If I had 20k BTC in a Bitcoin Max account, I either end up with 20k BTC if pirate pays out, or 0 BTC if pirate defaults.

Apparently I could sell it for 70% of its value.  That's what people are paying for them.  So I could sell it for 14k BTC.

If I then took 10k BTC worth of bets at 100% (x2) evens 100% ROI like Matthew is trying to do, I would end up with 14+10k = 24k BTC if pirate pays out or 14-10k = 4k BTC if pirate defaults.

Either way I end up with 4k BTC more than if I hadn't sold my account and made the bets.

I'm not sure about these statistics Erik.

•   Since Satoshi Dice started, there have been:
Blockchain Tx:  3365999  :  SatoshiDice Tx: 1943703  (57.7%)
Blockchain MB:  1405.6  :  SatoshiDice Tx: 795.6  (56.6%)

i.e. even since Satoshi Dice has been running it has 'only' been 57.7% of the transactions, so there's no way it's over 66% of all transactions ever.  Also, the blockchain is about 2600 MB and Satoshi Dice accounts for 795 MB of that.  That suggests that Satoshi Dice represents roughly 30% of all transactions ever, not 66%.

•   What is a 'top address'?  There are so many ways you could sort the addresses.  I guess you mean by number of transactions involving the address, but that's not clear.  Maybe by balance held, total coins moved through the address, etc.

•   I thought the millionth bet was in the 4th month not the 3rd.

•   9,300 bets per day times 31 days per month times 3 months is only 864,900 bets.  So how do you get a million bets in 3 months at 9,300 bets per day?

The performance of the site is certainly very impressive, but it would be better to have your statistics be correct too.  There's no need to exaggerate them.

The opposite had occurred to me in the past, that perhaps the sudden staggering losses were the site's way of hiding their obscene profits by deliberately winning 'from themselves'.  That wouldn't make any sense to do when trying to sell shares though of course.

Is this bounty still available?  It wouldn't be hard to implement if there's still interest in it.

Their profit from bets, after subtracting fees is 7333 BTC.  I don't know what they started with, or how much they've paid out on advertising, development, web hosting, etc.

From the blockchain, that's all we can tell.  We can't tell what the real profit is, since there are likely to be lots of hidden (to us) expenses.

If you send them one after another, then they are completely independent events.

What you can do, however, is send multiple bets *at once* by clicking 'add recipient' on the satoshi client's 'send' tab.  If you do that, then all your bets in that transaction get the same lucky number.

The lucky number is determined completely by just 2 things: the transaction ID of the transaction containing the bet, and the day's "secret".  If you send your bets separately, they'll have completely different transaction IDs, and so their lucky numbers won't be related.

It does appear that 'luck' runs in streaks though doesn't it.  I have to believe that's just an illusion though - that we focus on winning and losing streaks and ignore the win/lose sequences that are less patterned.

I always thought it was "I love black pussy".

I'm probably wrong though.

And the high-rollers have just become profitable for SD overall too.  At last!

I think 160 bits of security is plenty for now.

And apparently the 160 bits that bitcoin addresses give us is more secure than the 256 bits that private keys give us because the advent of quantum computing apparently would reduce the 256 bit private key security effectively down to 128 bits, whereas the 160 bit security of addresses isn't hurt by quantum computing (so long as you keep your public key private by only spending from each address once!)

That's a huge sudden profit for SD in the last 24h.  Most of the gains were made in blocks 194014 (500 BTC) and 194015 (1100 BTC).

It turns out these losses weren't the result of martingale betting.  It looks like it was someone just betting over and over on the high-reward bets, and losing every time.

The bets were:

     30 * 1 BTC on 'lessthan  128'       30

    127 * 1 BTC on 'lessthan  256'      127

    136 * 1 BTC on 'lessthan  512'      136

     22 * 1 BTC on 'lessthan 1000'       22
     20 * 2 BTC on 'lessthan 1000'       40
     20 * 3 BTC on 'lessthan 1000'       60
     50 * 5 BTC on 'lessthan 1000'      250

     20 * 2 BTC on 'lessthan 1500'       40
     20 * 3 BTC on 'lessthan 1500'       60
     50 * 5 BTC on 'lessthan 1500'      250

     20 * 2 BTC on 'lessthan 2000'       40
     20 * 3 BTC on 'lessthan 2000'       60
     50 * 5 BTC on 'lessthan 2000'      250

     40 * 3 BTC on 'lessthan 3000'      120

     40 * 3 BTC on 'lessthan 4000'      120
                                       ----
                                       1605



That's pretty unlucky not to have won a single one of those bets.  But most of them were multi-bet transactions, so it's not as unlucky as it looks at first glance.

Right.  It's the amount of profit they 'should' have made if they were really making 1% or 1.5% on turnover.  i.e. it's where the black (total profit) line should be.  When the dotted line gets steeper, it means the site is taking a larger volume of bets.  The dotted line is after fees have been paid, which is why multiplying the lower dotted line's height by 100 gets a little less than the total amount wagered so far.

I'll add that information to the daily report in future.

You're welcome.

The "lucky number" is determined from just 2 things: the transaction ID and the day's "secret".  You're allowed to put multiple bets in a single transaction, and they will all get the same lucky number.  That's probably what you saw.

Here's an example: http://www.satoshidice.com/lookup.php?tx=0f871a7bce540016cdf13e8ee0a4f92d433ee68013f9e77787590e12a2342712 - it's 6 bets all in the same transaction, so they all get the same (losing) lucky number.  Notice how the 'Bet Tx' field is the same for all 6 bets.  You can see the bet at blockchain.info here: http://blockchain.info/tx-index/15859914

If you make your martingale bets one at a time then you're unlikely (1 in 65536 chance) to get the same number on any pair of bets.  You wouldn't ever make your bets in the same transaction, since you need to see whether the first one wins before being able to decide whether the 2nd bet needs making or not.

What some people seem to do is run multiple martingales at once, for different odds.  Then you can place multiple bets in the same transaction.  But like I say, martingale is not a winning strategy, long term.

There's an added complication in that SD uses whatever secret is valid for the date they first see your bet, but I don't know when they saw each bet.  All I know is the timestamp on the block which contains the bet, which might be different, and so I may be using a different secret than SD are using.

Given that, here are the bets with magic number 0 that I've found (and also those with magic number 65535).  Check out the scary animated gif on the 0 pages!  


Some random facts, taken from the 817734 rolls I analysed.  Bear in mind that it may be slightly wrong since I can't be sure what date SD considers bets to have happened on:

All 65536 different lucky numbers have come up at least once each.
817734 / 65536 is about 12.5.  An averagely lucky number will have come up 12 or 13 times so far.
14777 is the luckiest number - it has come up 31 times.
62143 is 2nd luckiest - it has come up 30 times.
10470, 17257, and 26725 have come up 29 times each.

2436, 23243, and 60872 have each only come up once each.
14 different numbers have only come up twice each.
84 different numbers have only come up thrice each.

Number 0 has come up 10 times, as has 65535.
1 has come up 16 times.
2 and 3 have come up 7 times each.
4 has come up 11 times.
5 has come up 8 times.
6 has come up 21 times.
7 has come up 10 times.
8 has come up 16 times.
9 and 10 have come up 13 times each.  And so have 15, 18, 28, 29, 37, and many many more.

Enough trivial stats?