Bd3: Average probability: 0.655204
Qe2: Average probability: 0.613907
Resign1: Average probability: 0.077202

Bd3 wins.

@harfe can you N/A this and resign? thanks.

@deagol yes thanks for reminding me

good defense all 👍

@deagol what's up my friend

@wadimiusz sup

@deagol can you explain what you were trying to do? i don't really know about market manipulation and shit, and i'd like to learn. i'm assume one doesn't ask a real trader that, but hopefully you can explain a bit if we just trade fake internet money?

@wadimiusz i was propping this up so to try force the other move higher (I benefit the higher the score given my 200 YES on the previous move). And I could use all my YeS on this crappy move to dump against others who might want it to win. I was aiming for ~0.7 score, while holding NO @67% on the other move (thinking ahead). Last minute dump is just in case not tying my capital here

@wadimiusz I’m pretty sure this won’t average higher than the other, so resolves N/A and all profit/loss returned, as if nothing ever happened here

sorry, newb question here: is that normal when a market bounces up and down like a yoyo

@wadimiusz depends who wants it more i guess

@wadimiusz Well, if you can predict the bouncing you can profit from it.

@wadimiusz just notice these rules for Manifold Plays Chess game 3 are very particular to this structure @harfe came up with. Most other manifold markets have much simpler Y/N resolution, yet they can go also yoyo like crazy. Usually, if the resolution is related to a real world event, then you should be able to profit from some who hold crazy convictions.

so weird that the one most interested in this winning didn’t bet a single mana here

@JuJumper

Btw, why don’t you just ask to predict the probability of the market “Will white win game 3?” conditional on the specific move being made?

@wadimiusz that might be a good idea for the next iteration of Manifold Plays Chess. I understand that @citrinitas has similar ideas.

@wadimiusz @harfe I like that I can bet on a move’s merits independently of the game.

I understand the look-forward aspect of the winning conditional resolution in theory should make the market predictions back-propagate from the game result, but that doesn’t mean they should be equal at all times.

@wadimiusz if you do that, once the move is made you end up with a long tail of identical markets. If you keep them open, tons of arbitrage loss. If they're closed, casual bettors end up with a ton of capital locked in over a 20+ move game, which people don't like and also more risk of creating perverse incentives

I messed around with local stockfish and it seems to think this is a much worse move 🤔 That after 6. Bd3, white and black are more or less equal, but after 6. Qe2, the position is evaluated at -65 centipawn (I analyzed with depth 25).

I don't completely understand, but we seem to lose a knight soon, unless the opponent screws up? (I understand that we expect SF3 to screw up, but one shouldn't rely on that, right?)

Also, the WDL stats seem to indicate that after this sequence of moves, black won many times, but white didn't win a single game:

>>> from stockfish import Stockfish
>>> stockfish = Stockfish("/usr/games/stockfish")
>>> stockfish.set_position("e2e4 e7e5 f1c4 g8f6 b1c3 b8c6 g1f3 f6e4 c3e4 d7d5 d1e2".split())
>>> stockfish.get_wdl_stats()
[858, 142, 0]

(These three numbers are wins, draws, and losses for black)

@wadimiusz I see Qe2 much worse, loses the bishop with a loss of tempo or 2 after 6 …dxc4 7.Qxc4 Be6 followed by f5, black keeps developing while we retreat, eventually castles queenside, tons of center pressure, they end with a much stronger position, nice pawn structure with the B pair, all worth more than a pawn maybe 2.

Yes currently black’s d pawn is forking our B and knight, so it’ll capture one of those. Remember we took black’s knight at e4, so Bd4 equalizes, just traded knights.