This is my attempt to find Chess variants which are balanced while not being drawish: Draws are rare, and white does not have a significant advantage.
World chapion #3, Capablanca, proposed a 10x8 variant with two extra pieces: One that moves like a rook and a bishop, and another that moves like a knight and a bishop. Since there, there have been various proposals to change the opening setup of Capablanca.
Here is some of my study of these variants, using Fairy-Stockfish.
It’s actually possible to run commands directly from Fairy-Stockfish’s command line.
Let’s look at a variant:
xboard
variant capablanca
setboard ranbqkbncr/pppppppppp/10/10/10/10/PPPPPPPPPP/RANBQKBNCR w KQkq - 0 1
d
analyze
First number is plies. Second number is white’s advantage, in centipawns. Third number is elapsed time (centiseconds), fourth number is nodes searched.
As per https://maradns.blogspot.com/2009/10/capablanca-opening-setup-research.html here are some other set ups to look at:
setboard ranbqkbncr/pppppppppp/10/10/10/10/PPPPPPPPPP/RANBQKBNCR w KQkq - 0 1
setboard rqnbakbncr/pppppppppp/10/10/10/10/PPPPPPPPPP/RQNBAKBNCR w KQkq - 0 1
setboard rqnbckbnar/pppppppppp/10/10/10/10/PPPPPPPPPP/RQNBCKBNAR w KQkq - 0 1
setboard ranbckbnqr/pppppppppp/10/10/10/10/PPPPPPPPPP/RANBCKBNQR w KQkq - 0 1
setboard rcnbakbnqr/pppppppppp/10/10/10/10/PPPPPPPPPP/RCNBAKBNQR w KQkq - 0 1
setboard rcnbqkbnar/pppppppppp/10/10/10/10/PPPPPPPPPP/RCNBQKBNAR w KQkq - 0 1
Since Fairy Stockfish is deterministic, playing 1000 games will just result in the same game being played 1000 times. So, we will just perform a 30-ply analysis of the above six opening positions.
RQNBAKBNCR (+8% White, 13% Draws)
30 60 37392 81774242 42 218694 0 e2e4 e7e5 c2c3 h8g6 f2f3 d7d5 (etc.)
RANBCKBNQR (+8% White, 10% Draws)
30 71 36180 78484478 44 216923 0 d2d4 h8g6 e2e4 e7e5 d4e5 f7f6 (etc.)
RCNBAKBNQR (+16% White, 10% Draws)
30 75 26930 56640413 43 210323 0 e2e4 f7f6 d2d4 e7e5 f2f3 h8g6 (etc.)
RANBQKBNCR (+3% White, 12% draws)
30 87 43785 99279922 46 226740 0 d2d4 f7f6 e2e4 e7e5 d4e5 h8g6 (etc.)
RCNBQKBNAR (+5% White, 12% draws)
30 87 41632 86329874 47 207362 0 f2f4 f7f5 i1h3 c8d6 e2e4 f5e4 (etc.)
RQNBCKBNAR (+8% White, 10% draws)
30 114 30852 67721988 46 219499 0 i1h3 e7e6 e2e4 i8h6 f2f4 c7c6 (etc.)
To the right of each setup is the white advantage and draw percentage
from my 2009 research which used Joker80 (m here is the Rook + Knight
piece):
Setup Wins Losses Draws White Games played
ranbqkbnmr 46% 43% 12% +3% 1010
rmnbqkbnar 47% 42% 12% +5% 1017
ranbmkbnqr 49% 41% 10% +8% 1002
rqnbakbnmr 48% 40% 13% +8% 1006
rqnbmkbnar 50% 38% 11% +12% 1004
rmnbakbnqr 53% 37% 10% +16% 1011
Numbers may not add up to 100% because of rounding
As a point of comparison, here is what Fairy Stockfish has to say about standard western Chess:
RNBQKBNR
30 67 22524 58470334 44 259590 0 d2d4 g8f6 c2c4 d7d5 c4d5 c7c6
(Queen's Gambit Declined: Marshall Defense, Tan Gambit)
Let us start up the Fairy-Stockfish program from the command line. We should see something like this:
Fairy-Stockfish 11.2 LB 64 by Fabian Fichter
Let’s start off by typing this command:
setoption name MultiPV value 7
This instructs Fairy Stockfish to tell us the seven best possible moves for any given position.
We can also, if we have the memory, increase the hash size. For example, since my computer has 32 gigs of memory, we will use 16 gigs for the hash table:
setoption name Hash value 16384
Use a smaller number than 16384 if one’s computer has less memory.
Then, type in xboard and then variant chess to get in “xboard” mode
to play a game of standard FIDE chess.
The variant chess command gives us this output:
setup (PNBRQ................Kpnbrq................k) 8x8+0_fairy
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
piece P& fmWfceFifmnD
piece N& N
piece B& B
piece R& R
piece Q& RB
piece K& KO2
Then we type in d for “display board” and see this:
+---+---+---+---+---+---+---+---+
| r | n | b | q | k | b | n | r |8
+---+---+---+---+---+---+---+---+
| p | p | p | p | p | p | p | p |7
+---+---+---+---+---+---+---+---+
| | | | | | | | |6
+---+---+---+---+---+---+---+---+
| | | | | | | | |5
+---+---+---+---+---+---+---+---+
| | | | | | | | |4
+---+---+---+---+---+---+---+---+
| | | | | | | | |3
+---+---+---+---+---+---+---+---+
| P | P | P | P | P | P | P | P |2
+---+---+---+---+---+---+---+---+
| R | N | B | Q | K | B | N | R |1 *
+---+---+---+---+---+---+---+---+
a b c d e f g h
Fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
Sfen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR b - 1
Key: 9160F14A6C929FCE
Checkers:
Then analysis, and after about a day we get something like this:
1 110 0 119 1 39666 0 e2e3
1 102 0 119 1 39666 0 e2e4
1 95 0 119 1 39666 0 d2d4
1 79 0 119 1 39666 0 d2d3
1 70 0 119 1 39666 0 b1c3
1 63 0 119 1 39666 0 g1f3
1 25 0 119 1 39666 0 b2b3
[Lots of lines removed; following sorted by strength]
41 48 2757150 8488910265 57 307887 0 d2d4 g8f6 g1f3 d7d5 c2c4 e7e6 (etc.)
41 36 2757150 8488910265 57 307887 0 g1f3 d7d5 e2e3 g8f6 b2b3 c8f5 (etc.)
41 33 2757150 8488910265 62 307887 0 c2c4 e7e6 b1c3 g8f6 d2d4 f8b4 (etc.)
41 33 2757150 8488910265 69 307887 0 e2e3 d7d5 d2d4 g8f6 g1f3 c7c5 (etc.)
40 33 2363202 7271425238 61 307693 0 e2e4 c7c5 g1f3 e7e6 c2c3 b8c6 (etc.)
41 28 2757150 8488910265 65 307887 0 g2g3 d7d5 g1f3 c7c5 f1g2 g8f6 (etc.)
40 28 2757150 8488910265 58 307887 0 c2c3 d7d5 d2d4 g8f6 c1f4 e7e6 (etc.)
The only fields we will look at are the first one (the search depth, in plies), the second one (how strong the current player’s advantage is, in centipawns; when the other player is ahead, this is a negative number), and the fields showing the move on the right hand side.
Fairy Stockfish thinks 1. d4 then 1. Nf3 are White’s strongest moves, which is consistent with classical Chess theory. However, it considers 1. e4 somewhat weaker than 1. d4, about as strong as the English (1. c4), and thinks 1. e3 is just as strong as 1. e4 (???).