c-column closed? An open c-column remains permanently dangerous for both Kings. The player who has closed it favorably — with a piece on any Cc-square that also controls surrounding territory — holds a structural advantage. A Knight or Bishop on Cc1 is the ideal blocker; a pawn on Cc2 or Cc4 is acceptable but less flexible.
Are the Unicorns actively posted? Two Unicorns placed on complementary color classes — for example, Cc2 (class (0,1,1)) and Cd2 (class (1,0,1)) — together cover both Unicorn domains, projecting force along four distinct triagonals toward the opponent’s home territory. This coordinated placement is the supreme positional achievement of the Raumschach opening. Note that White’s Unicorns share their reachable squares with Black’s Unicorns — Unicorn-vs-Unicorn tactics are a genuine middlegame theme.
Which pawns are vulnerable? Due to three-dimensional capture geometry, pawns in Raumschach are more difficult to defend than in flat chess: they can be attacked from five different directions rather than two. A pawn on an exposed Level C or D square, without adjacent pawns on the same level, is a structural weakness. Middlegame strategy should target such weaknesses while avoiding creating them in one’s own position.
Raumschach has waited 119 years for its opening theory. The framework established in this paper — three foundational principles, a first-move hierarchy, eleven named openings with annotated variations (plus Joey’s Attack as a dedicated continuation), five key tactical and positional motifs — is a beginning, not a conclusion. Every line presented here requires computer verification, practical testing, and the kind of iterative refinement that flat chess theory has received over centuries of human play.
The five strongest first moves, in proposed order by principle:
A sixth line deserves mention on purely practical grounds. 1. ♕︎Bc1–Cb2 (Pluto’s Attack, the Cb5 Triple Fork) sacrifices adherence to the Level C Imperative and minor-piece-first development in exchange for a forcing tactical threat that, against an unprepared opponent, wins a Rook or Unicorn outright by move 4. Its weakness is recognizability: once the pattern is known, Black can defuse it by relocating the Rook or Unicorn on moves 1–2. But in games between players unfamiliar with Raumschach theory it is likely the single most dangerous three-move plan available to White — and it carries the distinction of being the first opening named from a recorded human-vs-engine game at raumschach.org.
The 50,000-game self-play corpus discussed in Section VII prompts four further notes, two confirming the geometric framework and two complicating it. Jupiter’s Attack (1. 🨢Bb1–Ca2) is catalogued as Opening 9 on the strength of an empirical result this paper’s geometric framework did not anticipate — the corpus’s second-highest first-move win rate, paired with a genuine spacemate mechanism once the Queen reaches the covered square at Ec4. Joey’s Attack, a bait-and-recapture combination available after the Unicorn Surge, stands as a dedicated continuation within Opening 6; the corpus’s poor score for the unprepared version of its central capture (44.1%) is the direct evidence for why the prepared version works. Against these, two newly catalogued openings sit in honest tension with the rest of this volume. The Long Ramp (Opening 10, 1. ♗︎Bd1–Db1) is the single most-played move in the entire corpus, yet its EV sits almost exactly at parity — a reminder that popularity and soundness are not the same measurement. The Queen’s Vanguard (Opening 11, 1. ♕︎Bc1–Cc2) is the corpus’s second most-played move and performs respectably, despite directly violating the Queen-development principle of Volume II and the hard rule Raumcapa enforces for exactly this reason — a discrepancy traced to the fact that the corpus predates the rule’s addition to Raumcapa’s own evaluation function, not to a different engine’s games. All four notes reinforce a point worth making explicitly: the corpus is a genuine second source of theory, not merely a confirmation tool — and it can disagree with the principled framework as readily as it confirms it.
Several research directions demand urgent attention. Computer engine analysis: dedicated Raumschach engines should verify or refute every evaluation offered in this paper. Deep variation theory: this paper carries analysis to approximately 4–5 moves; a full opening encyclopedia would carry each line to 10–15 moves with full branching. Black response catalogs: for each White opening, a complete Black response catalog should be developed; this paper provides 2–3 responses per system. The Dual Unicorn System with complementary classes: with White’s Unicorns confirmed on complementary classes (Bb1 and Be1), a dedicated investigation is needed into optimal Unicorn placement configurations — particularly the Cc2 + Cd2 coordination and the tactical implications of White and Black Unicorn domains being identical (including Unicorn-vs-Unicorn capture themes that the Unicorn’s color-binding might seem, at first glance, to rule out). Pawn endgame theory: which openings lead to winning pawn endgames for which side? The interaction between opening pawn structure and endgame outcomes deserves dedicated study. Counter-play against Pluto’s Attack: when Black relocates the Rook or Unicorn on moves 1–2 to defuse the fork, what structure results, and how should White continue? The full branching tree of the Cb5 system beyond the main line is unexplored. Deeper verification of Jupiter’s Attack: the corpus result is suggestive at depth 2; a higher-depth or tuned-weight re-run of this specific line, per the canonical tuning corpus planned for this engine, would substantially firm up its standing. Deeper verification of the Long Ramp and the Queen’s Vanguard: both stand on raw popularity rather than a verified advantage; a tuned, higher-depth re-run would settle whether either deserves a place among the five principled recommendations above.
Ferdinand Maack wanted a game that matched the true dimensionality of conflict — where attack could come from anywhere in three-dimensional space. In Raumschach, every classical principle deepens: center becomes a volume to be occupied, development becomes multi-directional, and the geometry of spacemate becomes a three-dimensional problem of cornering a king across 125 cells with no castling refuge. The theory established here is the first map of that vast, unmapped territory. May it invite many explorers, and may those explorers improve upon every analysis contained within it.