so, very confident, apparently. thanks, Mosier, that’s what I wanted to know.
speaking as one of the riffraff! 
The casinos want everybody to think there’s a trick, a secret, a lucky machine… and there isn’t. They make huge piles of money by making the games, at best, random, and then take a cut off the top. The only ways to make any money in a casino is to play a game where you get to play against another player, or you get to bet that the house will win (such as a crap game, or some roulette, and even then, you’ll probably just lose money more slowly than if you try to win).
I agree the “play that one” (wink) followed by an instant jackpot is implausible, but the idea that the waitresses might know which machines are “hot” seems vaguely possible to me. My understanding (and correct me if I’m wrong) is that with electronically controlled slot machines the computer has to occasionally manipulate the odds to keep the payout percentages within spec, and so the old gambler’s fallacy that “it’s due for a jackpot!” can be somewhat true. It seems like an astute waitress wouldn’t have too much trouble noticing which machines in her area were getting a lot of play without major payouts and would thus be likely to give slightly better odds at that particular moment.
They call it a fallacy because it is incorrect thinking. The results of any given spin are independent of any other spin. Slot machines do NOT occasionally manipulate the odds. Each spin has precisely the same chance to win no matter how long it has been since the last jackpot.
Isn’t it true that some slot machines are “looser” than others and casinos have some leeway in how they tune the payout structure for a particular slot machine at a particular time? For example, the casino can decide whether a machine has a payout of 88 %, 90 % or 95 %.
I’m assuming there are rules on how precisely this process works, i. e. it’s probably illegal to change the payout “on they fly” during play and it would certainly be illegal to tailor the payout to a particular player. But other than that, the payout scheme for a slot machine is a business decision: you increase the payout to attract customers on slow days or in second-rate locations and you lower the payout when demand is high (for instance the slot machines at McCarran International Airport where there are always enough individuals willing to feed their last remaining coins into the machines).
I think it’s a safe bet (heh) that there are approximately 0% of the machines which are set to pay out more than they take in, regardless of how slow a day is, so all we’re really looking at here is how fast the one-armed bandit takes your money.
This is venturing into an area I am not at all expert in, but I’ll hazard an explanation for why I think this is impossible (or at least, extremely unlikely).
Sure enough, slot machines are indeed programmed to deliver an expected payout rate over time. By analyzing data from a machine over a (very, very) long period of time it may theoretically be possible to determine that a machine has paid out at a lower rate than expected so far, thereby suggesting it ought to pay out at a higher rate in the future to compensate.
The problem is, this time period is measured in months, not hours. A cocktail server watching a particular machine can only do it for about 8 hours a day, and much of that time she wont actually be able to watch that particular machine because she’ll be working in other areas. This means she would need accomplices on other shifts to help her keep accurate records.
Also, this would only work for older machines.
Newer machines are programmed as networks, and the network of machines has an expected payout rate. Individual machines can vary wildly from the programmed payout rate, because other machines in the network can compensate for the difference. It is NOT always obvious which machines are part of a particular network, and certainly beyond the ability of any normal casino employee to observe and record.
Although it seems this ought to be true, it’s not. Machines are indeed programmed to deliver an expected payout rate. They aren’t actually random.
Yeah, looking into that a little further, it appears to me that in the US at least they machines are just 100% random, so nevermind! Although apparently some slots in the UK do fiddle with the odds to get an exact payout percentage.
Well, yeah, that’s what I was saying-- with something like a roulette wheel or dice or other physical device thinking the previous results affect the odds is obviously fallacious, but if there’s a computer adjusting the odds to get a certain payout percentage that wouldn’t necessarily be the case. But, yes, apparently they don’t do that in the US so ignorance fought!
There is a website with some interesting information:
See, yeah, I could have sworn I heard that from some authoritative source too, but googling it seems to be pretty clear that Turble is right that the odds don’t adjust to achieve a certain payout.
I think it might be a little confusing because when they’re calibrating the machines it sounds like they describe the odds in terms of that expected payout rate, but that means just what it says: the expected payout rate over time for a given set of odds. The machine doesn’t do anything to actually make sure it achieves that expected payout rate.
That can be confusing for some people.
The way I like to explain it is:
Let’s say I have a coin that comes up heads 90% of the time and tails 10% of the time.
It’s still 100% random, even if it’s not uniformly random.
Casino slot machines have random number generators inside. The payoffs for the slot machines in Atlantic City and in Nevada have to be at least a certain minimum percentage by statute. It is possible (actually done) to set the average outcomes from these random number generators to keep the average payouts at least 80% (usually more in many places) and less than 98%.
That’s still random, even if the setting an average rate of return does not seem random to some people. Particular outcomes aren’t known, even if average behavior is known.
Even for the trick coin or even a fair coin, I can’t say with any certainty what the result of a particular flip will be, even if I can tell you how many heads will likely show up over 1000 flips.
Right - the players do that, simply by playing a lot. Results that in the short term vary from the expected payout are not compensated for, they are overwhelmed - lost in the noise of many, many plays.
Right, I understand all that. The confusion was if you set the machine to, say, pay out 90% will the machine adjust the odds to assure it actually achieves 90% payout or does it simply set the odds to where it will theoretically achieve the 90% payout over time and not change them. The idea of a “hot” machine relies on it being the former, but it appears that at least in the US it’s the latter.
Looking at it as a coin flip being a very simplified slot machine:
You bet one dollar. Heads you win one dollar. Tails you lose one dollar. Over the course of thousands of flips you will sometimes be ahead and sometimes be behind, but after a large number of flips you will be very close to even.
Now, to make it a casino game … we don’t have to change the odds of the coin flip. We can change the payout. Instead of paying you a dollar when you win, we pay you 95 cents. And we ring bells and flash lights and call you a winner to take your mind off what actually just happened to you. For a small number of flips you might at some point be a little bit ahead, but after a large number of flips you will be very close to losing 5% of the money you wagered. And we can set the payout to any percentage we want without changing the coin one bit. On video poker machines, we don’t change the odds of getting dealt a Straight or a Royal Flush, we change the amount you get paid when you hit one.
If we are flipping an electronic coin controlled by a computer chip then yes, we can simply write the program to only show heads the correct number of times to earn our desired percentage … but the results are the same. No one can predict when the machine will hit, only that given a large number of plays the amount paid out will be very close to the Expected Value.
The odds of winning can be precisely calculated but the machine is still random. Any one play has the same chance of winning as any other play, regardless of how many wins or losses there have been in the machine’s recent past.
Some casino employees (just like any other people) may actually believe they can spot a hot or overdue machine but mostly they are just being friendly and polite when they wink and say “This one looks good.” Sometimes it does hit and they get a nice extra tip, but the results are random. If they don’t give the advice, they never get the extra tip.
It depends; how tight were the sweaters and how pronounced are your nipples? Pictures may be needed for absolute certainty.
I’ve read gambling books that recommend tipping the cocktail server to get information on “hot” slot machines. I don’t know why it would be a felony because there isn’t really such a thing as a “hot” slot machine since the outcomes are randomly generated every play. The customer and the waitress might believe a machine is “hot” as many people do, but unless the waitress took part in or knew of an illegal manipulation of the machine’s outcomes , I don’t think this “insider information” should be a felony, though the waitress could be subject to getting fired if against the casino’s policy.
Some video poker players in Washington state probably think they’re playing with actual video poker payouts, but there’s an extra wheel that adjusts the payouts to whatever the casino wants them at. I’m not sure of the bare minimum legal payout for slot machines in Washington state, but probably around 70-80% which is terrible for a video poker game.
This thread is an excellent study in why casinos are so incredibly profitable.
Emphasis on “almost”.
The gaming laws have established a system that is insanely profitable. Every casino’s best long-term interests lie in closely conforming to those laws.