Monte Carlo methods – Why it's a bad idea to go to the casino • Hacker News Board

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#41

Post by arsenide » Fri, Jul 05, 2019, 3:59 AM UTC

roenxi wrote: post #2

I have a pet peeve about Monte Carlo methods; although it might be more fairly characterised as a rookie mistake I saw once. MCM are not strong if the tail variance isn't an important feature of what is being modeled. I've seen simulations where the modeler starts with an analytic model - from which they could trivially calculate the mean and variance of a KPI - then used a MCM simulation to find out essentially what…

If there is an analytic model, using Monte Carlo methods is extremely useful to validate the theoretical result. If both agree, it is unlikely that there is a mistake in the theoretical result: which is valuable information.

For something as straightforward as Bernoulli, where you plug values into a formula and are done, there is no reason to use MCM. However, I am in agreement that in more complicated cases (e.g. a probability distribution is based on some result of one Markov or Markov-like calculation — or possibly multiple), and the theory is slightly or much more complicated then there is a lot of value in Monte Carlo methods.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#42

Post by bathyspheric » Fri, Jul 05, 2019, 4:05 AM UTC

roenxi wrote: post #2

I have a pet peeve about Monte Carlo methods; although it might be more fairly characterised as a rookie mistake I saw once. MCM are not strong if the tail variance isn't an important feature of what is being modeled. I've seen simulations where the modeler starts with an analytic model - from which they could trivially calculate the mean and variance of a KPI - then used a MCM simulation to find out essentially what…

Can I ask if I understand your use of `tail` in tail variance? Do you mean tail as in extremes of a distribution, or tails, as in losses. I'm very interested in using MC techniques to model extremes of a distribution, like 1-in-1000 year events from a Weibull distribution, but my (naive) algorithms spend a lot of time in the fat belly of the curve rather than out in the tails, but perhaps (probably) I'm holding it wrong. Is there a way to constrain the MC sampling to the tails of a distribution?

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#43

Post by TylerE » Fri, Jul 05, 2019, 4:07 AM UTC

dagenix wrote: post #24

blunte wrote: post #4

If you want to really know why you shouldn't gamble (for profit... gambling for fun, with discipline, can be satisfying), read this quintessential book: Scarne’s Complete Guide to Gambling What you will probably take away from this book is that the house wins when you win. And the reason they win is that they don't pay actual odds. Take the roulette wheel as a simple example. There are 37 (or 38 in the case of 0 and…

From what I've read, casinos are perfectly happy to let you count cards in blackjack. The advantage you get is so small, that even a few mistakes per hour means you'll lose a bunch of money. It's only if you count cards and win that you get kicked out.

Count? Yes.

USE the count? Typically no, unless you're betting very small and consistently.

Basically, the way to annoy them and either get kicked out or only allowed to flat bet is to vary your bet size with the count (E.g. bet a lot when the count is favorable, and the table minimum otherwise). That's the only way to really make money counting. Much easier to detect these days with the eye in the sky.

You can use the count to inform certain decisions that lower the house edge a bit, but typically not enough to actually be +EV. For instance, insurance is normally a very bad bet, but at certain counts it becomes +EV, but not +EV as a whole to make it worth...having to play blackjack.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#44

Post by jaimex2 » Fri, Jul 05, 2019, 4:08 AM UTC

If you go to a casino play poker with other humans. Nothing else is worth your time.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#45

Post by TylerE » Fri, Jul 05, 2019, 4:08 AM UTC

chimeracoder wrote: post #27

dagenix wrote: post #24

Earlier quoted context omitted.

From what I've read, casinos are perfectly happy to let you count cards in blackjack. The advantage you get is so small, that even a few mistakes per hour means you'll lose a bunch of money. It's only if you count cards and win that you get kicked out.

> From what I've read, casinos are perfectly happy to let you count cards in blackjack. The advantage you get is so small, that even a few mistakes per hour means you'll lose a bunch of money. It's only if you count cards and win that you get kicked out. It's not that the advantage is small (if it were, the MIT team would never have been so successful in the first place). It's that most people who try aren't good eno…

It's also worth noting that the MIT teams (and successors) stopped circa 2000. Times have changed. Casinos are much more on top of what's going on on the floor.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#46

Post by justinchan » Fri, Jul 05, 2019, 4:09 AM UTC

daveFNbuck wrote: post #40

55555 wrote: post #11

It's more complicated (worse) than this. That the house has a small edge only tells a small part of the story. When people go to vegas for a weekend, they might have a number in their head they are willing to gamble. If they dont, then that number is their bank account balance. Then they start flipping coins, and often they keep doing this until they go bust. They rarely cash out when up but they always lose when the…

People playing until they go bust doesn't actually affect the casino's take, except insofar as it means more people are gambling. If someone leaves when they're up and someone else takes their place at the table, the casino makes exactly as much money as if the first person hadn't left.

This would only be true if all the seats in a casino are at or close to fully capacity, which in my experience isn’t true. Also only true for games with a fixed amount to bet

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#47

Post by RomanBob » Fri, Jul 05, 2019, 4:10 AM UTC

55555 wrote: post #11

It's more complicated (worse) than this. That the house has a small edge only tells a small part of the story. When people go to vegas for a weekend, they might have a number in their head they are willing to gamble. If they dont, then that number is their bank account balance. Then they start flipping coins, and often they keep doing this until they go bust. They rarely cash out when up but they always lose when the…

This is a common misconception about why casinos win. That 'casinos have infinite bankroll so they cannot lose' isn't actually an argument because casinos have limited bankroll.

If they would allow me to play a +EV game I can crush them even if they had infinite bankroll by varying my betsize. The 'random walk' argument only works for fixed bet size.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#48

Post by TylerE » Fri, Jul 05, 2019, 4:11 AM UTC

freerobby wrote: post #9

fatbird wrote: post #6

Earlier quoted context omitted.

I was able to play martingale in a casino in a video game, so a limited wallet wasn't my problem, and I made money in the long run, but it exposed a subtler problem with Martingale: You only ever win your initial bet back; the rest is covering losses. If you have the cash to withstand 20 losses, then the initial bet, in comparison to your stake, is basically a rounding error. Who cares about winning $1 when you have…

Yes -- this is what is commonly misunderstood about Martingale. What Martingale really does is change the probability distribution of outcomes to be heavily weighted toward winning a very small amount of money. You can, for a given number of plays, use Martingale to make the odds of winning greater than the odds of losing. However what you can't do, is change the expected value of that set of plays. So what you've re…

What's kind of funny is that lots of people talk about playing the Martingale, which is dumb, but no one ever talks about playing the reverse Martingale - which is also dumb and losing, but at least has what I'd argue is a more favorable outcome distribution. Basically, with a reverse Martingale you'll usually lose a moderate amount (which, I mean, is what you'd expect since...casino) but occasionally win a larger amount when you hit your winning threshold and get, say, 6 positive results in a row.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#49

Post by chongli » Fri, Jul 05, 2019, 4:12 AM UTC

daveFNbuck wrote: post #40

55555 wrote: post #11

It's more complicated (worse) than this. That the house has a small edge only tells a small part of the story. When people go to vegas for a weekend, they might have a number in their head they are willing to gamble. If they dont, then that number is their bank account balance. Then they start flipping coins, and often they keep doing this until they go bust. They rarely cash out when up but they always lose when the…

People playing until they go bust doesn't actually affect the casino's take, except insofar as it means more people are gambling. If someone leaves when they're up and someone else takes their place at the table, the casino makes exactly as much money as if the first person hadn't left.

Yeah. I think casinos would prefer not to inflict financial ruin on their patrons. Apart from the bad PR (and potential for political interference), a bankrupt patron seems less likely to come back to the casino with more money.

Re: Monte Carlo methods – Why it's a bad idea to go to the casino

#50

Post by TBurette » Fri, Jul 05, 2019, 4:13 AM UTC

A while ago I wrote a similar interactive article to prove that the martingale betting technique at the roulette doesn't work : [1]. The martingale is when you bet 1, double the bet each time you lose, start again at 1 when you win.

The aim was to create a single page anybody could link to that proves without a doubt the abysmal odds of making money off the roulette with the martingale technique. Instead of pre-written probabilities, charts and 'obscure complicated math' that people could easily dismiss. I wanted to give people the ability to run simulations of playing at the casino in the browser. The idea being that people would end up building the probabilities/charts themselves through simulations.

It is running Monte Carlo simulations inside the webpage and outputs the result as charts dynamically using js and D3. What it does is it repeatedly simulates going to a casino with a certain starting amount of money and an objective of how much you want to win. It then plays the roulette using the martingale technique until you got your target winnings or you lost your money.

Unfortunately I don't think I wrote the article really well and the 'generate yourself the charts proving it doesn't work' doesn't come across really well. I should have put a 'run' button instead of needing a page refresh for a start.

[1] http://thomasburette.com/martingale/