Black Widow Boggled the Noggin!

[Inieas]

I have the strangest Problem on my one Black Widow machine. It has the SS6731 Software, but seems to pay excessively. It has been running for over 2 years and is holding -10%. I have Ram Cleared the machine and did the Setup and on Average if You insert 500 credits, the machine would cashout between 600 and 700 Coins Every time!

Please help as this is not normal. I have even done Kobetron Testing on the machine software and compared the result to the Licence of Certification and it is right!

[stayouttadabunker]

Did you use a SET chip to enter the denomination correctly?
How are you "inserting" 500 credits? Through the coin mechanism or bill acceptor?

[brichter]

How many games have you registered since the last RAM clear?

[Inieas]

31628 to be exact.

[stayouttadabunker]

I'm wondering how he's putting 500 credits on it.
It could be the incorrect bill acceptor or denom setting...?

[brichter]

From the PAR for that reel chip:
Confidence value, 10,000,000 pulls
You are less than one third of one percent through the cycle, to assume that you can judge whether it is on track to pay correctly might be a tad premature...  
Here is a little more info on where you are in the cycle:

                 90% CONFIDENCE VALUES     VOLATILITY INDEX =   16.413
   HANDLE PULLS     LOWER             UPPER
                            PERCENTAGE   PERCENTAGE
            1000.              40.64           144.45
          10000.              76.13           108.96
        100000.              87.35            97.73
      1000000.              90.90            94.19
    10000000.              92.03            93.06

So, you're a little outside the 90% confidence, but nothing that I see as being way out of whack.

You can go ahead and do a RAM clear, but it may reset you out of coming back in line with the 90% confidence values and put you back on a lower hold rate that could be as bad as -44%...  

[rickhunter]

I'm wondering how he's putting 500 credits on it.
It could be the incorrect bill acceptor or denom setting...?

I think he means if you insert 500 credits worth, by the time you are done playing, there's a 600 to 700 credit balance, in other words, the machine is one we would all like to play in the casino's.

I agree with Brichter's analysis though, it's probably a short term issue.  Change the SS chip, maybe you'll get different short term results?

[StatFreak]

Brichter, thanks for filling in and doing exactly what I would have done: looking up the volatility index and asking how many games were played.

I agree that a payback of 110% is not abnormal for any sequence of 32,000 pulls on this game. Also, keep in mind that since the machine's history is already on the high side, it will tend to stay that way as it comes back into the long-term expected range over time when all of the machine's history is viewed as a whole. Of course, if you take the bookkeeping readings right now, then play another 10,000 games and subtract the difference to look at just those 10,000 games, you could expect the payback during those pulls to vary wildly, as the volatility index indicates.

As far as a clear "resetting" the odds, we've rehashed that subject too many times. There is no conclusive evidence that a clear affects the odds, and plenty that says otherwise. The volatility index shows how much the machine can be expected to vary from the stated payback FOR ANY SERIES OF X SPINS, not just the first X spins after a clear.

Inieas, given the volatility of this machine, it is possible that you will still have a payback exceeding 100% after 100,000 pulls. If it's still over 100% after 250,000 pull, I would become concerned.


Stat

[brichter]

See that folks? I did done larned me a little something for all that time that I done been here...

Glad to help lighten your load, Stat!

[knagl]

I agree that a payback of 110% is not abnormal for any sequence of 32,000 pulls on this game.

Based on the chart posted earlier, though, that's well outside the 90% expectation, no?  At 10,000 pulls, the high end expectation is only 109% (108.96).  At triple the number of pulls, the 90% confidence value would be even less than 109%.

[brichter]

The 90% confidence is not an absolute, it's a prediction. The volatility index on this one is pretty high, but David knows more about this type of math than I do so I'll let him give the calculations. Paging StatFreak to the white slot stand courtesy phone... 

My forte is engineering, not pure math. I could call my stepdad, but he's probably asleep by now... he's got a Masters in Mathematics from a local college that starts with an S and ends in a car manufacturer's name, worked on some pretty cool military stuff in his time that I like to dabble in around the 4th of July. 

However, there's only about an 11% difference between 10000 and 100000 pulls, so that would only be a couple of percent when extrapolated to the 30000 pull number.

[StatFreak]

I agree that a payback of 110% is not abnormal for any sequence of 32,000 pulls on this game.

Based on the chart posted earlier, though, that's well outside the 90% expectation, no?  At 10,000 pulls, the high end expectation is only 109% (108.96).  At triple the number of pulls, the 90% confidence value would be even less than 109%.

Thanks Kevin. Now that you've forced me to do the math, it is outside 2 standard deviations, which is what most people would use as a yardstick. At 30,000, the 90% confidence value would be approximately 104%. 90% confidence represents about 1²/₃ standard deviations . Two standard deviations (about 95%) away is approximately 1.19 times the 90% confidence value. The expected payback is 92.544%. So:
104 - 92.54 =  11.46
11.46 * 1.19 = 13.64

With a sample of 30,000 trials:
1SD =   6.82
2SD = 13.64
3SD = 20.46

92.54 + 13.64 = 106.2%  2 Standard deviations (95%)
92.54 + 20.46 = 113.0%  3 Standard deviations (99.7%) (That's just under the 99.9^th percentile)

So 110% is close to 99.35% (99.7^th percentile). Put another way, a machine would be expected to fall outside that range (high or low) about once in 150 times one checked any sample of 32,000 spins.

Given the small sample size, I would keep a close eye on the machine and keep accurate (not approximate) records recorded from the bookkeeping meters in the machine.

The reason that I had originally said that I wouldn't be overly concerned if it were still just slightly over 100% at 100,000 is that since the machine has already paid out 110% thus far, it's more likely to remain above average for some time.

For example, even if the next 32,000 plays run normal at 92%, the overall payback at 64,000 would still be about 101%. If the machine continued to pay back, say 96% for the next 64,000 trials (well within normal), the overall payback would be at 100% at 96,000 (32k+64k). If the payback happened to drop below average in that time, then the overall payback percentage would drop into the 90s, but probably still be above normal. Also, the fact that the first 32,000 trials were well above average does not mean that the next 32,000 will be below average, since each trial is independent.

If the machine runs at 92.5% from now until it reaches 100k trials, the overall payback will still be almost 2 standard deviations above normal, at about 98.1%. There's a 50% chance that the overall payback at 100k will be higher than that, even if there is nothing wrong with the machine.

That's why we talk about the LONG RUN of millions of trials.

By the way, with a sample of 100,000 pulls:

1SD = 3.09
2SD = 6.17
3SD = 9.26

92.544 + 6.17 = 98.7%  (+2SD)
92.544 + 9.26 = 101.8% (+3SD)


If you really want to know if the machine is malfunctioning, keep playing it and DON'T CLEAR IT (That is, leave the books intact as they are). Post back at 100,000 games and we'll see where we stand.

<ADD> Inieas, be prepared to post accurate detailed records, not vague statements like "holding 10 percent"


StatFreak

[StatFreak]

...
Paging StatFreak to the white slot stand courtesy phone... 
...

I've been here the whole time on the other line!!   

[knagl]

...
Paging StatFreak to the white slot stand courtesy phone... 
...

I've been here the whole time on the other line!!

"Give me Ham on 5, hold the Mayo!"


Sorry, just saw Airplane at my local AMC theater (they're doing a special showing of it this past Saturday, and this coming Tuesday).

[StatFreak]

...
Paging StatFreak to the white slot stand courtesy phone... 
...

I've been here the whole time on the other line!!

"Give me Ham on 5, hold the Mayo!"


Sorry, just saw Airplane at my local AMC theater (they're doing a special showing of it this past Saturday, and this coming Tuesday).

 

Do ya' still want me to bring Ham, Oveur? The joint's been down for more than two days and Ham's been sitting on the tarmac since Saturday with a sundae.

[Inieas]

Hi All, Thanks alot for all the replies, I have been slightly swamped, recieved 60 Splus machines I had to refurb and get out. To answer some of the questions, I have now to date inserted 10000 credits by means of coins and still the machine is at 102%. Out minus in meter.

[brichter]

huh? 10000 shows the high water mark to be 108%, you're at 102... and I thought you were at 31000 before?

[IFFV68]

Side Bar,
I was at a machine many years ago in Cripple Creek Co.
It  was a .05 cent machine & if you put in a five, it would register 200  coins instead of 100 coins.  .

I feel bad that I cash out the $10.00 several times & left the Casino.

[StatFreak]

Hi All, Thanks alot for all the replies, I have been slightly swamped, recieved 60 Splus machines I had to refurb and get out. To answer some of the questions, I have now to date inserted 10000 credits by means of coins and still the machine is at 102%. Out minus in meter.

huh? ...

Inieas, I'm not sure that I understand what you said.

First, credits don't matter; the number of games played matters. (Okay, credits matter when we calculate the overall payback. )

Did you read my last post?

IF you played another 10,000 games and are saying that after a total of 41,000 games that the payback has now dropped from 110% to 102%, then I'd say that there is nothing wrong with your machine. In fact, the payback has dropped much faster than expected, so you (as a casino operator) must have seen good revenue from the machine over those 10,000 plays.

If you meant something else, you'll have to explain in more detail.

Stat

[Inieas]

Hi StatFreak,

Sorry for the tardy reply, Went on leave and had a holiday and then came back. Ok the machine has dropped down to 93% we have played a total of 1389562 Games, So I now know the machine is fine. Thanks again.

[StatFreak]

Hi StatFreak,

Sorry for the tardy reply, Went on leave and had a holiday and then came back. Ok the machine has dropped down to 93% we have played a total of 1389562 Games, So I now know the machine is fine. Thanks again.

QED



P.S. Glad to hear that things worked out with the machine.