Quote from: Tomatocage on January 24, 2013, 10:17:31 PM

Quote from: deepceleron on January 23, 2013, 01:12:09 PM

A "Martingale" is a failure betting fallacy, and a casino can only hope for players so stupid. It will result in a player going bust far quicker than the casino.

I can believe this. I played over a thousand rolls this past week alone in the 50% bracket. I always start out with .05 BTC, and I've gone on 7x losing streaks (Lose .05 -> .10 -> .20 -> .40 -> .80 -> 1.60 -> 3.20) about a half dozen times. However, after looking at my trade history, I've NEVER gone beyond 5 straight wins, ever. 50% my ass.

I believe you can select secrets that appear to have a lean, my own simulations definitely suggest this is possible.

Its as simple as generating secret, generating each result for the last 10 million bet txid and working out the distribution.
If numbers are on average higher than 32768 then its a keeper, if the number is not then you generate another and repeat until you do get a secret that leans the resulting hash the way you want.

Given that the SatoshiDice game you simply have to get a number lower than the number bet on it is easy to take advantage of this.
It does seem if you use a map that spreads out the results (ie even numbers win) it seems it is harder/ impossible to achieve.

Were a bitcoin casino to do this Id still argue that it is not cheating as it is not guaranteed, eg you may have a string of txid's that all pull low results despite the secrets lean.