# How to statistically beat this dice game?

There is a dice game on this site where you can bet a video game's currency in games. I was wondering if any of the more statistically minded could come up with a way of beating the system? The game's rules are as follows:

• The player may bet a certain amount of chips on rolling above or below a certain number on a 10,000 sided die, whereby the odds are the same, eg 45% chance of winning, above 55.00 or below 45.00.
• The player must make a MINIMUM bet of 2500.
• The player can set their own multiplier and odds. The relation is such that at a 2x multiplier of the bet, the chances of winning are 45% and at 5x it is 18%
• The maximum bet is such that the multipier*bet is equal to 1 billion.
• The player is every half hour given 100,000 chips to bet.

Here is my strategy that I have been using: Roll the 100,000 chips to 1 million. Using a 5x multiplier, roll until lose 12 rolls on the lowest bet possible, then bet .2% of the cash stack, doing so two times if the first lost, then doubling the previous bet and so on until profit is made. The reason I bet double every two is that, because the odds of getting 13 losses is 0.82^13, eventually a win should turn out and the risk of loss only lowers. I have made it up to 3 billion chips, then lost them all when I got up to 200 million chip bets, which was max and kept on losing them.

Assuming the system is not rigged, how would you go about beating this game consistently?

• Since the expectation of all admissible bets is negative, I'd bet as little and as slowly as possible in order to keep as many as possible of the $100000$ chips that I get for free every half hour. – joriki Mar 20 '16 at 9:09
• Your strategy of doubling your bet after loses is similar to Martingale Betting – Mark Mar 20 '16 at 9:12
• Isn't this some over-complicated variant of the this principle? If yes, it doesn't work. en.wikipedia.org/wiki/Martingale_(betting_system) – Peter Franek Mar 20 '16 at 9:14