Topic: bustabit – The original crash game - page 114. (Read 61394 times)

what did the user confirm?
 
the red trust for you is anyway a joke cause you did a new account to hide your original account. red trust cant hurt you

can RHavar confirm that a 2x or more KC was used or not. that would help all

A user confirmed the same thing but decided to delete it.  Probably because you're abusing your trust and insulting people to try and discredit them.

You going to insult me and negative tag me to try and discredit me and not answer the question?

Be honest for once.  What's the answers?

Yes. 2x Kelly.

At 2x Kelly, who are the people that stand to make steal money? 

I'll remind you of the breakdown.  Investors = 2x Kelly.  Players = -1% House edge. devans = 0.25% house edge automatically taken.

Without knowing the house edge in this scenario, it's not possible to compute.

If you are referring to bustabit, I believe it launched such that the worst case for investors (assuming multiple account aiming for max-profit) was a ~2x kelly. After the update, it was changed such that the worst case would be ~1.5x kelly.  All of these risks were clearly documented


That doesn't impact the kelly at all, as the site is risking based on your post-dilution fee bankroll.

And there it is!

Tell me RHavar...

What kelly are investors exposed to when they receive 0.75% of the house edge but a single player constantly bets to win 1.5% of the bankroll?

When you minus the dilution fees from the original investment, what does the kelly become greater than?

With this kelly and the house taking 0.25% of the house edge automatically, who are the only people to end up with the money in the end in this model (hint: they're both criminals)?

the house should protect their Bank Roll and Investors Bank Roll IMO and give the players a fair game

I think the way that makes most sense for me, is that the house needs to sort of assume a whale has infinite money. Like if you imagine dragonmaster2's "100% risk" scenario, it's just a matter of time before the infinite angry whale is guaranteed his win.

But if you invert the scenario, it doesn't really make sense to assume you can be a gambler with an infinite bankroll. And with any finite bankroll, all you can achieve is having a very large chance of busting the house -- but you can never turn that into a "profitable" scenario.


So when the house is risking more than a 2x kelly, you have a sort of weird scenario where it's bad for the house (it'll probably go broke) but also bad for the player (it'll still have an expectation to lose)

2x kelly will have no growth or decline in the bankroll on average long term (although incredibly wild swings in bankroll are expected). Anything greater than 2x kelly, and you can expect bankroll to decline; the larger the kelly, the faster the decline.

more than 2 house edges.

please explain in numbers what is "too big Kelly Criterion"? there are KC numbers which will ruin the casino as RHavar explained very well

Also, too big Kelly criterion does not necessarily lead to casino's loss.
Usually, to attack a such buggy casino, attacker will need much more money than a casino's bankroll. With no any garanties.
I think, it's more risky for attacker, than for a casino.

Ryan had already answered the other parts.

I just read your post. It's very similar to mine. What you're missing though is that in the rare chance that the house doesn't bust it has made such a massive profit that its expected profit is still positive.


The problem with these investing games is that we don't get to "keep playing until either you or your opponent is bankrupt". We only get to play for as long as the whale wants to play. He might only make a single large bet. What if you're playing the 60:40 coin toss game against an opponent who will call "game over" at some unknown point in the future? Would that change your strategy?

And...? Don't leave us hanging! How did you resolve this? If the house expects to lose money because it's over leveraged, and the player expects to lose money because of the house edge, where is all the money expected to end up???

My guess is that the house never "expects to lose money". It's just that the probability of going bust gets higher. For example if the house has no maximum bet, and is always willing to risk its entire bankroll every roll, it will go bankrupt as soon as a suitably rich whale wins a single bet. But if the house is paying only 2x for a 49.5% bet, the house still expects to profit by 1% of the amount wagered. Consider the case where the house starts with 1 unit, and the whale bets the whole bankroll against the house up to 3 times in a row or until the house goes bust:

There's a 0.495% chance that the house goes bust on the first bet, losing 1 unit.
There's a 0.505*0.495 chance that the house goes bust on the 2nd bet, losing 1 unit.
There's a 0.505*0.505*0.495 chance that the house goes bust on the 3rd bet, losing 1 unit.
There's a 0.505*0.505*0.505 chance that the house wins all 3 bets, profiting 1 + 2 + 4 = 7 units.

Expected profit = (0.495 + 0.505*0.495 + 0.505*0.505*0.495) * -1 + (0.505*0.505*0.505) * 7 = 0.030301 units.

So while there's a 87.12% chance that the house goes bust in the first 3 bets, there's a 12.88% chance that it wins 1+2+4 from the first 3 bets, which means the expected profit is positive.

One way to think about it that's intuitive to me, is to use extremes. So same +EV of 1% house edge, but increase allowed max profit to 100% of bankroll.

What's the chance the house will be bankrupt by a mega whale after the first bet? 49.5%

After 2 bets? 74.4975%
3 bets? 87.1212375%
4 bets? 93.496224937%
30 bets? 99.999999874%

Since anyone can challenge the house for 100% of it's bankroll at any time, bankruptcy is basically inevitable.

---

Another way to think about it that may be more intuitive, is what if max profit was 50% of the bankroll, but allowed max bet was 50x the bankroll. A whale could bet 50x the bankroll at 1.01, and basically have a 99% chance to win, and a 1% chance to lose.

Even though the whale could lose on the first round, and increase the house's bankroll by 50x, what will most likely happen is the whale will win about 50 times first. Each time the whale wins, the house bankroll will decrease by half. So by the time the whale actually loses, 50x the house bankroll will basically be nothing (1/22,517,998,136,852 of original bankroll).

---

A +EV edge has the largest impact if spread out over a large number of rolls; the lower the edge, the more it should be spread out. Let's say you played a coin toss game that's weighted 60:40 in your favor, you had $1000, and your opponent had $1000. You can choose how much you wager each toss, and you keep playing until either you or your opponent is bankrupt.

If you bet $1000 on 1 toss, that's basically a 60% you'll end up finishing the game with $2000, and 40% you finish with $0. Only an idiot would play like this (if you happened to only have 40% to win instead (-EV), then this is actually the optimal way to play).

If however, you bet $1 over several thousands of tosses, it's basically statistically guaranteed you'll bankrupt your opponent and finish the game with $2000.

Why did you only answer one small part of dooglus's questions?  Are you going to call him an idiot too to try and discredit him and ignore what he's saying?

You and RHavar are dishonest and shameless human beings.

2:1 is indeed the limit when adjusting your investment (although you won't be margin called if you later exceed it). I stuck with Johnson2239's example calculation anyway to show that it was off.

Almost the entire betting volume consisted of one "angry whale" on bustabit.  How else did bustabit end up with record recording volumes wagered?


You lied about investors being in line with the kelly criterion when it was 1.5x kelly.  You realized that it was a single player behind it all.  You did nothing about it.  You lied about refunding investors.

Well it completely depends on how you expect people to bet. If you are assuming the "worst case" of high-profit bets then Daniel is right. If there's more modest bets that aren't fully utilizing the bankroll, then the investor who decided to leverage will probably end up doing better (as their leverage will be closer to the kelly).

Probably Daniel should've added said: "Assuming an angry whale ..." to make it more correct, but honestly I think very few people really appreciate the harm in over-leveraging so I think the site has a duty to err on the side of pushing people into not using offsite for the purpose of leverage.


I remember after I sold MoneyPot to the current owners, they did a few little changes that resulted in the bankrollers risk being a worst-case going from a 1x kelly to a worst case 3.33x -- no matter how hard I tried (including even writing a simulator, that showed an angry whale would consistently bust them) they never listened. It's just not intuitive for people to realize that despite being +EV you can still expect to keep busting due to over-leverage (and funnily enough, even after they should've learnt the hard way and lost most of their bankroll they just resorted to hacks like limiting max-bet instead of addressing the core issue)

And to be honest, the whole idea of negative expected bankroll growth while having positive expected profit really screwed with my head. It took a lot of creating simulations to get a grasp on it. And the part that I found the most counter-intuitive is that if the casino is over-risking to the point that it expects to lose money, shouldn't it be profitable for a whale to play there?