[gus_massa]

In Chess960, castling has some weird rules. I don't like them, but people would expect those rules. (I'm not sure about what happens when the rock and king are not in the last rank. Probably not castling is fine.)

About pawns, it's strange because if it starts at the the first rank it can move 1-2-1-1-1-... What about no pawns in the first rank?

Anyway, the game is fine as it is and I'm a professional chess player and I'm not sure if they can be added to the engine.

PS: Another win and a few draws (and a lot of loses) with: 12 paws in the white squares in ranks 2,3,4 and 9 bishops in the black squares. I'm trying to swap a bishop for a rook or knight or a bishop of the other color, to avoid stupid draws.

PS2: I guess 8 pawns in the second rank and 8 bishops in the third rank and something to complete would be nice, but I had not tried it.

[gus_massa]

Too late to edit:

> I'm a professional chess player

I'm NOT a professional chess player

[chess39]

Spot on with castling and pawns! One idea is to allow pawns to jump from 1 to 3, I believe that’s how it’s done in horde.

Nice setup ideas! Beating the bot is quite hard, so maybe you are professional? :) If you want to make the bot even smarter, you can increase depth and think time in settings

[gus_massa]

Some will guesses, because guessing is free:

* I like the idea of using a lot of Bishops to snap a Queen or Rook in the first move. That's all my plan! And then try to lose material slowly, and win before the computer can compensate the difference.

* Looking at one of the fists game of Starcraft of AlphaStar, I guess that a lot of Bishops are overpowered for the computer, but difficult to use by a human. They are nice for the first few moves, but later they are too difficult to use for humans.

* A set of 13 Knight is probably better for the computer. It may make Magnus cry like a baby.

* I'm still surprised that Stockfish is so good in ridiculous scenarios.

[gus_massa]

* 13 Bishops at the 3rd and 4th rank + King at the 1st rank is very good. I'm getting probably a 75% winrate :) . Enough to be (slightly) angry when I lose or draw, instead of super mega happy when I ever win with the old strategy.

It depends on the current computer method that usually puts a Queen and a Rook on a vulnerable starting position, so I guess it can be counter.

Anyway, the bot plays badly when it is hopeless, like after loosing 2 Queens in exchange of 2 Bishops. At my level of play, the opponent can always make a mistake, so even in stupid bad situations a good idea is not to trade material when loosing. I'm not sure if it's possible to modify the bot to assume there is always a hope. (Anyway, the bot target my Bishops of the color with the smaller amount. If there are only one color of Bishops it's a draw.)