Probabilities

Zoozie said:
'Thunderstruck' is one of the most easy slots to calculate value of free spins for. I came up with very good estimate earlier. (45 times the bet size). I can change the value and see how much it changes the payout%. So this is not really an issue

33.5 or thereabouts in my test last year if memory serves.
 
Zoozie said:
Using 34 gives:
payback:0.9411552188552188

Thats gotta be pretty accurate then i'd say? My sample was only over 25,000 spins and if you assume a 95% payout...
 
Slotmachine said:
Looks promising, Zoozie! The payout seems likely to go down just the right % once you figure in the final reels. Calculating the effect of the free spins feature is tricky...I guess you could just rely on empiric data about the average return of the feature (KK should have some data). However, there must be a mathematical formula to figure in the exact effect the free spins feature has on the payout...?
SM
Sorry I can't help with Thunderstruck - I rarely play it cos it always MURDERS my bankroll :(

As for the free-spins payback - surely you can just work out the average return from 15 normal spins & then just multiply by 3?
Then, also knowing the odds of hitting free spins, you can calculate the odds of a re-trigger and the resulting extra return.

Zoozie, I haven't the faintest idea what you talking about with your maths, but it looks mighty impressive stuff to me! Well done! :notworthy
 
"As for the free-spins payback - surely you can just work out the average return from 15 normal spins & then just multiply by 3?"

The word 'normal' in your setence is the problem. We dont know what
the return% is before we know how much the free spins are worth.
I probably just need to think about it.

There also is the lazy way just by brute force. Playing a gazillion random spins
with the correct freespin feature implemented. Maybe the program can run tomorrow while I am at work. Nothing more fun than playing thunderstruck with
play-money mode AND ascii graphics on autospin
 
Ohh! I hit jackpot playing 1 line only after 1.3M spins, but I am still down :)

I J Q 9 L
T T T T T
S 9 S L C
new maximum win
totalWin:30000
multiplier:3
freespins left:5
iterations:1334533
payback:0.9679267730109015
 
After 145M spins the payback seems to be between 95.1% and 95.4% and staying in this interval. So there is no reasons to burn brain cells trying to figure out how much the free spins are worth, just let the program run for some hours and you will have 4-5 digits correct (95.36% etc).


I have attached the program if anyone interested in seeing the simulation. (and see how long it takes before you hit the 5 Thors during free spins).

0) You need to have Java installed. But you probably have this allready even though you dont know it.
1) Unzip
2) click 'start_simulation.cmd'

Every few minutes it will print the current state of the simulation. (payback%)

Zoozie
 
Last edited:
It does not work for me. :(
java.lang.NoClassDefFoundError: SlotTestCaseRunner
Exception in thread "main"

I would be really interested in seeing this work.
 
I had a thought on calculating the free spins multiple. Ignore the fact the slot may or may not be "weighted":

If you start with $9 million balance, do 1 million "ordinary" spins at say $9 (IE: taking the 3x free spins multiple out of the equation) take the balance left at the end & divide this amount by 1 million (no. of spins), and you have the average return per spin. Then multiply this by 45 (15 spins x 3x multiplier).

That gives you the total average amount won in the bonus round, so divide by 9 (wager amount) and you have the average multiple which you can stick back into the program (ie 35x).

Probably best to do it over 10 million spins or even 100 million if poss.

I think thats right anyway. There may be a simpler way - its late :D
 
Very impressive, Zoozie. I am working on the template of the reels for Harvey. So far almost reel 1 finished; ran into a snag with one (valid) group of 3 that i cannot seem to fit in anywhere. I was interested in this since the so-called "scatters" really are just cheap stuff but the bonus round can be very rewarding. Of course, the two occasions when i played for real money, i never saw a bonus round :( If i ever manager to finish this, do you want me to post something?
 
GrandMaster said:
It does not work for me. :(
java.lang.NoClassDefFoundError: SlotTestCaseRunner
Exception in thread "main"

I would be really interested in seeing this work.


Hi GM.

You are sure you unzipped all files in same directory? You have Java installed, so
there should be no problems.
 
Managra:
This version can not handle "Harvey's". It is too special and does not fit into same template as ThunderStruck.
1) scatters must be alligned on left reels
2) champagne on reel 2 and 4

I could make the program know this template but it would take some hours.
And from my experience the feature multipliers are heavily weigthed (you rarely get *25). So I will not be able to simulate it 100% like thunderstruck where there was no weights involved.
 
Last edited:
I had not expected my program to run that fast. After 1 hour the maximum
value for the 'int' was reached.

This happens after 2147483647 spins! (2^31 -1)

So it is not really important unless you want to find the payback% with 4 digits. The program still shows how fast your money converge towards the payback%, and that there is no way to get back on tracks again :)

But it was CERTAIN that the payback% is between 94.9% and 95.0%.

It will only take 1 minute to 'upgrade' the int to long in Java but will need
to attach the program again in a new post. Think I will improve it with better logging and more lines and maybe more slots if someone makes the template. Now the program is made I should be able to make these changes fast.
If you have any ideas for improvement/logging, just tell me.
 
Zoozie, in my opinion you're turning out to be a star on this forum and I hope you stick around.

Nice to see some positive and informed action on a rather esoteric subject, and I know that you will enhance the contributions frequently made here by experts like Grandmaster. WTG!
 
jetset said:
Zoozie, in my opinion you're turning out to be a star on this forum and I hope you stick around.

Couldn't have said it better. I'm glad this thread is getting sustained interest and people are busy analyzing the slots :)

When we play BJ or VP, we know what the return% is. It's only fair that we slot players can make some informed decisions, too.

I find it comforting that T'struck is close to the 95% mark (even though I know this is terrible from a BJ player's point of view), I was afraid it might be a lot lower, like 91-92%.

Many thanks Zoozie :notworthy for proving that practically any gambling related question can be answered right here in the CM forum!:thumbsup:

Cheers,
SM
 
*blush* - thanks for the kind words.

I have been running the test over and over all day and they all converge to the same result. (as they should do mathematically, you cant beat the casino)

Payback% = 95.0% +- 0.1%. (94.9-95.1).

However this result is only valid IF we got the reels correct and there are no bugs in my program. The last part I can check myself again.

Anyone downloaded it and started the simulation?
In the start it will log lots of spins and you can check that the
pay involved is correct. (it only play 1-line). Just stop it and start over. (and see how long it takes before you hit the jackpot)

Zoozie
 
9-lines now implemented. This is the best line combination in ThunderStruck:

T T T T T
S 9 S L C
9 R 10 J S

New maximum win
totalWin:10264
scatter win:0
Line 1:0
Line 2:10000
Line 3:0
Line 4:4
Line 5:10
Line 6:200
Line 7:0
Line 8:50
Line 9:0
 
This is new version now using all 9 lines. Same result as before (as expected).

1) unzip all files to a new folder.
2) click 'start_simulation.cmd'

I only have one question. Are the reels 'looped'? (as going in a circle.)
Ie. if a reel is like this : 1,2,3,4,5,6,7 (7 different symbols, each only once)

is 7,1,2 and 6,7,1 then a 'legal' spin?

In my program I have assumed this, but it would not take long to fix if I was
wrong.

Zoozie
 
Slotmachine said:
Damn! Now THAT'S what I call Autoplay!!!!:eek:

Cheers,
SM

Yeah. My computer can do 6M spins/second.
So if you play 0.09$ pr spin, your hourly expected loss would be 1.6M $ at 95% payback!

Maybe I should sell this autoplay program to the casinos :)
 
Last edited:
Zoozie said:
Yeah. My computer can do 6M spins/second.
So if you play 0.09$ pr spin, your hourly expected loss would be 1.6M $ at 95% payback!

Maybe I should sell this autoplay program to the casinos :)

LOL.. Microgaming might send you a big $$$ check just for not doing this sort of thing anymore:D

Your program is definitely eye-opening. Playing 200 million spins in a matter of minutes....hitting the jackpot once...well, it sort of makes me wonder...It's a deep look into the metaphysical nature of slot machine gambling....and into the quantum and probabilistic nature of life, the universe and everything...

Instead of gambling for years to come, I could just let your program run for 10 minutes, see how I did, then move on to other things..

Cheers,
SM
 
Actually I was wrong (sorry)

It is you expected loss each MINUTE that is 1.6M$.
Hourly rate is 97M$.

Not that it makes it any better......

I agree it is kind if depressing to see the simulation.. Just no way to get back
up after allready 100000 spins.

I got an idea for next modification. I will also log whenever your payback%
makes a new record, so we can see how late this can happen.

Edit: new program attached with this improvement
 
Last edited:
Zoozie - did you include the formula for calculating the free spin probability as mentioned last night?
 
Simmo! said:
Zoozie - did you include the formula for calculating the free spin probability as mentioned last night?

No. There really is no need. But I can have the program calculate the
avg. win for a feature. I will do that tomorrow. I will of course make
possible respins wins still count as the original respin wins.

This program is an random spin autoplay simulation that also emulates the
free spins and *3 multiplier. Just let it run for some hours if you want to
see if the excact payback is 95.04% or 94.99 %. The longer it runs, the more
digits you will see not changing.
 

Users who are viewing this thread

Meister Ratings

Back
Top