White 476 vs Black 821
White starts from setup 476 (RBNNKRBQ), Black from setup 821 (RKBBNQRN). In the encyclopedia since 2026-11-01: neither side starts better.
|cp| ≤ 5 at 400M · 800M · 1.6B · 4B nodes, two independent engines agree
How to play with different setups
White sets up the pieces by position No. 476, Black by position No. 821, mirrored on their own side; the pawns stand as usual.
Moves as usual: White starts; everything else follows the Chess960 rules.
The two setups
White: setup #476 R B N N K R B Q
King on the e-file, rooks on a, f, queen on the h-file; no bishop starts in a corner; the knights start on central files. Castling follows the Chess960 rule: king and rook end on the classical squares. In ordinary symmetric play setup 476 gives White +0.25 (Morozov Balance Index, 400M nodes).
Black: setup #821 R K B B N Q R N
King on the b-file, rooks on a, g, queen on the f-file; no bishop starts in a corner; the knights start away from the centre. Castling follows the Chess960 rule: king and rook end on the classical squares. In ordinary symmetric play setup 821 gives White +0.23 (Morozov Balance Index, 400M nodes).
Why the pair is balanced
At the deepest rungs (400M, 800M, 1.6B nodes) the evaluation never left ±4 centipawns of equality, reaching search depth 54.
Admission rule: |cp| ≤ 5 at four budgets up to 4 billion nodes, no one-sided drift along the ladder, and confirmation by two independent engines — Berserk and Lc0 — each below its own median on symmetric positions. The same corridor is required at every deeper budget (8 G, 16 G) at which the pair has been evaluated.
The shallow rungs wandered between -50 and +15 cp before settling — a reminder that quick engine checks overstate imbalances, and why the ladder goes deep.
For comparison, in symmetric play setup 476 gives White +0.25 and setup 821 gives White +0.23: with both sides on the same rank the first move is worth something. In this pair Black's own setup absorbs that edge completely.
White is not forced into a single path: the best first move is f3, and 2 first moves are within 10 cp of it.
Along the engine's main line (30 moves) the largest swing was 52 cp (around move 11), ending at -20 cp — the position stays playable for both sides.
Engine self-play agrees: 10 games, White won 1, drawn 8, Black won 1 (White scored 50.0%).
Verification ladder
Nine independent searches of the starting position, each with a larger node budget. Evaluation in centipawns from White's point of view.
| Stage | Nodes | Depth | Eval, cp | Main line |
|---|---|---|---|---|
| S0 | 100 000 | 13 | -50 | 1. c4 f5 2. a4 a5 3. f4 e6 4. g4 Ng6 |
| S1 | 500 000 | 20 | +1 | 1. c4 f5 2. Nc3 g5 3. g4 e6 4. gxf5 exf5 |
| S2 | 2 000 000 | 22 | +14 | 1. f3 e6 2. c4 f5 3. Nc3 g5 4. d4 g4 |
| S3 | 8 000 000 | 28 | +8 | 1. c4 f5 2. Nc3 e6 3. f3 Ng6 4. Nd3 b6 |
| S4 | 32 000 000 | 31 | +15 | 1. f3 e6 2. c4 f5 3. Nc3 Bf6 4. g4 g5 |
| S5 | 128 000 000 | 38 | +14 | 1. c4 f5 2. Nc3 Ng6 3. f4 e6 4. g3 b6 |
| S6 | 400 000 000 | 41 | +4 | 1. f3 f5 2. e4 b6 3. Ne2 Ba6 4. a4 e6 |
| S7 | 800 000 000 | 45 | 0 | 1. f3 f5 2. a4 a5 3. Nd3 g5 4. g4 f4 |
| S8 | 1 600 000 000 | 54 | 0 | 1. f3 f5 2. a4 a5 3. Nd3 g5 4. g4 f4 |
Deep verification
The search ladder reaches 1.6 billion nodes; admission needs a 4-billion rung and two further engines on top of that. Beyond even that, every pair was re-checked at budgets up to ten times larger and in thousands of self-play games. This section shows what each check returned for this pair.
| Nodes | Eval, cp |
|---|---|
| 4 000 000 000 | 0 |
| 8 000 000 000 | not checked |
| 16 000 000 000 | not checked |
This pair stays within the ±5 cp corridor up to 4 000 000 000 nodes.
Independent engines, run on the same position:
- Berserk 14: 0
- Lc0 0.32.1: -15
Independent confirmations: 3 of 3. An engine counts as confirming when its reading of the pair is closer to equality than its own reading of a typical symmetric Chess960 position (+10 for Berserk, +18 for Lc0). Where only one engine confirms, the pair is balanced according to that engine alone, and this page says so.
Own weight of each setup, measured against the classical array: White +26, Black +42, difference -16. A balanced pair needs White's setup to be weaker by about the price of the first move.
Extended self-play: 10 games, 1 White wins, 8 draws, 1 Black wins, White scored 50.0%.
Best first moves
The 3 best first moves for White, searched together at 400 000 000 nodes.
| # | Eval, cp | Depth | Main line |
|---|---|---|---|
| 1 | +1 | 39 | 1. f3 e6 2. c4 Ng6 3. a3 f5 4. Nc3 b6 5. d4 Ba6 |
| 2 | -8 | 38 | 1. a4 a5 2. Nc3 f5 3. Ba2 e6 4. f3 b6 5. h4 Ng6 |
| 3 | -13 | 38 | 1. c4 f5 2. Nc3 Ng6 3. Bc2 b6 4. Nd3 e6 5. O-O-O Bf6 |
Main-line play-out
30 moves, 5 000 000 nodes per half-move. Largest |evaluation| 52 cp, mean 20.7 cp, final -20 cp.
| # | White | cp | Black | cp |
|---|---|---|---|---|
| 1 | c4 | +6 | f5 | -5 |
| 2 | Nc3 | 0 | Ng6 | 0 |
| 3 | Bc2 | -11 | b6 | -19 |
| 4 | Nd3 | -22 | e6 | -22 |
| 5 | O-O-O | -26 | Bf6 | -17 |
| 6 | f3 | -20 | Bb7 | -17 |
| 7 | Kb1 | -14 | Qe7 | -29 |
| 8 | h4 | -32 | Bxh4 | -31 |
| 9 | Qh2 | -32 | Rf8 | -27 |
| 10 | a4 | -38 | Qg5 | -51 |
| 11 | Qh3 | -52 | Bg3 | -45 |
| 12 | Bh2 | -40 | Bxh2 | -40 |
| 13 | Qxh2 | -42 | O-O-O | -38 |
| 14 | c5 | -48 | d6 | -38 |
| 15 | cxb6 | -31 | axb6 | -31 |
| 16 | Bb3 | -34 | Kb8 | -44 |
| 17 | Bxe6 | -39 | d5 | -42 |
| 18 | g4 | -41 | f4 | -23 |
| 19 | Bf5 | -28 | Nd6 | -25 |
| 20 | Nb5 | -16 | Qf6 | -19 |
| 21 | Rh1 | -14 | Rf7 | -10 |
| 22 | Rc1 | -11 | Re8 | -20 |
| 23 | Nxd6 | 0 | Qxd6 | 0 |
| 24 | a5 | 0 | c5 | 0 |
| 25 | e3 | 0 | c4 | 0 |
| 26 | Qxh7 | 0 | Nf8 | 0 |
| 27 | Qg8 | 0 | Rfe7 | 0 |
| 28 | Nxf4 | 0 | b5 | 0 |
| 29 | Qh8 | -6 | b4 | -12 |
| 30 | Qh2 | -16 | b3 | -20 |
Engine self-play
10 games at 500 000 nodes per move: White won 1, drawn 8, Black won 1 — White scored 50.0%. Average length 46.4 moves.
Sample game 1 (1/2-1/2)
1. f3 e6 2. c4 Ng6 3. Nc3 f5 4. Nd3 Bf6 5. a4 a5 6. g4 d6 7. Bc2 b6 8. Qg2 Qe7 9. gxf5 Nh4 10. Qh1 Nxf5 11. O-O-O g5 12. e4 Nd4 13. Bxd4 Bxd4 14. Qg2 Rg7 15. Nb5 Bf6 16. e5 dxe5 17. Qg3 Nd6 18. Nxe5 Bb7 19. Kb1 h5 20. Ng6 h4 21. Qe1 Qe8 22. Ne5 O-O-O 23. d4 Nxb5 24. axb5 Kb8 25. Rg1 Rg8 26. Qe3 Qe7 27. Nc6+ Bxc6 28. bxc6 Qb4 29. Qc3 b5 30. Qxb4 axb4 31. cxb5 Bxd4 32. Rg4 Be3 33. Rxb4 Rxd1+ 34. Bxd1 h3 35. Bb3 Rh8 36. Bxe6 Bf4 37. Rb3 Rh4 38. b6 cxb6 39. Rc3 Bxh2 40. f4 Bxf4 41. Rxh3 Rxh3
Sample game 2 (1/2-1/2)
1. c4 f5 2. g4 e6 3. Nc3 g5 4. gxf5 exf5 5. f4 gxf4 6. Nd3 Ng6 7. Nxf4 Nxf4 8. Rxf4 Bg5 9. Rf1 Qh6 10. Qd5 Nf6 11. Qd3 d6 12. Bc2 Bh4+ 13. Kd1 Bg5 14. Qd4 f4 15. Ne4 a5 16. Bd3 Nxe4 17. Qxe4 Bh3 18. Kc2 c5 19. e3 Bxf1 20. Rxf1 Rf8 21. h4 Bxh4 22. exf4 Ka7 23. Be3 Rae8 24. Qd5 Qe6 25. Qxe6 Rxe6 26. Bxh7 Rh8 27. Bd3 Ree8 28. f5 Bf6 29. Bf4 Rd8 30. Rg1 Rhg8 31. Rg6 Rxg6 32. fxg6 Bg7 33. Be4 Kb6 34. Kb3 Rf8 35. Bxd6 Rf2 36. Kc2 Bh6 37. Be5 Rxd2+ 38. Kc3 Rd7 39. Bd5 Re7 40. Bf6 Re3+ 41. Kc2 Rg3 42. Bf7 a4 43. a3 Rg2+ 44. Kd3 Kc7
Play it in the CH960 app
Open this page on a phone with the CH960 app installed, or find pair 476 / 821 in the app's encyclopedia, then choose how to play.
- Local game — two players on one device, with a clock.
- With a mentor — a clearly labelled training bot with hints, at your level.
- Challenge a friend — an online game by link; both players get the same pair, colours swap on rematch.
Play this pair against the computer, right here Play a friend
FEN (Shredder, castling rights by rook files): rkbbnqrn/pppppppp/8/8/8/8/PPPPPPPP/RBNNKRBQ w AFag - 0 1
Mirror twin: White 684 vs Black 138 — a different position (castling squares are fixed), not evaluated by the search.