Topic: Satoshi Dice -- Statistical Analysis - page 24. (Read 192920 times)

Lots of unprocessed bets, and my script timed out getting the result for many of them too, so there's lots of scope for error here.

Trick question: VirtEx.

So who is stacking up the most BTC? SatoshiDice or MtGox!

To make it more understandable, every time you send 1.00 BTC to Satoshidice, they send you back 0.981 BTC. If you send the 0.981 BTC to them again, you now have 0.9624 BTC. They have just added variance to this process to keep you entertained for a while. The more you bet, the more you reduce the variance and approach the above scenario. The more you bet, the more you compound your loss.

The only winning move is not to play.

And they spent most of their money again so everything is good with the world:

You're forgetting about variance. Variance is why you might not notice the house edge in the short term. There's no 50 game cut off, but as you play longer, your losses approach 1.9% of the total you bet (assuming an infinite bankroll).

A game in which you are paid double for a win and in which you have 49.05% chance of winning results in an expected win percentage of 98.1 - house odds of 1.9%.

For 50 turns of this game the 95% confidence interval for wins is 18 to 31, which is a gain of  ~ - 28% to + 24%. This interval is many times larger than the expected loss of 1.9% and so the effect of the negative expected value is very small by comparison, and not noticed. But there is still a significant gap between the bound for the losses and the bound for the gains. I think I'd notice that 4% difference over a number of 50 turn games.

After 10000 of this game, the 95% confidence interval for wins is 4807 to 5003, a gain of ~ -3.86% to + 0.06%. Still an approximate difference of 4%.

The house edge is there. You might not notice it, but it has an effect on each and every roll you make.

For the sake of this argument, lets say SD's house edge is 2%.

For every 100 games played, you win 49, you lose 51. Therefor, for every 49 games played, you win 24 and lose 25 (Give or take .1 of a win lol). That's one extra loss than win every 50 games on average. In 50/50 gambling, in theory, you will not make any profit or loss. So for 49/50 of the rolls you do with SD, their house edge is not affecting you what so ever. If you're betting an average of 1% of your money every time you bet, it's going to take 1000 rolls for SD's edge to take 20% of your money. Therefor, with an average of 1% of your money (Not compounding) gambled each time, it'll take 5000 rolls before you are 'cleaned' due to SD's house edge.

Obviously the 50/50 can have you go either up or down a lot during those 5000 rolls. But that's 50/50, and has no EV apart from entertainment.

With such a low house edge as 2%, the variance involved when gambling is still massive. The player has enormous chances at making a lot before house edge even comes into play.

It's unbeatable when you look at everyone who has ever gambled there, and then add all their bets together. That's when house edge starts really showing. But to the individual bettor who's looking for a bit of fun, it's more or less unoticeable.

I just feel as if you guys are making it out to be a lot bigger than it really is (The house edge).

This sounds quite dubious to me. Can you prove your assertion?

It can be beaten . 1.9% house edge is TINY. Chances are, if you played less than 50 games of SD, their house edge hasn't even affected you, and you were playing 50/50.

It's only people who play THOUSANDS of games that end up losing everything due to house edge.

Some casino's take up to 10% margins, SD takes 1.9% which is extremely fair.

Obviously though, in the long run they profit, but so do a lot of other people

Wish I had $20k to blow playing SD.

I take that back, looks like they won one:


199.88505000 BTC on a .05 lessthan 16 bet.

Looks like there is a bot playing from this address 1MMqBiVZiciztWqL635TLVMnTXeqSQ8jLU.

blockchain.info shows hundreds of low probability bets in just the past hour or so.

If one of them hits, say bye-bye to all the profits so far this month.

I do really think some of those big wins of last month coming from double spending exploit, forcing sdice to wait for 1 confirm. Bad luck shouldn't happen so often from now on 

Dont you have played lottery never ?

Because humans can decide to be no better than this:

why people smoke, you can't beat cancer ?

Just curious, ¿why people play a game that CAN'T be beaten?

A pretty quiet 24h: