Michael's Deconstructing Jackpot Party Post

I was fascinated by Michael Shackleford's post entitled Deconstructing Jackpot Party. My question (for Michael, or others that are knowledgeable) is whether or not the same method could be used on a progressive slot machine to calculate the jackpot size at which the machine will be returning more than 100%. I'm referring to strictly to slot machines, not video poker, where these values can be easily calculated.

It seems to me that if were possible to accurately calculate this value on progressive slots with only a few hundred spins on the target machine, you could just run around with a list of machines and optimal jackpot sizes. Instead of having complicated VP strategies to memorize, you'd simply go to a machine and hit the "spin" button until you or someone else wins and the jackpot resets. This would result in other benefits too, since slot play normally yields much higher free play and comps than VP. Earnings would be higher as well, since there are zero mistakes to make and slot play is generally faster than VP play. Also, I've never heard of a slot player being asked to leave :).

Any info would help. This could be a fun new approach to advantage gambling.

Edit: If Michael pops in on this, I'd also like to know if you believe it's possible to accurately determine the weighting of symbols on single line slots, and are you absolutely certain that there is no weighting on multi-line video slots?

As is usual with higher mathematics, I did not quite follow the deconstruction of the straight video machine and so doubt that I would fully grasp the deconstruction of a prgressive video machine.

The slot machine has a certain payout percentage but not an expectation as to when it will take place... so in reality the information would be meaningless to any drunken coin-dropper merrily going about her business. She already knows the slot machine is rigged to take rather than give, she knows she can watch all the doodads and have some excitement and she knows that they will soon be sending a cocktail waitress around again with another free drink for her. So whether the machine is 88.89 or 91.02 makes little difference in the price of her next "free drink".

Now a prgressive that is somehow "due" might be of more interests but such information is already "known" its simply imprecise.

Quote: FleaStiff

As is usual with higher mathematics, I did not quite follow the deconstruction of the straight video machine and so doubt that I would fully grasp the deconstruction of a prgressive video machine.

The slot machine has a certain payout percentage but not an expectation as to when it will take place... so in reality the information would be meaningless to any drunken coin-dropper merrily going about her business. She already knows the slot machine is rigged to take rather than give, she knows she can watch all the doodads and have some excitement and she knows that they will soon be sending a cocktail waitress around again with another free drink for her. So whether the machine is 88.89 or 91.02 makes little difference in the price of her next "free drink".

Now a prgressive that is somehow "due" might be of more interests but such information is already "known" its simply imprecise.

Progressives don't become "due". As the progressive rises, it simply makes it more valuable when you win. The odds don't change. If the progressive value gets high enough, the game will have a mathematical return of more than 100%. You just have to be able to calculate what "high enough" is for the game you are playing. If it is possible to use Michael's strategy to calculate that point for any given progressive slot machine, then it opens up a whole new (at least to me) form of advantage gambling.

Any other responses to my original post above?

Thanks.

Thanks for the kind words. I could indeed "reconstruct" any video slot as long as pertinent bonus rules were provided.

The book 'The Secret World of Video Poker Progressives' talks about also playing reeled slots if the jackpots got above a certain point. The book didn't get into how they knew when a jackpot was big enough, but they probably used crude techniques of recording at what point the jackpot usually was hit and meter growth.

There are probably other teams, and perhaps recreational players, who already do this.

And then there is the tale of the infamous "Lion's Share" $1.00 progressive slot at MGM.

Lions'sShare

Quote: MrV

And then there is the tale of the infamous "Lion's Share" $1.00 progressive slot at MGM.

Lions'sShare

I wouldn't be able to reconstruct that one, because those single-line games have unequally weighted stops.

Quote: Wizard

Quote: MrV

And then there is the tale of the infamous "Lion's Share" $1.00 progressive slot at MGM.

Lions'sShare

I wouldn't be able to reconstruct that one, because those single-line games have unequally weighted stops.

Actually, you could use the same technique. While the weightings aren't known, what you *do* know is the physical strip -- at least after just a few observations. Most physical reel machines are 22-stops long (11 symbols, 11 blanks). So you'd just use frequency analysis: how many times did stop N appear on reel 1, etc. The only part where it gets tricky is with the progressive symbol or other rare symbols. Those don't show up on paylines much, but they should show up one-off paylines as the adjacent blanks appear. For starters, just assume a weight of 1, and see whether it makes sense with what you calculate the denominator (length of virtual reel) to be.

Quote: MathExtremist

Actually, you could use the same technique. While the weightings aren't known, what you *do* know is the physical strip -- at least after just a few observations. Most physical reel machines are 22-stops long (11 symbols, 11 blanks). So you'd just use frequency analysis: how many times did stop N appear on reel 1, etc. The only part where it gets tricky is with the progressive symbol or other rare symbols. Those don't show up on paylines much, but they should show up one-off paylines as the adjacent blanks appear. For starters, just assume a weight of 1, and see whether it makes sense with what you calculate the denominator (length of virtual reel) to be.

That's what I did with my Slot machine appendix 1, and I found even the 4000 spins I recorded were not enough. There seemed to be a 6 to 1 ratio in highest to lowest reel weighting.

I recall hearing that some slots were programmed to display or stop on "near misses" instead of coming to rest on a random, worthless stop. Is there any truth to this? If so, wouldn't that make tracking patterns worthless?

I was of the impression that "near miss" displays were considered psychologically effective and Gaming Board regs banned the practice. Anyone able to find the latest regs on it?

The appropriate sections of Reg 14 were enacted in 1988 after Universal did exactly that:

Reg 14.040
All gaming devices submitted for approval: (3) Must display an accurate representation of the game outcome. After selection of the game outcome, the gaming device must not make a variable secondary decision which affects the result shown to the player.

I happen to have that quote handy because it's in a patent specification I drafted, but that's a different conversation.

Quote: Wizard

Thanks for the kind words. I could indeed "reconstruct" any video slot as long as pertinent bonus rules were provided.

The book 'The Secret World of Video Poker Progressives' talks about also playing reeled slots if the jackpots got above a certain point. The book didn't get into how they knew when a jackpot was big enough, but they probably used crude techniques of recording at what point the jackpot usually was hit and meter growth.

There are probably other teams, and perhaps recreational players, who already do this.

Thanks for the response! I had also thought of simply recording jackpot levels over the course of time, and assuming that if the jackpot was in the 90th percentile historically, that it might be an advantage play. However that method is far from foolproof, and you'd really have no way of knowing if you were playing with a true advantage. That was what struck me about your post - it was the first time I had ever seen the possibility to accurately determine machine return and payout odds.

Are you absolutely certain that there is no weighting on any multi-line video slot? Also, do you have any idea approximately how many spins it would take to determine the weighting on a single line slot? Even 10,000 spins at a 5% advantage (expected loss of 500 units) would be an acceptable investment to me if I could get the optimal jackpot size.

As for Lion's Share, someone needs to check those EPROMs. There is a machine in a lonely corner of the Grand Sierra casino in Reno that has a sticky note on it saying "maximum progressive jackpot $80,000" and the jackpot is at $80,000, so there is no meter rise anymore. It's a 2 coin $1 machine and appears to be older. I would assume its probably at the 100%+ point, or something is wrong with its EPROM :).

Quote: MathExtremist

The appropriate sections of Reg 14 were enacted in 1988 after Universal did exactly that:

Reg 14.040
All gaming devices submitted for approval: (3) Must display an accurate representation of the game outcome. After selection of the game outcome, the gaming device must not make a variable secondary decision which affects the result shown to the player.

I happen to have that quote handy because it's in a patent specification I drafted, but that's a different conversation.

That's true, but they could just program the game to have a statistically high percentage of outcomes that appear on the reels to be "near misses". That would indeed be an accurate representation of the game outcome. For example, the odds of a megabucks symbol appearing on reels 1 and 2 might be set to 1 in 5 spins, and reel 3 might be set to 1 in 2,000,000. The odds of having all three appear on the pay line would be 1 in 50,000,000 spins (1/5*5*2,000,000). However, once every 25 spins you'd see 2 megabucks symbols on the pay line, and if you didn't know better you'd think you were about to win - especially since some of them light up and make a noise when they line up on the pay line! :)

Quote: thefish2010

That's true, but they could just program the game to have a statistically high percentage of outcomes that appear on the reels to be "near misses". That would indeed be an accurate representation of the game outcome. For example, the odds of a megabucks symbol appearing on reels 1 and 2 might be set to 1 in 5 spins, and reel 3 might be set to 1 in 2,000,000. The odds of having all three appear on the pay line would be 1 in 50,000,000 spins (1/5*5*2,000,000). However, once every 25 spins you'd see 2 megabucks symbols on the pay line, and if you didn't know better you'd think you were about to win - especially since some of them light up and make a noise when they line up on the pay line! :)

That's right, and that's exactly what they do -- albeit not to the degree you're suggesting. That was the tradeoff allowed by the GCB in the 1980s when other gaming vendors started making noise about doing their own "near miss programming" games. Read Chapter 4 of Jeff Burbank's "License to Steal" for the whole story.

Quote: MathExtremist

That's right, and that's exactly what they do -- albeit not to the degree you're suggesting. That was the tradeoff allowed by the GCB in the 1980s when other gaming vendors started making noise about doing their own "near miss programming" games. Read Chapter 4 of Jeff Burbank's "License to Steal" for the whole story.

Interesting, I'll have to pick that book up. Any thoughts on my question about whether or not multi-line video reels are weighted? Some casino hosts will tell you what your theoretical loss for the day was if you ask (mine always have). I'm thinking that if you stick to one machine for a set number of spins, get the theo from the host (which gives you the return for that machine, since you know how much action you gave it), and combine it with analysis of the results as Michael did with Jackpot Party, you could come up with an accurate profitable jackpot level even if you couldn't perfectly work out the weighting based on results (the machine's return, which will be dead accurate, might fill in the blanks on the weighting).

Social engineering + math for the win?

Quote: thefish2010

Interesting, I'll have to pick that book up. Any thoughts on my question about whether or not multi-line video reels are weighted?

They are not. In order for multi-line slots to be weighted, you'd have to calculate the RTP of each payline independently, and they would all be different. That would be a ridiculous overcomplication yet would serve no practical purpose. I have never done, nor been asked to do, math for a weighted multi-line game (and I've done a *lot* of math for multi-line games). In fact, most of my clients' game engines couldn't handle a weighted multi-line game. They'd have to rewrite the code to account for it.

Not saying you couldn't do it, but it would be far more trouble than its worth -- especially in regulatory costs. Consider that a game with 50 paylines would have 50 different RTPs and therefore would require something close to 50x as much time to regulate, so you're looking at almost 50x the lab fees. That'd never pay off.