20

Being a computer programmer I assumed that every single spin of a poker machine was pre-calculated by the machine thus avoiding the chance of multiple large pays being payed out in a short period of time. e.g Some guy puts in 5 bucks but spins 3 x $10,000 in three spins.

I am just reading a government designed pamphlet and I am surprised to see the following:

There are millions of possible combinations on modern poker machines and every spin has an equal chance of bringing up a winning combination

It does say later that the operator can change the bias of the machine so that there are fewer chances to win, that is adjusting the probability, but the above statement suggests that the provided payout is still left to chance. Meaning you could effectively get multiple large payouts in a row.

Most slot machine players I have met suggest that a machine "goes quiet" after a medium/big payout and you are better off pulling your money out and switching machines because the machine knows it has already paid you.

Are poker machine payouts for each spin calculated on the fly or is purely down to statistics?

Flimzy
  • 15,520
  • 14
  • 63
  • 132
going
  • 18,069
  • 18
  • 86
  • 151
  • 5
    On the last line, the answer to both options is Yes! – Oddthinking Apr 13 '12 at 04:46
  • @Oddthinking - I think I worded that wrong, I mean pre-calculated at each turn, as in decided in a deterministic way. – going Apr 13 '12 at 05:35
  • 18
    Often, even after explanation, people have trouble grasping the statistical concept of a series of independent events. Instead, gambler's lore of various kinds is used to explain supposed patterns in the events as they are realized. – Paul Apr 13 '12 at 06:10
  • 1
    Why should they be designed like this? The chance that this happens is small at best and if the machine doesn't have enough money left to pay out a large win it still can alert an employee. – Martin Scharrer Apr 13 '12 at 14:53
  • @Martin No slot machine in a casino I've ever seen still pays out cash. They give you a slip of paper with a cash value that you can either cash in or feed to another slot machine. Or you have a card, which has a cash value. Machines take money off the card for each spin and put the winnings on the card. You can get the cash value of some or all the money on the card at any time. – Sam I Am Apr 13 '12 at 19:05
  • Thanks @Sam, I already thought something like this is used. (I don't have much experience with casinos myself, because they are very rare in Germany.) This makes the claim even less likely, right? – Martin Scharrer Apr 13 '12 at 19:07
  • @Martin, yeah, less likely if not having the winnings on hand was the reason for the assumption. Every casino I've ever been to (3 of them) the winnings were paid by the cashier, not the machine itself. I'm sure there probably was a time where machines spit out money, but not the modern ones here. I can't imagine why anyone would assume results being pre-calculated was ever the case. I'm also a programmer, and I never thought that. – Sam I Am Apr 13 '12 at 19:15
  • @SamIAm: It doesn't make sense for me as well and really sounds like some gamblers lore as said above. BTW, I remember seeing gambling machines in the UK which used real cash (.1 or .5 pounds or so), but these machines were for small winnings only and never had any big jackpots. – Martin Scharrer Apr 13 '12 at 19:27
  • Casinos bear this risk with traditional games. I don't see their motivation to mitigate it with slots. –  Mar 16 '18 at 16:14
  • 2
    Depends on whether you'd consider a Pseudo-Random Number Generator with a seed value a "pre-determined" calculation. It is a deterministic calculation and the same RNG with the same seed should produce the same number sequence. – GordonM Mar 16 '18 at 16:55

2 Answers2

28

No, it is not the case that results are pre-calculated in gaming machines. The result of each game is calculated independently and on-the-fly.

Gaming machines' randomness is validated statistically by independent companies such as Gaming Laboratories International. [Warning: Site has annoying voice-over.]

At GLI, we have one job: to test electronic gaming equipment. In fact, our clients are gaming regulators in jurisdictions all over the world, more than 450 in all.

They test against standards such as Gaming Devices in Casinos, although the appropriate standard depends on jurisdiction and there is not yet a universally accepted standard.

Example rules from the Gaming Devices in Casinos standards:

Each possible permutation or combination of game elements that produces winning or losing game outcomes shall be available for random selection at the initiation of each play, unless otherwise denoted by the game;

[...]

The RNG [Random Number Generator] shall be cycled continuously in the background between games and during game play at a speed that cannot be timed by the player.

[...]

Unless otherwise denoted on the payglass, where the gaming device plays a game that is recognizable such as Poker, Blackjack, Roulette, etc,, the same probabilities associated with the live game shall be evident in the simulated game. For example, the odds of getting any particular number in Roulette where there is a single zero (0) and a double zero (00) on the wheel, shall be 1 in 38; the odds of drawing a specific card or cards in Poker shall be the same as the live game. For other gaming devices (such as spinning reel games or video spinning reel games), the mathematical probability of a symbol appearing in a position in any game outcome shall be constant.

Note: Some machines manage a growing jackpot. Where there is a jackpot, the expected payouts between games are not independent.

Disclaimer: I have a friend who used to work in the gaming validation industry, but not for this company. I have no connection with Gaming Laboratories International.

Oddthinking
  • 140,378
  • 46
  • 548
  • 638
  • 1
    Would be interested to know why this is necessary: "at a speed that cannot be timed by the player". Surely that sucker flies (randomly not cycling 1,2,3,4) at phenomenal pace, not repeating itself at regular intervals? – going Apr 13 '12 at 05:32
  • 2
    Possibly in response to [Michael Larsen on Press Your Luck](http://gscentral.net/larsen.htm)? Longer, better-told version by [This American Life (Act IV)](http://www.thisamericanlife.org/radio-archives/episode/412/million-dollar-idea). – Oddthinking Apr 13 '12 at 05:44
  • 3
    @xiaohouzi79 - If you can time the updates, you can predict future numbers, as the RNG is really a pseudo RNG, thus allowing a player to improve his/her odds. – Rory Alsop Apr 13 '12 at 08:01
  • 1
    @Rory Are you sure about this? Why would modern casinos not use real RNGs? Such devices are cheap enough, and even normal computers can produce non-predictable random numbers (`/dev/random` on Unix …) using several sources of environmental noise from device drivers. Either way, the non-timed speed requirement is unnecessary. – Konrad Rudolph Apr 21 '12 at 01:32
  • The RNGs are, as far as I know, all linear congruential RNGs. The RNG must also preserve its state across power outages/resets. The reason is simple - you don't just need random numbers, you need to guarantee that *every* number will eventually turn up. Linear congruential generators have that characteristic. If some numbers might not occur in the sequence, those values may correspond to the range of values that produce a top-tier win on the game, which would invalidate the pay tables. – Dan Haynes Jan 03 '13 at 22:04
  • @DanHaynes I don't understand your "The reason is simple": IMO to simulate roulette you should to be random, and leave it to chance whether any given number eventually turns up; and to simulate poker, you should randomly select from the deck of remaining cards. – ChrisW Nov 03 '13 at 09:45
  • 1
    @ChrisW Picture an RNG that gets re-seeded with a random value every third time you use it. There are numbers that generator will never produce in a million years. If one of those numbers maps to a royal flush, the player would never hit one. But the true odds of getting a royal flush in poker is fixed. If you play enough hands, it *will* happen. It has to be preserved across power outages/resets otherwise the players will place a bet, draw some cards and then pull the plug if it's a losing hand. Power it back up, draw again... repeat until a winning hand shows. Yeah, it really happens. – Dan Haynes Nov 13 '13 at 22:54
  • 3
    @ChrisW: These things are regulated by law and your specification is not enough by law to simulate poker. To properly simulate a deck shuffle, if you run a million games then the distribution of which card is on top of the deck should be equal - that is, you must guarantee that the random shuffle should generate an equal probability of getting king of club or three of diamonds or any of the other cards as the top of deck. Indeed you need to guarantee this for all positions in the deck. In gambling, the statistical distribution is as important as unperdictability – slebetman Feb 08 '17 at 14:27
  • How, exactly, would one "prove" (short of examining the logic/mechanism) that a slot is truly "random", vs using a pre-programmed sequence? – Daniel R Hicks Mar 17 '18 at 18:40
  • 2
    @DanielRHicks: They don't have to stop short of examining the internals. There are also statistical tests for looking for biases and non-randomness. – Oddthinking Mar 17 '18 at 23:39
  • @Oddthinking - Consider that it's often impossible to decipher the logic of a computer program when there was no intent to conceal the algorithm. Don't you think any reasonably competent 7th grade programmer could hide "cheats"? – Daniel R Hicks Mar 18 '18 at 12:45
  • What incentive is there to cheat when they can just make the game statistically favor the operator? – Tgr Mar 18 '18 at 19:09
  • @DanielRHicks: [Let's take it to chat](https://chat.stackexchange.com/rooms/74749/discussion-on-answer-by-oddthinking-are-slot-machines-purely-random-or-do-they-p). – Oddthinking Mar 19 '18 at 11:53
-4

According to this source, it is possible to pre-calculate the outcome of a slot machine and "rob the bank".

So you can decide whether you beleve if it's pre calculated or completely random and "pre-calculatable" on the fly.

pericles316
  • 22,676
  • 2
  • 84
  • 161
Frezzley
  • 253
  • 1
  • 8
  • 7
    -1: That source - a Wired article - is not a reliable one; it does not contain sufficient detail or references to ensure what they are relating is true. I was going to demonstrate this by showing another user had [asked a question about that exact article](http://skeptics.stackexchange.com/questions/37172/did-russian-hackers-beat-slot-machines-with-an-app), but it turns out that was *you*! Even **you** don't trust this source. – Oddthinking Feb 08 '17 at 14:38
  • FWIW, Schneier seems to have taken it at face value.. – Benjol Feb 09 '17 at 10:46
  • 1
    This is interesting but not really related to the question. OP asked whether machines "decide" in advance that the player is going to lose. That the internal random number generator has a weakness and can be predicted is very different from not being based on a random number generator at all. – Tgr Mar 18 '18 at 19:08