Probabilities

[SlotsWizard]

I have been analyzing the Harveys slot for the past few nights. You can find an attachment at the bottom of what I believe to be the painstakingly accurate reel layouts.

The scatters and bonus symbols are very easy to calculate the probability of, because you can do so without regard to where they land.

But when it comes to the actual paylines (25 in this case), what is the proper method for calculating the odds? There must be some sort of special consideration for the various different paylines, correct? Do you calculate the odds for a single payline and just use that, or do you multiply that result by 25, or something entirely different?

I had written a program to iterate through all 58.1 million combinations of reel stops and store the winning payouts in a database, but that was taking too long to run and I would prefer to do it mathematically anyway.

Can anyone here help?

A note about the attachment: there is a symbol that looks like it could either be 2 lobsters, 2 shrimp, 2 popsicles, or some other type of dessert. It is this symbol that I have labeled as "Mystery Food".

[brianzz]

Those are lobster tails, lobster tails, with no butter. See the little feet and the tail curled up beside them and the antenna sticking out of what used to be the head.

When I tested real video slots at real casinos to determine the payouts using PAR sheets I used the 5 symbols that appeared across the reels, thats how IGT programmed the games, using that and having the par sheets allowed me to determine a machines return within 5 or 10 plays of the game playing just 1 line.

Just from my experience from testing video slots, the ~ symbols, or blanks, are where the "reel strip" would begin, ie shorter reel strips on certain reels.

I'm not sure what you're meaning by the payline layout, it's really insignifigant if the reels are not weighted. All the paylines return about the same amount in all multi-payline games with the exception of a game with a progressive that pays only on a certain payline or paylines, ie Wheel Of Fortune Video Slots, and the return on that line is increased by 1% which is the games contribution to the progressive, 1% of each bet.

If you're just trying to figure out the probability of a winning combination hitting on a specific line or number of lines you just do the math. I'm at work and keep getting interupted so I can't give you the exact numbers this moment but on most 5 reel games is somewhere in the neighborhood of the top jackpot paying out ON AVERAGE every 2.5 or so million plays no matter which line it lands on. So theoretically you could hit the jackpot 25 times in a row on a 25 line video slot. This isn't really unheard of, I've seen a 3 reel machine hit the top jackpot of 10,000 coins on 2 spins in a row.

In the world of slot machines you're always cheating yourself by not playing max paylines, the coins really don't matter because if a machine "accidently" overpays what it was supposed to, ie the machine pays more than the coin in total, it'll always tighten up and get it back later. Again at land based casinos it's not out of the oridinary to have a video slot to return 250 to 300% of the coin in for a month or two and then for the next 3 to 4 months it'll return 80% or less to get back in check, this usually happens on very volitale games.

I'm sure internet slots are alot different than those in the land based casinos, to be sure of all the speculation someone would have to get with a programmer and talk to them for awhile and maybe get some PAR sheets for a few of the games.

When you're trying to figure out the probabilities you really need to consider how many symbols per reel there really are in the RNG, usually on the reel strip the number of symbols is multiplied 2X or 3X in the RNG to make it really random and harder to win. The newer IGT 3 reel machines have 128 stops per reel which makes the odds of a top jackpot astronomical, 1:2 million spins.

[KasinoKing]

I have been analyzing the Harveys slot for the past few nights. You can find an attachment at the bottom of what I believe to be the painstakingly accurate reel layouts.

Excellent!

[Zoozie]

I have been analyzing the Harveys slot for the past few nights. You can find an attachment at the bottom of what I believe to be the painstakingly accurate reel layouts.

But when it comes to the actual paylines (25 in this case), what is the proper method for calculating the odds?
I had written a program to iterate through all 58.1 million combinations of reel stops and store the winning payouts in a database, but that was taking too long to run and I would prefer to do it mathematically anyway.

Good work on Harveys. When it is possible to play Harveys ín the flash client, I have a way to help you check if you got the reels correct.

It does not matter how many lines you play, the payout% is the same. In my program I just play 1 line when analyzing payout% to speed up the process. It is pretty simple to understand why it does not matter. When iterating over all combinations with 1 line, you will hit every combination once. If you play 2 lines, you will hit all combinations twice. But this will 'cost' you twice as much, but the wins are also double compared to 1 line. So this gives the same. Besides this simple explanation my programs shows excacly the same.

Actually only the number of times each symbols occours on each reels is important. Changing the order does not change the payout%. It does change the maximum win combination that my slot also can find though.

This only gives what I call the basepayout%. If you want to take the free spins into account, you have to use the formular I have written earlier in this thread. And in the case of Harveys, this is a problem since there is a random weight factor that each win are multiplied with during the freespins. Since we dont know this, we can not find the payout%. If we know what the feature would pay on average (in bet size) we could calculate the payout%.

It could be interesting just try find the free spin multier (on average). Since these slots have payout in the 95-96% range, we could work backwards and find the free spin multiplier.

I made some small improvements to the GUI of my slot simulator. (TallyHo and Loaded included also) I will post it again later with your the Harvey reels.

To find the payout% as you tried, the ONLY way is to iterate through all combinations and adding together all the wins. But when same symbols occours more times on the same reel, it is possible to make some SMALL performance improvements, but it is not worth it. (Same principle as Video poker analyzer programs, where this gives a gigantic improvement). It only takes 2 minutes to analyze a slot with 5 million combinations (Loaded) allready.

This is what my program does besides taking the free spins into account also.


Zoozie

[BlackjackInfo]

In the world of slot machines you're always cheating yourself by not playing max paylines, the coins really don't matter because if a machine "accidently" overpays what it was supposed to, ie the machine pays more than the coin in total, it'll always tighten up and get it back later. Again at land based casinos it's not out of the oridinary to have a video slot to return 250 to 300% of the coin in for a month or two and then for the next 3 to 4 months it'll return 80% or less to get back in check, this usually happens on very volitale games.

Amid an otherwise interesting post, this part is utter nonsense, and I expect you know better yourself. This is akin to saying that coin flips will trend toward heads if you've seen a propensity of tails over the last month or two.

[brianzz]

What is non-sense about it? I hate you tell you but it's called a random number generator for lack of a better name. How can it be random if it's programmed? Has nothing to do with coins. It's programmed to win a majority of the time for the casino but in short terms some RNGs win for the player more than was played in the machine. It's a proven fact, I've seen data sheets from machines that do it with my own eyes. This might not happen at online casinos because most slot machines have returns set so low that something like this will never happen. At real casinos where players expect competition casinos have to put machines on the floor that have a hold of 1% or 2%, online casinos don't have to worry about this because they know other players can't walk up and watch you play this machine so they can get away with machines that hold 20% 30% 40%, they don't have to answer to anyone so they can practically do what they want.

Online casinos can throw those PWC reports up every month but I'm not impressed by that, in fact it proves nothing to me with the exception of adding bounses and progressive wins can balance the books. Now if a casino wants to publish a monthly report represents depostied monies - bonuses given by the casino = actual coin in + wins to the player - progressive jackpot wins = actual return of the machine. Awhile back I noticed the PWC reports showed a small spike in "returns" when several casinos were offering buy $50 get $200 free and a little less spike when they offered "play an hour on us" for a deposit bonus.

[BlackjackInfo]

Perhaps I wasn't clear in what I was contesting in your post...
Yes, games sometimes pay unusually high amounts. In your example, a month where the aggregate was 250%. That's randomness for you.

What I disagree with is that the machine will subsequently "tighten up and get it back later". Unless the casino management changes the game to be tighter, the game's not going to change it's expected hold. If the game is set for an average return of 97% and one month it has a 200% payout instead, the next month's AVERAGE return is still the same 97%. Of course, in reality it may pay out 90% or 200% or 500%. The more play the game gets, the more likely the result will be near the expected 97%.

Steering this thread ever further from the original topic, I'll point out an interesting irony that results from this fact. Many Microgaming casinos publish the payout percentages from PWC. If we assume that all Microgaming slot payouts are set the same (I don't know if that's true or not, but it seems likely), then we're virtually guaranteed to find the highest PWC-audited payout percentage at the less popular casinos.

Why? The popular casinos get enough play that their game payout percentage should be near the theoretical hold, while the casinos with fewer players stand a better chance of bucking the odds and posting a big payout number.

[Zoozie]

Here is the newest version of the slot analyzer program and it is now fully functional.
The GUI is still butt-ugly because I am not a GUI programmer.

It comes with the following slots: Tally Ho, Thunderstruck and Loaded.
But you can always define you own from scratch.

And it has 3 functions:
1) Analyze payout% excact
2) find maximum win combination for a given slot/number of lines played
3) Full simulation. (Do a few million autospins, and see how much you lost, or maximum bankroll and when it happened etc.)

In next version it will be able to simulate Harveys. I need to change the core program so it can handle the left to right 'scatters' which is a new type of symbols. But it will be at least 2 weeks before this is ready. I also want to add a few more functions.

Then I will probably declare it finished.

Please report any input/feedback/bugs before the final release. I just need 1 or 2 people to give it a small test.

Unzip all files (it creates a new folder). No virus/adware/registry entries. Pure Java!

Zoozie

[tiger2006]

Great work.

Can you do the same with the Moonshine Slot?

LG tiger2006

[Awai]

Hi,
not sure whether it matters, I use a tool called Hypercam (www.google.com) to create Screenshots and Screenshot movies . Saves all spins or any motion on screen as AVI/MPG movie including chats with support folks or whatever
Cheers
Feif

very interesting topic guys...i've just run a test using screen cap software on my editsuite (i run a multimedia company) and am able to frame accurately scan the reels - i did the test with Thor but in theory it should work with an online slot which doesnt "mask" the spin with cartoony fx. i'll check the layout i create tomorrow against a 50 spin play test - if the results are of interest or use i'll post them here.

in theory if it works it'll cut down the play time for "cracking" the reel sequence to about 20mins...

Awai.