Yes. You limit the maximum profit per bet using the Kelly criterion.

Do you pay bug bounties at all? I didn't find a way of cheating, but I did find some factual errors in the FAQ.

Please note that I do not vouch for yolodice.

Well, let's look at a simpler case:

Is there any tangible difference between:

a) betting 1 BTC at 50% (1.98x)
b) betting 0.5 BTC at 50% (1.98x) twice

Both risk 1 BTC, so both have an expected profit of -0.01%.

Let's look at the possible outcomes:

a) has 50% chance of profit 0.98 BTC and 50% chance of profit -1 BTC
b) has 25% chance of profit 0.98 BTC, 25% chance of profit -1 BTC, and 50% chance of profit -0.01

So the maximum and minimum profits are the same, but b) (betting a smaller amount, more times) fills in the middle. The extreme outcomes become less likely the more you subdivide your bet, and the expected outcome becomes more likely. In the extreme, if you bet a single satoshi 100 million times (ignoring rounding), you will be very likely to have a profit very close to the expected profit of -0.01 BTC

For the players, their only hope of winning is to beat the expectation, and so their best strategy (assuming they have a fixed amount to bet, and have to bet at a certain chance of winning) is to make a single large bet.

We could plot the outcomes against the chance of them happening. Both charts would have the same "center of mass" (ie. the expected profit, -1%), but the chart where the bets were divided up more would be more pointy, with the outcomes concentrated around the expected -1% point.

Edit: I missed the end of your question:


I think the player is at a disadvantage if you limit them to 100x. They effectively pay 1% of everything they stake.

If they can bet 1 BTC at 10,000x, their expected profit is -0.01 BTC.
But if they can only bet at 100x, they have to bet 1 BTC at 100x, then 100 BTC at 100x. That second bet has an expected profit of -1% of 100 BTC, ie. -1 BTC.

The strategy "bet 1 BTC at 100x, and if I win, bet 100 BTC at 100x" has an expected profit of -0.01 BTC for the first bet, and -0.0099 for the second bet (0.99% chance of having to make the 2nd bet, and an expect profit of -1 BTC if you have to make it, for an overall expectation of -0.0099 BTC), for a total of -0.0199 BTC. That's very nearly twice as bad for the player as if you let them make a single bet.

Update:

 1  2013-11-30  1132.29
 2  2013-12-04  1111.56
 3  2013-11-29  1065.36
 4  2013-12-03  1050.57
 5  2013-12-02  1000.49
 6  2013-11-28   998.56
 7  2013-12-05   975.06
 8  2013-12-01   945.67
 9  2013-12-10   941.56
10  2013-11-27   937.09
11  2014-01-06   936.62
12  2016-12-23   897.77
13  2013-12-06   894.36
14  2013-12-11   890.42
15  2014-01-11   883.00
16  2013-12-13   882.99
17  2014-01-05   881.82
18  2013-12-09   867.49
19  2013-12-12   862.85
20  2013-12-14   860.41

Source - click 'Load raw data' under the chart; use the last column ("Weighted Price"); don't use the last row because that's the current day

I don't think that's true. You would only need to recalculate the share for each investment tier, ie. 10 numbers.

I assumed that's what you were doing.

So your "leverage" is exactly the same as Just-Dice's "offsite investment" feature, rather than true leverage. The "loan" amount (or "offsite amount" in JD terms) is constant and doesn't grow or shrink as people play.

[Edit: I guess that also means the "Deposited" column doesn't show how much the investors deposited, but rather shows how much they have now. And the "In the bank" statistic doesn't show how much is in the bank, or how much is available to be won because most of it is made up of coins which don't actually exist (and were 'loaned' into existence). The same goes for the "over 226BTC already invested" - most of that isn't real coins, but the result of pretend loans. Right?]

Please note that I do not vouch for yolodice.

My bad. I missed a zero.

I was describing a reverse martinglae where you double on win, not on loss.

Using a reverse martingale it's possible to turn 1 satoshi into a million satoshis, by doubling 20 times. Since you would be 1000000x'ing your starting balance would you be attempting to break the rules? That was what I was asking.

But it sounds like the rules aren't really decided upon yet, so there's no point trying to get clarity until they are.

What's going on with these numbers?



Look at the 2nd row. 2 times 15.275 isn't 30.275
Look at the last row. 10 times 10.638 isn't 103.338

It's hard to figure out how the math is done. The 7x row is weird: 0.01019145 times 7 != 0.07019145
It's like you're multiplying the first few significant figures, and leaving the rest not multiplied.

Please note that I do not vouch for yolodice.

Maybe I'm misunderstanding how it works but I thought the app just does whatever the user asks it to do. The user asks the app to play two 1000x bets in a row, so the app passes the requests on to MP. How is MP meant to know whether the bets were initiated by a real user or by the app itself? And why would it care?

If MP investors are happy to bankroll a 1000x bet of 100 satoshis, paying out 100k satoshis, and also to bankroll a bet of 100k satoshis paying out 1 BTC, would they really not be willing to bankroll those two bets one after the other? It seems unfair to cut off an app because one of its users got too lucky.

Suppose I start with 100 satoshis and try to repeatedly double it on 2x bets. After my 10th win in a row I will have 1024x'ed my starting bet. Will the app have its MP service cut off because it has circumvented the 1000x limit rule? I think the new rules need to be more clearly stated.

Are you saying that if I make a small bet at 1,000x and win, then place the winnings on another 1,000x bet I would be "attempting to circumvent your rules", because the net win would be 1,000,000x?

do you call this scam FARE !!!!!!!

(just kidding)

There was only one seeding event. Only a single block hash was used.

The client seed is 000000000000000007a9a31ff7f07463d91af6b5454241d5faf282e5e0fe1b3a, as mentioned in this post.

Did you read my post which tries to make the process clearer? Does it help?

And I am not ethan_nx.

I'd also like to state that I do not vouch for yolodice.

I expect I'll need to quote that in the future to defend myself against a QuickSeller attack.

40% of the house edge goes to app owners
10% of the house edge goes to MP
so 50% of the house edge is going to the investors

of the profit investors make from their 50% of the house edge, app owners and MP each get 10%; that represents a further 5% of the house edge for each

so the investors end up getting 40% of the house edge

---

For your specific examples:

when the player wins a 1 BTC bet (at 2x) the app gets 0.004, MP gets 0.001, and the investors pay 1.005
when the player loses a 1 BTC bet the app gets 0.004, MP gets 0.001, and the investors get 0.995

but then, at the end of the week, the investors pay 20% of their net profit: 10% to the apps, and 10% to MP.

So suppose there were 1000 bets in the week. All of them were for 1 BTC at 2x, with a 49.5% chance of winning. A house edge of 1%.
Suppose 495 of them won and 505 of them lost. That's the expected result.

On each of the 495 winning bets the investors pay out 1.005; they pay a total of 1.005 * 495 = 497.475
On each of the 505 losing bets the investors earn 0.995; they earn a total of 0.995 * 505 = 502.475
Their net profit is 502.475 - 497.475 = 5 on a total amount wagered of 1000. That's 50% of the house edge.

At the end of the week they pay 10% of their profit to the apps, and 10% to MP; that's 0.5 BTC to each.
They are left with a profit of 4. That's 40% of the house edge.

Get it now?

Oh yes, that makes sense.

So the expected total commission has dropped from 70% (50+20) to 60% (45+15).

I hadn't realized before that you can't calculate the expected commission by simply summing the 'on house edge' and 'on profit' percentages.

For example, BetKing used to take 25% up front (on the house edge) and another 25% of the actual profits. I always assumed that represented a 50% expected commission (25+25), but now it seems like it was 'only' 25 + (0.75*25) = 43.75%. Do I have it right now?

So the house edge used to be split 50% to app, 20% to MP, leaving 30% to investors if the actual profit matched expectation.
And now it's split 40% to app, 10% to MP, 40% (10% more than before) to investors? Where's the missing 10% going now? (40+10+40 = 90, not 100)

Did you mean to say this instead?

> Investors will stand to make an additional 20% (50%) of the house edge.

Or did I misunderstand?

How about this for a possible explanation:

While the business is profitable, most of the profit has been from a few large whales. Whale sightings have been getting rarer and rarer. DN worries that BTC whales are going extinct, which would leave him with very little volume, and so he figures it's better to sell 30% of the business for 2000 BTC now while he can boast of relatively high profits over the last year.

That would seem to make sense. While he could end up missing out on a bunch of value if betting volume doesn't die off, it's a hedge against that eventuality.

Much the same as when the commission model switched from taking a percentage of investor profits to taking the commission up front, as a percentage of the *expected* profits. That was a hedge against the site performing badly. It turned out to be a bad decision, since the site performed way better than expected, but it did limit the site's worst-case commission take.

tl;dr: selling 30% of the business now makes sense as a hedge against possible future outcomes

I know. Rather than having the site be down for a week while I migrated the database offline I figured I could do it while the site was up and running. But I was doing it too fast and it was interfering with the bet speed.

It should all be good now though.

I'm not sure what he was asking. Maybe he's asking whether it's still possible to claim the original free CLAMs. If so, yes, it is.

It turned out it was going to take about 8 days to fix the database table that contains the old bet archive. So rather than waiting for that to happen I made a new empty bet archive table. Betting is active again now, but you won't be able to look up details of old bets until I migrate them into the new table.

That will happen gradually over the coming weeks (and hopefully not months, but we'll see...).

None of the data is lost, it just needs a little massaging.

Database rows are typically a fixed length. Why use 8 bytes per betid if you think they will always fit into 4 bytes. When you're storing 2 billion of them it makes a difference. Either way you need to decide what size each field in each database table needs to be.