Topic: bitZino - Bitcoin Casino - Blackjack, Roulette, 3 Card Poker, Slots and more! - page 35. (Read 82361 times)

I would suggest also adding the ability to lower the maximum bet. 

We have officially launched Roulette!





I'm really proud of the interface we developed for this game. We have been trying to break away from the common paradigm of simply simulating physical games on a screen. We think we can create much better user interfaces when we utilize all the features that a computer has to offer.

For example, we have been trying to not introduce the chip metaphor into any of our games - we think it's much easier to type a number than drag an drop various chips around the screen. This did pose as somewhat of a challenge for roulette, where chips play an integral part of both how you place your bets as well as how you visualize your bets. But in the end, we think we've created a much better digital representation of roulette than any we've played. Try it out, and let us know if you agree!

Also, like all of our games, roulette is Provably Fair. We utilize cryptography to make it computationally impossible for us to cheat you. You are guaranteed a fair game!

As always, I'm here on the forums, and very receptive to your feedback. If you have any comments at all, please don't hesitate to let me know!

Heh, this is a definitely very interesting question, and I'm happy to talk about it Risk management is one of the more complicated aspects of running a casino, and I have certainly learned a lot about it since launching bitZino. Fortunately, we've got some smart minds with a background in risk management. Before launching any games we do extensive analysis of the game to ensure that our risk of ruin is infinitesimal. This process is fresh in my head because we've been doing extensive analysis of our soon-to-launch Roulette game (which poses unique challenges with its 35 to 1 bets (more on this below)).

This is a paramount process for any honest casino: if the casino's risk of ruin isn't effectively zero, then you as a player are not only playing against the house-edge, but you're also playing against the risk of the casino going broke.

The statistical math behind risk of ruin calculations is incredibly complicated. While we do utilize concepts such as binomial distributions to double-check certain assumptions, we primarily calculate our risk of ruin by running Monte Carlo simulations. A Monte Carlo simulation is effectively just simulating a random event enough times so that you have an idea of the probability of that event. If you do enough simulations you can get a very accurate estimation of the true odds. This reduces the need for incredibly  complicated math, while also providing us more testing of our actual code (since we run our Monte Carlo simulations against our actual code for the game in question).

So, what odds are we calculating when we do our Monte Carlo simulations? We're not just calculating the odds of the game - we already know that. We are calculating the odds that a player's balance gets above a certain point at any time while continually playing the game and betting the maximum.

So, for example, to set up a Monte Carlo simulation for our single-0 roulette game we create a new player, give them a balance of 0, and have then constantly place a bet of 1 on a single number (which pays 35 to 1). We have them do this 1.5 million times (which is enough for their balance to go irrevocably negative). All the while, we are tracking the maximum value that their balance gets to. We then repeat this entire process 100k times. We can then plot of histogram of the maximum balances of the player, which gives us the effective risk of ruin for various bankrolls:

Risk of Ruin for Single-0 Roulette (player betting 1 on single number)

Bankroll	Risk of Ruin
0	81.605%
1000	17.216%
2000	3.547%
3000	0.75%
4000	0.258%
5000	0.083%
6000	0.025%
7000	0.008%
8000	0.001%

This chart indicates that if a casino wants to have a less than 0.001% risk of ruin, they must have a bankroll of more than 8000x the maximum bet on a single number.

If you run the same simulation on roulette with the player betting on a 1 to 1 bet (ie, betting on red or on even), the bankroll requirement for the house is significantly lower. We find we only need 300x the maximum bet to keep our risk of ruin well below 0.001%.

I found it interesting how much of a difference a 35:1 bet versus a 1:1 bet affects the risk of ruin. If we used our risk of ruin estimation of the 1:1 bet to calculate the bankroll for our 35:1 bets (ie, if we only had 300x the bankroll of the maximum 35:1 bet), then our risk of ruin would be over 50%!

When we launch roulette, we will initially be conservative and will allow bets of up to 10 BTC on the outer bets, and up to 1 BTC on the inner bets.


Exactly! 0.001% is small, but it's not infinitesimal. In order to make our true risk of ruin truly infinitesimal, we have the ability to lower the maximum bet as necessary. We prefer not to rely on this mechanic though, which is why we shoot for a very low risk of ruin without it.


While this is true, we don't rely on this at all when computing our bankroll requirements. Because at any point a player could come and start betting the maximum and playing perfect strategy.

As you can see - we have thought extensively about this problem

That was fun to write, but now it's time to get back to launching roulette!

You can have a lot higher max bet and lower risk of ruin if you are willing to reduce it as you lose.

Also the edge in blackjack can be massive if some players play poorly.

Or he might have a massive roll :-)

I'm curious too but wouldn't have had the balls (or lack of tact -- take your pick) to ask. 

Out of curiosity - what are the maximum bets for your games? Your OP claims 10 BTC as the maximum, but the bankroll required for you to operate with people betting that much has to be massive. Is 10 BTC really the max bet and do you think that's sustainable? Do you actually get much action that high? As I said, I'm more curious (or nosy -- take your pick) than anything.

I would prefer not to rely on a third-party if possible. This may end up being one of the most practical solutions though, given the current feature-set of bitcoind.


Armory does seem quite promising, but I think they still require the entire blockchain to be stored in ram (which makes sense, how else would they be able to quickly get the balance of any address). I also am wary of the security implications of Armory. It definitely doesn't have as many eyes on it as bitcoind itself has, and just because it uses bitcoind for the underlying networking doesn't mean it is fully secure.

Thanks for the advice everyone! I definitely didn't think this feature would have so many pros and cons involved in the decisions we'd have to make! Right now, I think running 2 copies of bitcoind is sounding like the best solution, but I'm not 100% sold yet.

Armory, which uses the blockchain from bitcoind appears to provide this:


 - https://github.com/etotheipi/BitcoinArmory/blob/master/createTxFromAddrList.py

Which can be called using a script from:

 - https://github.com/etotheipi/BitcoinArmory/blob/master/extras/sample_armory_code.py

What about using blockchain.info's API to get the balance of the key?

I agree - this was what got me originally thinking about just running a second bitcoind for the purpose of importing the private keys. We could just throw away the wallet from the 2nd bitcoind periodically, or immediately after the funds are sent to the primary bitcoind wallet.


The problem is that with the standard bitcoind, it seems that you can't even figure out the balance of a private key without either importing it first, or importing the blockchain into a separate external data structure. So, since you don't know the balance, you can't really create a valid transaction


Thanks for the link! That does seem like a cool set of tools. However, the requirement to import the entire blockchain into a separate database is starting to border on overkill.

To be honest, none of these solutions is ideal. They all pose very real costs: both in terms of server resources and engineering effort. They also pose a potential security risk: I'd much rather be relying solely on bitcoind, since it is simply the most well-tested client out there.

I really do want to support importing private keys - I think the ability to for people to use bitcoins in a similar way to how they currently use paper cash is a worthy goal that will certainly help bitcoin's adoption. Ideally, I'd like there to be a sendprivatekey API call in bitcoind, which would really trivialize the whole operation.

I haven't given up hope though! I'll keep exploring all the potential options.

Well, you don't need to import the key into the wallet.  You are simply doing a one-time redeem to spend any funds on that coin.  And because the key is compromised (as it comes from an external source), I certainly would never mix it in with a wallet holding other funds.

I just assume you create a spend transaction with the output address being one that is tied to that user's account.  The Bitcoin.org client v0.7 has raw transaction support.  You could also build a transaction and use Blockchain.info's pushtx I believe.

 - https://people.xiph.org/~greg/signdemo.txt

Or use genjix's Subvertx:
 - https://bitcointalksearch.org/topic/m.604667

Thanks for keeping us on track We haven't forgotten about it, we just ran into some technical problems, which have slowed us down.

I thought this would be simple to implement by using the importprivkey API call, but it turns out that method is really slow to run, and basically locks up bitcoind during the time that it's running. It seems like the only solution may be to run multiple copies of bitcoind, or use some sort of modified client - both of which don't seem ideal. Do you have any technical advice on a good way to implement this? It seems like a lot of other services do this, so there must be a good way!

Any chance that is coming anytime soon?

Thanks you guys, I'm glad you're enjoying it!

We'll be launching roulette at the beginning of next week, which I'm really excited about! We're also getting close to launching craps.

Also, we've made some substantial changes on the backend which significantly improve our system's reliability and loss-prevention. All of our data is backed-up in realtime and stored in geographically diverse locations. So now, even if our primary data center completely falls off the map, we'll be able to failover with no data loss and minimal downtime.

Tried this out on my iPhone today with play money. Really cool.

Great site and impressive response to customer feedback. Well done!

really cool site.  Just fired off .13 bitcoins.  extremely simple, clean interface + love that it's 30 second anon account / pure bitcoin play.

glglgl

Thanks Kaji!

Stay on the lookout for Roulette and Craps - those will likely be the next 2 games we launch.

I've been playing on bitZino for a while now.  I've had to contact customer support a few times in the past for password rests and have nothing but great things to report.  Bravo on an awesome and well developed product.  My only complaint is I SUCK at BlackJack .  I still play from time to time tho.

My request for the developers.  Craps!  I would be so down for investing in on such a product for development.  Just let me know where to contribute.

-kaji

Thanks, I'm glad you like it!

We are in active development, and constantly working to make the site better. If you have any feedback for us, let me know!

Found your site recently, and I am hooked!
Its clean, easy to use, fast, and well uses BTC!

Keep up the good work!