Topic: Satoshidice house edge (Read 1178 times)

The number of bets isn't important, because the vast majority of the bets are relatively small, and have little effect on the bottom line.

Satoshi Dice's overall profit is bigger than it 'should' be thanks to a very rich player who bet huge amounts of money on the site earlier this year and ended up stopping after a bad run of luck.


It's possible to run 'provably fair' poker games using a system where each player encrypts each card in the deck and shuffles it before passing it on to the next player.  That way nobody can see cards they aren't supposed to.  You can read about it here: http://en.wikipedia.org/wiki/Mental_poker

One problem with such a scheme is that if any player drops out of the game mid-hand then the game stops, because every player is needed to decrypt the cards.  Another problem is the bandwidth needed to pass all the cards around continually for encryption.  I think.  Maybe it was CPU, I forget...


There is a 1-in-2^256 chance of finding another secret with the same hash each time they try.  That's as close to 0-in-1 as you can imagine.  It would take trillions of years to cheat a single time.  It's harder for SatoshiDice to cheat by finding an alternative daily secret than it is for them to break the security of Bitcoin itself.

Edit: Maybe I'm not being clear.  They have a lot of combinations to run through, but *none of them have the correct hash* so we would know immediately that they're cheating if they used a different secret than the one that has the hash they published in advance.

If your talking about a roulette wheel, it won't be perfect (maybe .0001% off after infinite spins), but that would be to do with a slight flaw with the wheel.  If you're talking about computer calculations, they should be perfect (definitely not .189% off!).

They may have published the hashes but they haven't published the secrets which look like this: P2BhTxFmqnRtmjAFM3nWGdOAuHNvDv9YBazK7jpiXIhdMtCyLapmiJwLL6QV1cCF meaning they have quite a lot of combinations to run through to get the most beneficial result for the house.



Still, when you add time into the equation you can then establish that it isn't really mathematically proven/honest (what I have come to realize)(citation needed)

Oops, I was talking about the odds of cards being drawn.  say the odds of drawing pocket aces is 221:1 (.45%).  I'm saying that after 10,000 hands, you will see that you have drawn pocket aces almost exactly .45%.  The more hands you play, the closer to .45% it should be

Still, that's only a 10% increase in profit, for the risk of having their scam exposed to the world?  It's not worth it, is it?
It does seem like a lot though.. in my experience, after many many online poker hands, actual statistics always reflect the odds with much more accuracy when you get up to 10,000 hands.  And how many satoshi bets have there been?  damn.. come to think about it, 0.189% is quite a bit..

Well, the house edge is 1.9% so it's an extra 0.189%...


Seems like a lot of steps to go to but I guess that could work.

Just curious, no agenda, not trying to disprove them in public they are pretty cool.

However the fact that most people think that they are so bad for laundering is what makes them so great for laundering, one could separate the sum of money he would like to launder into dozens of even amounts (mix them beforehand for added security?) then submit them systematically on a low chance winning bet since statistically you would win every X amount of bets, then when the question arises where that individual has received the funds he could confidently claim that he had won all of them while not mentioning all (or most) of the losing bets and mentioning all the winning bets of course.

Then would both render the coins origin as explainable (e.g. one could claim to be a professional player or a genius mathematician after prolonged use) and untraceable to the real origin.

they must be pretty desperate for that extra 0.089%!!! 

Their daily secrets are hashed with sha256.  That's currently thought to be impossible to find collision for.  The 48 hour delay is irrelevant.  They've published the hashes of the secrets they'll use for the next 10 years.  They could start working today on finding an hash collision for the secret they'll use in a year's time.  But they'll have to use the same fake secret for every bet they process on that day, and due to the large number of bets they get both the real and the fake secret will likely give them something like 1.9% profits on the day.  Not that they can find such collisions anyway.  If they could, they could easily just steal your bitcoins by making a private key whose public key hashes to the same as yours, and then spend directly from your wallet.  ie. they can't.

And as for money laundering, satoshi dice is terrible for that.  Every amount you win has your original bet as one of its inputs.  So every payout is tied directly to the bet.  That makes it the worst possible site to use for laundering.  If you want to launder bitcoins just send them to mtgox and then withdraw them again.  Nobody but mtgox themselves will be able to tie your deposit to your withdrawal, especially if you make the amounts different.  Deposit 10 BTC and withdraw 4.5 and 5.2 BTC say, on different days.

So far you've raised 3 separate and easily disproved complaints against satoshi dice.  What's your agenda here?

Whats also interesting is that they claim "fair play" yet they sometimes delay payouts up to 48 hours which then gives them the option of bruteforcing different secret phrases for the maximum profit to the house. Also one could easily launder money if one was inclined enough.

As others have pointed out, house edge is not the same as actual profits, and also neither is anywhere near 80%.

As of today SD has returned just 2.089% less in payouts than they've taken in bets.  That figure is slightly more than the house edge says it should be, but they've been lucky.

See https://bitcointalksearch.org/topic/m.2078900 for today's report.  I update it in that thread every day.

Yes but the idea that as time goes by the total amount of funds won would get closer and closer to 98% of the total amount of funds bet.

does vegas lower their edge after being around a long time
i wouldn't

Yes it is, but if you take the sum of bets 3557272/4447462=0.7998 then 0.7998*100=80% so wouldn't it become closer to 98% on a long enough period?

I believe their house edge is currently 2% actually.

How come the satoshidice house edge is currently 80% even though they have been around for so long?