Just-Dice.com : Invest in 1% House Edge Dice Game

[stripykitteh]

Correct me if I'm wrong, but with 550,000 btc gambled, the expected house take right now is about 5500 coins, right?

Has anyone calculated out standard devs and such?

In lieu of detailed stats on bet sizes and such, I can tell you that the site was about 1800BTC up when turnover was at 180KBTC, about a week ago. I think until that point almost all of the betting was from small fish.

Since then several whales have been betting big and doing well.

You might be interested to see the results of this simulation that I ran a few days ago. It basically takes as a starting point that there is a whale with a bankroll comparable to the house's bankroll, and the whale bets the maximum until one side or the other goes broke (i.e., the most extreme and aggressive whale scenario theoretically possible).

https://bitcointalk.org/index.php?topic=242962.msg2762706#msg2762706
https://bitcointalk.org/index.php?topic=242962.msg2765772#msg2765772

It is not uncommon for the house to go way, way down to such a whale (like 80% or more). So the house being 5,500BTC down vs expectation isn't particularly shocking, if most of the turnover was coming from max bets, or close to max bets (without seeing the stats no-one can say if this is true or not). The good news is the house always wins in the end.

[dooglus]

Is a list of (bet size) and (number of times that bet size happened) enough?

Or do you want it grouped by (bet size) and (chance)?  Or even by (bet size) (chance) and (win/loss)?

I'm not sure what you need exactly to work out whatever you can work out.

I made 3 dumps of the database.  "bet" is in satoshis, and chance is 1e4 times the percentage chance.

Here's the first 10 lines of each of them:

https://just-dice.com/bets-by-bet-and-chance.zip - 5804 bets of 0 satoshis on 0.0001%, etc:

bet   chance   count(*)
0   1   5804
0   3   4
0   4   8
0   10   231
0   20   1
0   40   102
0   45   17
0   99   50
0   100   4355

https://just-dice.com/bets-by-bet.zip : 1950643 bets of 0 satoshis, etc:

bet   count(*)
0   1950643
1   2026402
2   589524
3   213438
4   348768
5   215525
6   123034
7   82488
8   370996

https://just-dice.com/bets-by-chance.zip : 1006277 bets at 0.0001% chance, etc.

chance   count(*)
1   1006277
2   944
3   221
4   332
5   1432
7   69
8   164
9   801
10   26995

[dooglus]

As for the poor dude who claims to have lost his bitcoins or his phone compromised, that's his fault.

He asked me to address you:

Send a mass message to investors saying this.
As I don't really feel like giving out my forum name to the public on there.

" I contacted the Mcdonalds that I was at, and they said that nothing was caught on tape, however the phone WAS later returned that SAME day. I called the last 3 places and the 2nd was Mcdonalds who did have my phone. Upon arrival I checked wallet and found my bitcoins gone. I believe I have like .004 left. Anyways, the only transaction I saw was a 30 BTC to one I didn't recognize. I checked all over the phone and found in the search history that he went on just dice. So yeah, I don't really have anything to lie about? I mean, I have a family to take care of, I know investors recently had quit a hard loss with that one cici person winning alot. If I ever did gamble on this website(wont ever happen, after what just occured) or any other website I would NEVER go in with ALL my money. That is crazy, as that was pretty much a huge investment flip for me. even a 50% portion of the money( i understand its coming from your pockets) would do wonders, as any money getting back is really nice. I have just set a lock on my phone, along with deleting google chrome, and found a new alternative web browser that does not save passwords"

This guy is still hassling me pretty regularly trying to get a refund for the lost bets.

Most recently he added me on skype posing as a camwhore.  Either that or he really is a camwhore, I'm not sure which.

[Logik]

Get his name and address and report him to the police? Dude is clearly a crazy scammer. You aren't his personal insurance service, or private investigator. 

[Bowjob]

I think the real question is.. what is dooglus doing on OkCupid.

[petrescuerz]

I think the real question is.. what is dooglus doing on OkCupid.

Yes, that is curious, isn't it Bow!   

Doog?

[dooglus]

I think the real question is.. what is dooglus doing on OkCupid.

Yes, that is curious, isn't it Bow!   

Doog?

Probably that's a different dooglus.  It's a very common name?

[bitfish]

Is a list of (bet size) and (number of times that bet size happened) enough?

Or do you want it grouped by (bet size) and (chance)?  Or even by (bet size) (chance) and (win/loss)?

I'm not sure what you need exactly to work out whatever you can work out.

Hi dooglus,

best thing would be a sizeable database dump (~ 50000 bets) of single bets as they were done in sequence:

some_player_id   lucky_number   target   bet   payout   profit

A cumulated list of bets is omitting important  information.


Enjoy!
bitfish

[Rannasha]

Is a list of (bet size) and (number of times that bet size happened) enough?

Or do you want it grouped by (bet size) and (chance)?  Or even by (bet size) (chance) and (win/loss)?

I'm not sure what you need exactly to work out whatever you can work out.

Hi dooglus,

best thing would be a sizeable database dump (~ 50000 bets) of single bets as they were done in sequence:

some_player_id   lucky_number   target   bet   payout   profit

A cumulated list of bets is omitting important  information.


Enjoy!
bitfish

If the goal is to calculate the expected standard deviation on the house profits, then the info provided by dooglus is sufficient. If my napkin-math is correct (which is not certain considering the time my kid woke me up this morning) then the standard deviation is given by:

where i iterates over all the lines in the bets_by_bet_and_chance file, c_i, p_i and b_i are the count, chance and betsize respectively.

(edit: there should be brackets around 0.99 b_i)
(edit 2: equation is not correct, see further post for fixed version)

[bitfish]

If the goal is to calculate the expected standard deviation on the house profits, then the info provided by dooglus is sufficient. If my napkin-math is correct (which is not certain considering the time my kid woke me up this morning) then the standard deviation is given by:

where i iterates over all the lines in the bets_by_bet_and_chance file, c_i, p_i and b_i are the count, chance and betsize respectively.

(edit: there should be brackets around 0.99 b_i)

But if you want to check whether RNG  (and site) is working properly it is not enough

[Rannasha]

If the goal is to calculate the expected standard deviation on the house profits, then the info provided by dooglus is sufficient. If my napkin-math is correct (which is not certain considering the time my kid woke me up this morning) then the standard deviation is given by:

where i iterates over all the lines in the bets_by_bet_and_chance file, c_i, p_i and b_i are the count, chance and betsize respectively.

(edit: there should be brackets around 0.99 b_i)

But if you want to check whether RNG  (and site) is working properly it is not enough

To check the "RNG" you can simply implement the algorithm yourself and run a huge number of test-runs.

My equation isn't entirely correct by the way (which is a friendly way of saying that it's wrong). Now that I'm running the data provided by dooglus, the obvious errors become apparent.

[Rannasha]

Allright, I believe that the correct equation for the JD-standard deviation is:



For an individual bet, sqrt(p * (1 - p)) is the expected standard deviation of the chance to win. The expected standard deviation of the return is obtained by simply multiplying by the expected return, which is 0.99 * b.

Summation of standard deviations for independent random variables is done through taking the root of the sum of the squares. Each term is multiplied by how often it occurs (c).

That brings me to the numbers obtained from the dataset posted by dooglus:
# of bets = 33,257,221 (currently posted on site: 33,524,569)
Total BTC bet = 556,351 (currently posted on site: 557,909)
Expected BTC return for gamblers = 550,787
Expected house profit = 5564 (if only ...)
Expected house edge = 1.0000881%
Expected standard deviation of amount returned to gamblers = 2754

Current house profit is at -228 BTC, which is 2.1 times the standard deviation away from the expectation value. If the profit follows a normal distribution, the probability of this deviation occurring through sheer chance is approximately 2.5%.

Conclusion: The site has been very unlucky, but the deviation is not large enough to be implausible.

[dance4x]

site seems to be down right now...

[BigBitz]

site seems to be down right now...

down for me also

[molecular]

Current house profit is at -228 BTC, which is 2.1 times the standard deviation away from the expectation value. If the profit follows a normal distribution, the probability of this deviation occurring through sheer chance is approximately 2.5%.

Thanks for calculating this!

This reminds me of the times when pools had streaks of really bad luck (with similar probabilities) and people started accusing pool operators of cheating.

As probably everone here knows, we are trusting dooglus to not use his knowledge of the secret random seeds to predict bet outcomes and play anonymously on his own site, reaping "lucky" profit.

The site being down doesn't help confidence

Just to be clear: I'm not making any accusations and 2.5% is not unlikely.

[bitfromit]

site seems to be down right now...

down for me also

also.

[dance4x]

site is back up.

[infested999]

Code:

Your IP address has been blocked for an hour for suspicious activity.  Sorry for the inconvenience.

[qxzn]

Allright, I believe that the correct equation for the JD-standard deviation is:



For an individual bet, sqrt(p * (1 - p)) is the expected standard deviation of the chance to win. The expected standard deviation of the return is obtained by simply multiplying by the expected return, which is 0.99 * b.

Summation of standard deviations for independent random variables is done through taking the root of the sum of the squares. Each term is multiplied by how often it occurs (c).

That brings me to the numbers obtained from the dataset posted by dooglus:
# of bets = 33,257,221 (currently posted on site: 33,524,569)
Total BTC bet = 556,351 (currently posted on site: 557,909)
Expected BTC return for gamblers = 550,787
Expected house profit = 5564 (if only ...)
Expected house edge = 1.0000881%
Expected standard deviation of amount returned to gamblers = 2754

Current house profit is at -228 BTC, which is 2.1 times the standard deviation away from the expectation value. If the profit follows a normal distribution, the probability of this deviation occurring through sheer chance is approximately 2.5%.

Conclusion: The site has been very unlucky, but the deviation is not large enough to be implausible.

Thanks for this analysis! Can any other statistically minded folk comment / confirm?

[dance4x]

I am currently looking for a way to borrow from my investment in Just Dice. Here is what I propose:

1) I pay you $50.
2) You create a Just Dice account and put 17 bitcoin in it.
3) 30 days later I buy the total value of that Just Dice investment from you for 17 bitcoin. You keep the $50.

This is a 40% annual return for whoever wants it. You will need to have some kind of reputation to be a part of this deal. PM me if interested. Thanks.