Sometimes people can be too smart for their own good. I was sent a write-up linked to a single player who claims to be a statistician. Unfortunately, while this person is probably incredibly math savvy presuming their self-credentialing is accurate, they clearly don’t understand the basics of how a slot machine works.
The original post was so long that even I struggled to stay focused on it because within a few sentences the player had already made critical errors in how they came about to their arguments. I’m going to cite specific sections, and then if you’re really curious will incorporate the full post at the end so you can attempt to follow along with it.
It is my understanding that the law requires gaming machines to possess a minimum payback percentage to all players. This percentage is derived through the algorithm’s source code as an overall average calculated over a defined amount of spins. While all the machines pay back may be compliant on this requirement, there is no regulation on HOW this requirement can be achieved.
The player’s first mistake is not understanding the concept of a random number generator (RNG) and its critical involvement in the overall payback. In this case the player is writing about Massachusetts casinos.
The Massachusetts basic rulebook is nearly identical to Nevada’s with just a few modifications. Nevada requires all outcomes to be possible on any push of the button and the corresponding call to the RNG. By that nature, that is how that requirement MUST be achieved, and so slots can’t weight to some outcomes vs. others.
All the possible outcomes are established upfront, which is how the payback is achieved, and it’s selecting among them, whether on a per-reel basis, in the case of a reels first machine, and the number of options on each reel available, or in the overall set out outcomes, in the case of a prize first machine.
As such, that regulates how this requirement can be achieved, and so the game can’t go off script as it were and do what it wants.
As one example, a machine can read a player’s club card and determine the player’s frequency in their play and automatically adjust the standard deviation accordingly.
Players cards have no impact on a slot’s payouts. The players card systems don’t have access to the payback settings, and cannot change them. Furthermore, the slots are simply designed with x payout options, not multiple “standard deviations” with the same payouts that it can switch between.
For example, a low volatility slot can’t suddenly be transformed into a high volatility one because a players card was inserted. Changing things like hit frequency to accomplish this would basically break the feel of a game, which is a very important part of a slot’s design. Just because something is theoretically and mathematically possible doesn’t make it true.
Other variables that I have noted this pattern for are: time of day, day of the week, holidays, the time delay in between spins, and even the players age group.
The only way to detect this would be to examine the source code and verify all the variables defined. I was told that this kind of information is under NDA, thus there is no way for a player to know for sure.
Yes, slot machine source code is under NDA for a variety of reasons, not the least of which is that imperfect beings can implement imperfect solutions. Even the random number generator isn’t truly random, but pseudo-random, and knowing how it was developed could yield ways to beat the machine. There’s precedent for this.
However, it is a bit concerning to me that when I ask to see the source code for gaming machines, I am not allowed. When I bring this up to state authorities, my emails have gone unresponsive for about 6 months now. We need to raise a red flag over this and demand proof.
Slot machines go through a testing phase before they can be put on casino floors, through companies like Gaming Labs International (GLI). They test to the exact regulations I mentioned at the outset. So to be approved in Nevada, they have to pass Nevada’s regulations; to be approved in Massachusetts, they also have to meet Massachusetts’ nearly identical, but somewhat different, regulations.
Usually a game will be designed to clear the hurdles of many states at the same time, maximizing a game’s sales potential within the market(s) it targets. They’re not going to hand the source code to any random person who asks, but the games are required to be reviewed by a third party prior to hitting casino floors.
So Twitter manipulates their algorithms to suit their needs….but Encore would never manipulate their source code to make higher profits.
First off, a social media algorithm and a slot machine is comparing apples and oranges. Second, a casino doesn’t have access to the source code either – just the final compiled and encrypted software loaded onto a machine. They don’t hack slots and can’t do whatever they want – they choose from a half dozen or so payout options, set the denomination and location, and let it make them money.
It’s important to remember that to the slot makers, the game needs to feel like the game regardless of where it is; letting a casino set a payback absurdly low would hurt them too. The game would be practically unplayable below a certain payback. The house edge on slots is high enough and they’re the most profitable gambling on the casino floor – they don’t need to do a thing to make plenty of money.
Knowing math is helpful to knowing how a slot machine works, but you can’t just throw out some aspects because it’s inconvenient. More importantly, as I always advise, if you’re this distrusting of a heavily regulated industry (which is done so specifically to keep things legit and fair), you shouldn’t be stepping into a casino and gambling at all.
OK, now that lengthy rambling post I promised – good luck!
Hi guys…..so I have been skeptical about the way slot games are managed for a long time and it is my suspicion that the gaming authority and their partners manipulate the statistics player by player to maximize profits. It is my understanding that the law requires gaming machines to possess a minimum payback percentage to all players. This percentage is derived through the algorithm’s source code as an overall average calculated over a defined amount of spins. While all the machines pay back may be compliant on this requirement, there is no regulation on HOW this requirement can be achieved. There are many combinations of numbers that one can average together to generate the same value. The statistical value that governs this behavior is the Standard deviation of the payouts over a range of spins. The higher the standard deviation, the more spread out and extreme the values are. If the standard deviation is lower, the values will tend to be closer to the average. A machine set with a high standard deviation will likely pay out zero for most spins, but then has the chance of paying out an extremely high amount to balance out all the zeros and still be within the required average. A machine set with a lower standard deviation will tend to be more reliable and will pay out at a rate closer to the requirement for smaller sample sizes. Using my expertise in physics and mathematics, I have observed and noted the behavior of many machines over a 1 year period. I have noted patterns against different variables. My theory is that it is mainly the standard deviation that changes according to the aforementioned variables. As one example, a machine can read a player’s club card and determine the player’s frequency in their play and automatically adjust the standard deviation accordingly. It appears that people with a higher frequency of play have a higher standard deviation setting than those who appear to be new players. Other variables that I have noted this pattern for are: time of day, day of the week, holidays, the time delay in between spins, and even the players age group. Testing the machines for minimum payback would not catch this change in standard deviation necessarily. For example if the minimum payback is 80% and the machines are set to 85%, the standard deviation can be allowed to vary within a certain bandwidth and still test within the required 80% minimum. This variation ultimately means that not every player is enjoying the same exact behavior on every spin at all times. While the machine may be paying back the required average over time. The instantaneous odds can vary greatly, thus changing the overall entertainment value of each visit and making the overall experience very unreliable. To make an analogy, It is not as reliable as saying that a poker game has the same 52 card deck on every deal. Or every craps table has 2 6-sided dice on every throw. The only way to detect this would be to examine the source code and verify all the variables defined. I was told that this kind of information is under NDA, thus there is no way for a player to know for sure. If I ask a craps worker to let me examine the dice for integrity that’s not a problem. If I ask a poker dealer to count all 52 cards in front of me and show me their hands are clean, that’s not an issue. However, it is a bit concerning to me that when I ask to see the source code for gaming machines, I am not allowed. When I bring this up to state authorities, my emails have gone unresponsive for about 6 months now. We need to raise a red flag over this and demand proof.
So Twitter manipulates their algorithms to suit their needs….but Encore would never manipulate their source code to make higher profits. They are 100% trustworthy! Everyone should be video recording every spin at this point so that machine behavior is properly documented and saved. Anyone that can video record a machine relentlessly give zeros for 100s of spins on end should save this and share.