UK "compensated" slots -- Does anyone really get what is going on?

In my blog entry on Gambling in the United Kingdom I tried to shed some light on the various classes of slots they have there and what they mean to the player. I admittedly did a lousy job of it, because I still don't understand it very well.

Any of this could be wrong, but my understanding is that slots in the UK at one time, sometimes referred to as fruit machines, were "compensated," meaning they were programmed to achieve certain return percentages in the short term. So, if a machine had been running cold then it would be designed to loosen up to achieve it's desired percentage.

On the contrary, "random" machines have the same odds every spin, which is how they work in the US.

This page at the UK Gambling Commission explains that class B3, B4, C, and D machines can be compensated OR random, while B1 and B2 must be random. The machines there did identify what class they were.

When I was there I noticed a lot machines stated "The outcome of any game or feature is not necessarily that shown by the odds displayed." in the rule screens. I tend to think these are the class B3, B4, C, and D machines, but could use a confirmation on that.

I guess my other main question is if someone is playing a B3, B4, C, or D machine how can he tell whether is he playing a compensated or random machine? As noted above, it can be either.

I've read a lot of legal mumbo jumbo on this topic, but does anyone really understand what is going on, who can explain it in plain simple English? I'm really looking to explain to the world if these "compensated" games still exist, and if so, how to identify them.

Thanks in advance for any help.

To answer my own question, I think that the "compensated" games are what we would call "pull tab" games here. They are most likely to be found in bars, fish-n-chip shops, and other places that don't want to take the chance on a fully random game.

Source: http://www.casinolifemagazine.com/article/gambling-professor-explains-british-casino-culture. In particular

Quote: Dr Mark Griffiths

Men gravitate towards skill games. It’s interesting, because in Britain men are more likely to play slot games than women, whereas in other countries it’s the same. We’re the only country in the world which doesn’t use a random number generator. This means that there’s a level of skill involved – people can watch the machine fill up with other people’s money, then come in and get the pay-out.

Quote: Wizard

To answer my own question, I think that the "compensated" games are what we would call "pull tab" games here. They are most likely to be found in bars, fish-n-chip shops, and other places that don't want to take the chance on a fully random game.

Source: http://www.casinolifemagazine.com/article/gambling-professor-explains-british-casino-culture. In particular

I didn't -think- fruit machines run like a pull tab, with a deck of results in the memory. I do know that people seem to know/think they know when a machine is 'ready to drop'.

Quote: thecesspit

I didn't -think- fruit machines run like a pull tab, with a deck of results in the memory. I do know that people seem to know/think they know when a machine is 'ready to drop'.

Can you elaborate on HOW the fruit machines work then?

This article also seems to say that "compensated" slots act like a pull tab game.

Quote: uk-21.org

These machines are known in the gaming trade as "B3", or "compensated" games - "compensated" meaning that once a win combination is used (paying out a small win or, at the other end of the scale, a jackpot), this combination is removed from the total number of outcome combinations that could be achieved by the random number generator. In short, previous payouts are compensated for and will affect future prospects of winning.

Here's what I know:

If the game is random, it must say "This Machine is Random."

If it is compensated, it must say "This game is compensated and may be influenced by previous play."

If there is any feature of the game, including reels, that could indicate odds that don't match the actual odds, then the game must say "The outcome of any game or feature is not necessarily that shown by the odds displayed." This would apply to mechanical reels that are weighted, for instance. I'm guessing they could use dice that are not fair if they include this statement.

As far as I know, all awards must be available at all times, but they can make the game looser (enrichment periods) or tighter (raking periods) based on the current return of the game.

Quote: CrystalMath

but they can make the game looser (enrichment periods) or tighter (raking periods) based on the current return of the game.

Can you elaborate on how they do that?

Quote: Wizard

Can you elaborate on how they do that?

I spent six weeks in Spain working on some slot machines for a Spanish manufacturer. Although the machines I was working on were typical Class 3 slots they also manufactured AWP (Amusement With Prize) machines. I was told those machines had multiple payback percentages and that if the actual results were outside some parameter they would change to a tighter or looser schedule to get back within the desired payback percentage range.

Quote: DRich

I was told those machines had multiple payback percentages and that if the actual results were outside some parameter they would change to a tighter or looser schedule to get back within the desired payback percentage range.

How confident are you in this information? I'm not saying you're wrong, but this does contradict that last two sources of quoted, which could be wrong.

Do you know if the loosest of these three settings has a theoretical return over 100%?

Quote: DRich

Quote: Wizard

Can you elaborate on how they do that?

I spent six weeks in Spain working on some slot machines for a Spanish manufacturer. Although the machines I was working on were typical Class 3 slots they also manufactured AWP (Amusement With Prize) machines. I was told those machines had multiple payback percentages and that if the actual results were outside some parameter they would change to a tighter or looser schedule to get back within the desired payback percentage range.

Although I've never been involved with one of these games, I would anticipate doing exactly what DRich said. This way, all prizes are still available, just at a different distribution.

Quote: Wizard

Can you elaborate on how they do that?

These are all regulated by the UK government, right? Do they make their regulations public?

Quote: AxiomOfChoice

These are all regulated by the UK government, right? Do they make their regulations public?

Yes, most of it is on the Gambling Commission's website, although hidden in reams of documentation. I can see they allow "compensators" for C and B3/B4, but those sections do not seem to be in the B1 or B2 documentation. As has been stated previously only casinos are allowed B1 machines, with their larger top prizes, and indeed in most casinos all the machines are B1; only a few casinos have the £500 ones.

See http://www.gamblingcommission.gov.uk/gambling_sectors/gaming_machines/about_gaming_machines/about_gaming_machines.aspx
or http://www.gamblingcommission.gov.uk/shared_content_areas/gaming_machines_technical_stan.aspx

This one ( http://www.gamblingcommission.gov.uk/gambling_sectors/gaming_machines/about_gaming_machines_fruit_m/gaming_machine_categories.aspx ) has the current list of categories and maximums.
A's aren't found anywhere;
B1's only in casinos;
B2's are the ones typically found in bookmakers;
B3 (and I'm guessing the B's) are sometimes found in bingo halls (and some "regional casinos");
C's are typically found in pubs;
and you can imagine the D's are amusement arcades, given the description "penny drop", "crane grabber" etc.

The most interesting requirement I've stumbled across so far is Section 5.8. of http://www.gamblingcommission.gov.uk/pdf/Machine%20standards%20category%20C%20June%202012%20revision%203.pdf . There's a similar section for B3/B4.

Quote:

5.8 Use of compensators and/or regulators
The use of compensators or regulators to determine any stage of the game outcome is permitted,
except in the case of ‘pre-gambles’, provided that the following rules are complied with:
a.
each possible permutation or combination of game elements that produces winning or
losing game outcomes must be available for selection at the initiation of each play;
b.
the outcome of any gamble must not be predictable to the player;
2
See Regulation 2, Gaming Machine (Circumstances of Use) Regulations 2007
3
Within the meaning of the Category of Gaming Machine Regulations 2007
14
c.
cyclic periods of play must not deliberately be introduced and due care must be exercised
to prevent their inadvertent occurrence;
d.
the chance of winning a prize must not be so altered as deliberately to create a series of
losing (raking periods) or winning games (enriched periods); and
e.
any sequence of wins must not exceed that to be expected from a random machine of a
similar payout profile and running at the same payout percentage.
A gaming machine or device must not present a losing game result which indicates a ‘Near Miss’;
for example where the odds of the top award sy
mbol landing on the pay line are limited it must
not frequently appear above or below the pay line.
A compensated game must clearly display to the pl
ayer on the face of the machine at all times
(when in operation) or at the point the game is selected for play (where operated on a multi-game
terminal) the following statement:
THIS GAME IS COMPENSATED AND MAY BE INFLUENCED BY PREVIOUS PLAY
Where a machine operates in such a way that a particular feature (such as Hi/Low or gamble)
may invite a player to make a choice in circ
umstances in which they have no chance of success
(defined by the probability of a win being reduced to less than 20% of that required to achieve the
target percentage payout) then the following statement must be substituted for the above:
THIS GAME IS COMPENSATED AND MAY BE INFLUENCED BY PREVIOUS PLAY AND
OFFER THE PLAYER A CHOICE WHERE THERE IS LITTLE CHANCE OF SUCCESS

Nice find charlie. I have dug through British regs before and never found the technical standards. Apparently because I never did stumble across the gambling commission website...oops

Quote: CrystalMath

Quote: DRich

Quote: Wizard

Can you elaborate on how they do that?

I spent six weeks in Spain working on some slot machines for a Spanish manufacturer. Although the machines I was working on were typical Class 3 slots they also manufactured AWP (Amusement With Prize) machines. I was told those machines had multiple payback percentages and that if the actual results were outside some parameter they would change to a tighter or looser schedule to get back within the desired payback percentage range.

Although I've never been involved with one of these games, I would anticipate doing exactly what DRich said. This way, all prizes are still available, just at a different distribution.

Good call guys.

This statement seems to argue the machines are pull-tabs

Quote: Regs

cyclic periods of play must not deliberately be introduced and due care must be exercised to prevent their inadvertent occurrence;

I interpret that to mean that the outcomes must be well mixed up, and the maker can't put outcomes in a certain pattern, to try to cheat and play the game when he knows a jackpot is forthcoming.

Another reason in favor of pull tabs is I understand these "compensated" games are meant for small businesses, like bars and fish-n-chips restaurants. Even at a low return setting of say 50%, the game could still perform four standard deviations above expectations and cause a big loss for the proprietor in the short run. However, a pull tab game is totally safe as long the proprietor has the patience to play through at least one batch.

An argument against pull tabs is that every outcome must be possible. Perhaps the regulations meant that each outcome must be possible at the start of a batch. I've personally made pull tab games for the US market and that is how we interpreted such a requirement. However, if they means always possible, then what about when the game has one entry left in a batch. Clearly not every outcome would be possible. A solution would be to mix in a new batch when at least one possible outcome of the old batch has been exhausted.

I think this particular regulation makes the machine function like DRich initially described.

Quote: British Category C Regulations

e. any sequence of wins must not exceed that to be expected from a random machine of a
similar payout profile and running at the same payout percentage.

This would rule out the idea of pull tabs, wouldn't it?

I haven't been involved in the UK machine market, but from what I've read in this thread the compensation is simply an evaluation that if the recent actual payback has been outside some target RTP range, the game swaps in a tighter or looser set of reels for a while until the actual RTP gets back within tolerances.

Also, finite pool games are operated in the UK under category B3A. I don't believe any of the rest of the categories are based on finite pools, and a finite pool game doesn't need to compensate anyway because the total pool payback is known in advance. That's one of the primary reasons to using a finite pool rather than an independent random game. The deal may be shuffled so winners come out early or late, but the total payback is constant so you wouldn't ever need to compensate for anything.

Quote: tringlomane

This would rule out the idea of pull tabs, wouldn't it?

I wasn't sure what that regulation was saying. Perhaps it was referring to a pull tab game and that the wins couldn't be too clumped together.

Quote: MathExtremist

I haven't been involved in the UK machine market, but from what I've read in this thread the compensation is simply an evaluation that if the recent actual payback has been outside some target RTP range,

I'm not saying you're wrong, but it does contradict the sources I sited earlier who said that compensation games are "non-random."

Quote: Wizard

I'm not saying you're wrong, but it does contradict the sources I sited earlier who said that compensation games are "non-random."

I think you may be conflating pull-tab games with skill-stop/pachislo games. In the skill-stop games, you determine when the reels stop, not an RNG. The details are a bit fuzzy, but I understand them to be variable in terms of payback in the same way you're describing the compensating feature.

On the other hand, pull-tab games actually are random, they're just dealt from a depleting pool of tickets before starting over. There is no skill involved in video pull-tabs, and I think the source you cited was referring to skill-based games. Like I said before, you don't ever need to compensate for anything if you're operating pull-tabs. There is zero variance in the net results of the whole pool and the operator always knows what's coming.

Quote: MathExtremist

On the other hand, pull-tab games actually are random, they're just dealt from a depleting pool of tickets before starting over. There is no skill involved in video pull-tabs, and I think the source you cited was referring to skill-based games. Like I said before, you don't ever need to compensate for anything if you're operating pull-tabs. There is zero variance in the net results of the whole pool and the operator always knows what's coming.

Wouldn't there be a certain "skill" in watching games and playing ones that haven't paid out much? You know the high-payout tickets are still in there.

I have no idea if it could get to +EV territory without the details, but it seems you could at least cut in to the edge. Assuming they are mandated to deal out ALL tickets before a reset. I have heard of B&M pull-tab joints (they have them in bars around here) pulling a bucket and just trashing all the tickets if they notice a bunch of the big-paying tickets are left and it's close to the end. Definitely shady, whether or not it's illegal I have no idea.

Quote: AcesAndEights

Quote: MathExtremist

On the other hand, pull-tab games actually are random, they're just dealt from a depleting pool of tickets before starting over. There is no skill involved in video pull-tabs, and I think the source you cited was referring to skill-based games. Like I said before, you don't ever need to compensate for anything if you're operating pull-tabs. There is zero variance in the net results of the whole pool and the operator always knows what's coming.

Wouldn't there be a certain "skill" in watching games and playing ones that haven't paid out much? You know the high-payout tickets are still in there.

I have no idea if it could get to +EV territory without the details, but it seems you could at least cut in to the edge. Assuming they are mandated to deal out ALL tickets before a reset. I have heard of B&M pull-tab joints (they have them in bars around here) pulling a bucket and just trashing all the tickets if they notice a bunch of the big-paying tickets are left and it's close to the end. Definitely shady, whether or not it's illegal I have no idea.

I've dealt with pull tab games in Washington state, and there would be no way for the player to know what's coming, because they would constantly create new instances of the pool depending on how depleted the existing pools are. There could be hundreds of pools simultaneously.

I think you would have as much luck doing the same thing with scratch off lottery tickets.

As for the B&M joints, that's definitely shady. On the other hand, if players knew that all the good tickets had already come out, then they wouldn't buy the remaining tickets, so the place could get stuck with a loss on that batch.

Quote: CrystalMath

I've dealt with pull tab games in Washington state, and there would be no way for the player to know what's coming, because they would constantly create new instances of the pool depending on how depleted the existing pools are. There could be hundreds of pools simultaneously.

In the late 1990's I worked on some systems for Washington state and we only drew 50% of the "tickets" from the pool. After 50% we always started a new pool but the customers would not know when a new one started.

Quote: AcesAndEights

Wouldn't there be a certain "skill" in watching games and playing ones that haven't paid out much? You know the high-payout tickets are still in there.

I have no idea if it could get to +EV territory without the details, but it seems you could at least cut in to the edge. Assuming they are mandated to deal out ALL tickets before a reset. I have heard of B&M pull-tab joints (they have them in bars around here) pulling a bucket and just trashing all the tickets if they notice a bunch of the big-paying tickets are left and it's close to the end. Definitely shady, whether or not it's illegal I have no idea.

I don't believe that you could meaningfully exploit it, other than having a general idea of when NOT to play...which is probably never.

The reason is that, I would assume, for a certain pay to be advertised, then that pay probably has to be possible, IOW, at least one occurrence in every set, if not more. Let's say you have pays on a machine of 500-FOR-1, 100-FOR-1, 50-FOR-1, 25-FOR-1, 10-FOR-1, 5-FOR-1, 2-FOR-1 and 1-FOR-1...Okay, now for every single one of those things to hit once (100% payout, which certainly isn't the case) it would be 693 plays. Furthermore, the lower pays are certainly going to occur more frequently, by design, than the higher pays, so you're obviously going to have far more than 693 plays.

If you look at Wizard's analysis of the fruit machine he designed:

https://wizardofodds.com/games/slots/appendix/4/

Note: From this point forward and only for purposes of my example, I am treating Wizard's fruit machine as a single-line game.

You'll see a paytable similar to the theoretical one I posed above would complete a pre-set cycle at 8000 spins based on the way Wizard set that machine up, however, Wizard's machine also has an extremely low house edge of 3.5%. I have no way of knowing whether or not these British fruit machines are lower in the pubs, but if I had to guess...

As has already been mentioned, new seeds can probably be added while a current seed is in progress, furthermore, a machine may be able to start with multiple seeds programmed into it in the first place. In the unlikely event the house ever gets way behind on the machine, assuming no programming errors, they would simply be able to let the seeds already loaded finish playing out before adding new ones.

Therefore, unless you watch the machine from the installation of it, know how many seeds the machine started with having been loaded, know the fixed RTP of each individual seed, and track all of the spins, then you have no way of knowing when you are at an advantage.

For instance, let's go back to Wizard's fruit machine. If we assume that only one seed is loaded into the machine at a time and is allowed to play all the way out, (and this is really a best case scenario for someone hoping to track it) then you could watch 2,000 spins with returns as follows:

One Cherry 400/2000 = 800
Two Cherries 30/2000 = 150
Three Cherries 2/2000 = 20
Three Oranges 7/2000 = 105
Three Bells 10/2000 = 200
Three Plums 2/2000 = 100
Three Bars = 0
Three Globes = 0

Okay, so the overall RTP in this 2,000 spin sample was 1375/2000 = .6875 or 68.75%

A single starting seed would have 8000 * .965 = 7720 in pays and 8000 in bets

If we subtract the 2,000 spins from both sides, we are left with 6,000 in bets and 6345 in pays, based on what you have seen. The remaining RTP, based on what you have seen is 105.75%, which is clearly awesome. Everyone likes trading 6,000 for 6,345, and actually, since we have this information, we know to quit ANYTIME we are more than 345 units ahead!

Okay, so you decide to play out the remaining 6k in bets, however, the previous day two people came in and played 800 total spins with the following combined results:

One Cherry: 190/800 = 380
Two Cherries: 20/800 = 100
Three Cherries: 1/800 = 10
Three Oranges: 7/800 = 105
Three Bells: 4/800 = 80
Three Plums: 0/800 = 0
Three Bars: 1/800 = 100
Three Globes 1/800 = 500

That's 1275 in pays v. 800 in plays, which brings us to 5200 remaining in seed plays and 5070 in seed pays, and the RTP on the rest of the seed is back down to .975 or 97.5%, which is still better than it started, but the player is at a disadvantage, and unbeknownst to the player, the best pays remaining are 7/9 Three Plums.

CONCLUSION

These things could theoretically be exploited, but a ton of information is needed, the more information the better, but at a minimum:

1.) What is the fixed RTP, per seed?
2.) How many seeds did the game start with, or how many seeds are left?
3.) Since the beginning of the current seed(s), how much has been bet and how much has been returned?

The probability distribution of results would also be nice to know, but strictly speaking, isn't necessary as long as you know how much the pool started with in terms of RTP and how many total bets that reflects, (i.e. House Edge) as well as the results of every spin.

If there is a jurisdictional minimum RTP on these things, then you could always assume that, but you'll not know where you are in a seed on anything except for a brand new machine, and even then, you may not know how many seeds have been loaded at a time.

Simply put, if you watch someone play one of these machines at a lesser return than jurisdictional minimum, or a return less than the base House Edge, (if known), but all that will really tell you is if the House Edge is better or worse than when you walked in.

If you have a barkeeper friend, however, who is in charge of seeding the machines, collusion could be possible just based on knowledge they would have of the machine's take (during a seed, if loaded one at a time) vs. the machine's pays for that seed.

I've been doing some more digging and found some patent numbers that address these "compensated" slots. They are GB2185612 and GB2087618. Based on my layman's reading of the latter, it seems the way these fruit machines work is they keep a running record of the return over the past x spins. Then it adjusts the probability of certain bonus features of the game according to whether the game needs to pay out more or less to get closer to its desired return percentage. It seems to me that there isn't just three settings (tight/average/loose) but a sliding scale according to the probability of the bonus feature(s).

That is about as specific as I can get. GB2185612 was not very helpful and I think what I did get of GB2087618 was not the full thing. American patents tend to run hundreds of pages, by comparison. If anyone wants to try to dig deeper, I used this site.

I know folks who do 'spy' fruit machines in UK pubs and reckon they can tell when they will 'drop' the jackpots. I have no idea if they are full of it or not. They will certainly spend a bunch of time on them/

Quote: thecesspit

I know folks who do 'spy' fruit machines in UK pubs and reckon they can tell when they will 'drop' the jackpots. I have no idea if they are full of it or not. They will certainly spend a bunch of time on them/

Considering the typical payouts of these things, I would be shocked if they weren't full of it. If the average payout is 80%, you don't have to make the game +EV to catch up.

When I was a kid, the service clubs (like the Elks/Eagles/Rotary/Moose) used to always have a dusty punchboard on the bar. It was usually dull yellow. And always dusty.

Anyway, the board would have a listing of all the major prizes on the board. You would take a litle wire and punch through however many holes you paid to punch. There were several hundred to a thousand punches available on each board, and the per punch price was at different denoms. Little rolled slips would fall out the back. And you would unroll them and read off the amount of your prize(s), if any.

You were not allowed to do this without the bartender there, because the bartender took your money, counted how many punches you used, and gave you your prize money. They also marked off every prize on the face of the board as it was won. They were not allowed to retire a board until all prizes large enough to be printed on the face were marked as found. But once a board had awarded all listed prizes, it was retired as too bad a bet for the players, no matter how many holes were left unpunched (and there were many minor prizes not listed). No idea what the actual edge was. But it was always a treat to get to buy punches.

Good web site on fruit machines: http://www.clubmachines.co.uk/mech.htm.

It is more on advice for playing them than how they are programmed. From the web site I get the impression that a jackpot comes along once every 3 to 10 days and the pros try to play when it is due.

I just rewrote the section in my London blog entry about slots as follows. I welcome all comments.

Normally I don't like to update a blog entry, but I got so much feedback about the slots in the UK that I had to revise what I wrote originally. I've been writing about slot machines since 1997 and one of the most frequent questions I get is about UK "fruit machines." In particular some machines there do not offer the same odds every spin but adjust the odds continuously to achieve short-term return targets. I must have been asked specifically how they work hundreds of times. While I can't answer the question completely, let me tell you what I do understand.


First, not every slot machine in the UK is what I'll refer to as a Fruit Machine, which use some form of "adaptive logic" at achieve short-term win goals. The ones in the casinos are much like the ones you see in Las Vegas, although with some limits on the maximum bet and jackpot size. To be specific, there are various classes of machines, as follows:

Class A -- Don't exist
Class B1 -- These are the most Vegas-like slots and found in casinos only. Maximum win of £10,000.
Class B2 -- These are multi-game machines often found in race and sports betting parlors. A reeled slot game in this class will have smaller jackpots than a B1 machine. Maximum win of £500.
Class B3, B4, C -- These are the Fruit Slots typically found in pubs.
Class D -- These are amusement only games that you might find in an arcade. For example a claw game where you try to win a teddy bear.


"So, Wiz, how do these Fruit Machines work?," you might ask. I've read through lots of web sites and read lots of Emails and forum posts about the topic and I'm still not exactly sure. Two terms you hear come up over and over are "compensated" and "adaptive logic." Apparently, the goal of these games is to ensure the proprietor that the game will make money consistently on a weekly basis. The game has a memory and keeps track of how much money it has been taking in and giving out. If it has been overpaying beyond its target return percentage then it tightens up. Likewise, if it has been underpaying then it will loosen up.

"But how does it do this?," I hear you asking. Other web sites contradict each other about this so I looked at an actual patent that covered the technology, UK 2087618-A. It mentions that the probability of the player achieving certain features and bonuses in the game is dependent upon the recent payout return percentage. Thus, the player is more likely to win the features and bonuses if the game has been underpaying in its recent history. How far back it goes, I don't know, but believe it to be about a week, depending on how much the machine gets played.

One of the most convincing web sites I've seen on the topic of Fruit Machines is Club Fruit Machine Guides (http://www.clubmachines.co.uk/mech.htm). As a resource for players, this site doesn't get too deep into the programming of the machines, but says the net effect is that the games hit a jackpot on roughly a weekly basis. The trick is to know when a game is ripe to hit. That web site gives some hints I won't get into but emphasizes knowing individual games and how frequently they hit a jackpot. He also emphasizes tipping the bartenders, who can be more forthcoming with information about the games in their bar according to how much they like you.

You can expect to find other web sites out there claiming Fruit Machines are entirely non-random, implying or stating outright the games have a predestined sequence of wins. Others say that these machines have a pool of possible wins in a predefined batch that come out one at a like, like a US pull-tab game, or "drawing lots" in the bible. According to this theory, once an outcome is removed it can't be won again but the game otherwise randomly chooses from the remaining outcomes. Let me say that I believe these sources to be in error. For one thing, they contradict the patent I linked to earlier.


In conclusion, that is about the best I can do on that topic. Would I like to explain exactly how these machines work, as I have done for several American slot machines? Yes. Can I? No. If any slot engineer who has ever worked on a Fruit Machine should read this, please contact me. I would love to pick your brain on the topic. While I haven't taken this as far as I wanted, I hope you knew more about Fruit Machines than you did ten minutes ago.

Hi,

I haven't read through every post of this whole thread in great detail, but I'm very flattered that Mr S has made a reference to something I wrote with regard to UK slots (on the www.uk-21.org site). I captured this from a forum response on a similar topic that I found whilst researching the subject back in late 2012, made (allegedly) by a currently employed slot technician here in the UK. Regrettably, I've never been able to find this particular response since. I've never gone to the extent of reviewing the details of the patent(s) that may exist and been filed by the software houses that generate the code that underpins B3 gaming machines.

Something that may be of interest is the document at the end of this link (apologies if this has been referred to already), published by the UK Gambling Commission:

http://www.gamblingcommission.gov.uk/pdf/Gaming%20machine%20technical%20standards%20categories%20B3,%20B4,%20C%20and%20D%20legacy%20machines%20-%20July%202007.pdf

This document states in para 6.1 (page 11) that the RTP should be achieved within 100,00 cycles of the game, although it doesn't state in detail how this should be achieved. There is also the issue that, to the best of my knowledge, the UKGC doesn't have any active programme of auditing this requirement with licencees (of gaming machines), ie undertaking a statistical analysis of takings, payouts and the retained bit to see that the RTP is achieved, and that the results fall within statistical norms. So the whole issue of "compensated" gaming results behaving as they should is a regulatory black hole.

So the reference to "a week" in the posting above ( "... How far back it goes, I don't know, but believe it to be about a week, depending on how much the machine gets played . . ."), will largely depend on the volume of play put through the chip since it's last "reset" (presumably this must happen at some stage?) after having hit the pre-set RTP.

For clarity I would also suggest that B3 "compensated" gaming machines can be found in all UK casinos, in addition to slot arcades, pubs, local bookmaker offices and other retail outlets who may have applied for a licence to operate one. The B2 games. are "uncompensated" ones and rely on the pay-table to create the house edge - and as such are subject to unadjusted variance in the results experienced. They are also referred to as "Fixed Odds Betting Terminals" within the gaming and bookmaking industry, with the most commonly played game on them being electronic roulette - and the subject of much debate in the UK at present.

Sincere apologies if I've stated something that has previously been covered.


- - - -

On a completely unrelated note, I did read on Mr S's "Gambling in the UK" blog entry (https://wizardofodds.com/blog/gambling-united-kingdom/) made in February that most blackjack games in the UK are hand-shuffled. This is no longer the case, since the Grosvenor Group absorbed the former Gala Casino venues into it's portfolio, and as all Grosvenor and Genting casinos I've ever visited use the ubiquitous 1-2-6 CSMs, hand dealt games across the UK are now the exception rather than the rule. During my last casino visit, just two weeks ago, this particular Grosvenor outlet did have a single shoe-dealt blackjack game, although it carried a UK£25.00 per hand minimum - every other table card game was serviced with a 1-2-6. The LCI operated venues do, I'm told, offer 8 deck shoe games, although I've never played at one (as there is still a better game on offer in the Capital, at Napoleon's Casino in Leicester Square).

Be lucky !

Hi all

I'm a games producer in the social slots market but I started some 17 years ago designing and programming uk pub fruit machines so I hope I know what going on :)
I've just been pointed to this thread by a friend having just got back from Macau for G2e so I'll try and reply to all the posts on the train home as I'm on my phone at the moment but I might pass out from jet lag!

Dave

Just found this. Good Article.

Just a comment about the tipping signs. Tipping of gaming staff used to be prohibited in the UK. It has been allowed for approximately 2 years. The signs are there to educate customers of the fact tipping is now allowed. The vast majority of staff dont hustle for tips, but do accept them gratefully
and on a personal note I dont agree with the signs.

Quote: Wizard

In my blog entry on Gambling in the United Kingdom I tried to shed some light on the various classes of slots they have there and what they mean to the player. I admittedly did a lousy job of it, because I still don't understand it very well.

Any of this could be wrong, but my understanding is that slots in the UK at one time, sometimes referred to as fruit machines, were "compensated," meaning they were programmed to achieve certain return percentages in the short term. So, if a machine had been running cold then it would be designed to loosen up to achieve it's desired percentage.

On the contrary, "random" machines have the same odds every spin, which is how they work in the US.

This page at the UK Gambling Commission explains that class B3, B4, C, and D machines can be compensated OR random, while B1 and B2 must be random. The machines there did identify what class they were.

When I was there I noticed a lot machines stated "The outcome of any game or feature is not necessarily that shown by the odds displayed." in the rule screens. I tend to think these are the class B3, B4, C, and D machines, but could use a confirmation on that.

I guess my other main question is if someone is playing a B3, B4, C, or D machine how can he tell whether is he playing a compensated or random machine? As noted above, it can be either.

I've read a lot of legal mumbo jumbo on this topic, but does anyone really understand what is going on, who can explain it in plain simple English? I'm really looking to explain to the world if these "compensated" games still exist, and if so, how to identify them.

Thanks in advance for any help.

Hi Wizard - good site, so glad I've been pointed to it.

In terms of compensated fruit machines, most of the posts on here are way off the mark... some are pretty close.
I'll answer each post in turn (where relevant) and then i'll post a "layman's guide" to compensation at the end.

All games have to state whether they are compensated or random on the glass, or loading screen in the case of video terminals.

They certainly do exist, and make up the majority of category D and C games.

Category C games are £100 jackpots and £1 stake (maximum) and are found in pubs, service stations and bingo halls.
They range from simple three reel, 1 winline "bingo" Cat C's (these are very simple games with a single feature) to the very complex ones you get in pubs.
I've designed, programmed, and written the control logic for all types of Cat C machines.

Category D machines are the low stake and prize machines, which are normally found in seaside arcades - these are the only Category under 18's can play, and include low stake fruit machines (identical to Cat C other than the stakes and prizes) as well as Crane Grab machines, Pushers, etc...

Dave

Quote: Wizard

To answer my own question, I think that the "compensated" games are what we would call "pull tab" games here. They are most likely to be found in bars, fish-n-chip shops, and other places that don't want to take the chance on a fully random game.

Source: http://www.casinolifemagazine.com/article/gambling-professor-explains-british-casino-culture. In particular

They most definitely aren't operated as a "pull-tab" system. There were a few manufacturers who did do their B3 random machines like that, which is actually against the tech standards, as for a game stated as Random must have the same chance of every outcome on every spin (that doesn't mean every outcome has to have the same chance of course).

Someone later in the thread mentioned Category B3A machines, which are only found in working men's clubs really, and do operate as a "lottery" or "pull-tab" system with a pre-set seed

Dave

Can you explain exactly how "compensated" slots work? And, if they are not like pull-tabs, how are they different?

Quote: DRich

Quote: Wizard

Can you elaborate on how they do that?

I spent six weeks in Spain working on some slot machines for a Spanish manufacturer. Although the machines I was working on were typical Class 3 slots they also manufactured AWP (Amusement With Prize) machines. I was told those machines had multiple payback percentages and that if the actual results were outside some parameter they would change to a tighter or looser schedule to get back within the desired payback percentage range.

This is kind of true, and those who don't fully understand the way we compensate games often think this is what we do (i.e have one set of maths at 86%, one at 94% and one at 106% for example) but i've never come across a UK company that does this. I'll explain more shortly....

Ooops, duplicate post....

Quote: MathExtremist

I haven't been involved in the UK machine market, but from what I've read in this thread the compensation is simply an evaluation that if the recent actual payback has been outside some target RTP range, the game swaps in a tighter or looser set of reels for a while until the actual RTP gets back within tolerances.

Also, finite pool games are operated in the UK under category B3A. I don't believe any of the rest of the categories are based on finite pools, and a finite pool game doesn't need to compensate anyway because the total pool payback is known in advance. That's one of the primary reasons to using a finite pool rather than an independent random game. The deal may be shuffled so winners come out early or late, but the total payback is constant so you wouldn't ever need to compensate for anything.

This is a lot closer to what actually happens, but get "sets of reels" out of your head... it's nothing to do with sets of reels, excel, standard deviation, etc... most programmers who do UK fruit machines have never even heard of standard deviation! :)

Quote: Mission146

I don't believe that you could meaningfully exploit it, other than having a general idea of when NOT to play...which is probably never.

The reason is that, I would assume, for a certain pay to be advertised, then that pay probably has to be possible, IOW, at least one occurrence in every set, if not more. Let's say you have pays on a machine of 500-FOR-1, 100-FOR-1, 50-FOR-1, 25-FOR-1, 10-FOR-1, 5-FOR-1, 2-FOR-1 and 1-FOR-1...Okay, now for every single one of those things to hit once (100% payout, which certainly isn't the case) it would be 693 plays. Furthermore, the lower pays are certainly going to occur more frequently, by design, than the higher pays, so you're obviously going to have far more than 693 plays.

If you look at Wizard's analysis of the fruit machine he designed:


Note: From this point forward and only for purposes of my example, I am treating Wizard's fruit machine as a single-line game.

You'll see a paytable similar to the theoretical one I posed above would complete a pre-set cycle at 8000 spins based on the way Wizard set that machine up, however, Wizard's machine also has an extremely low house edge of 3.5%. I have no way of knowing whether or not these British fruit machines are lower in the pubs, but if I had to guess...

As has already been mentioned, new seeds can probably be added while a current seed is in progress, furthermore, a machine may be able to start with multiple seeds programmed into it in the first place. In the unlikely event the house ever gets way behind on the machine, assuming no programming errors, they would simply be able to let the seeds already loaded finish playing out before adding new ones.

Therefore, unless you watch the machine from the installation of it, know how many seeds the machine started with having been loaded, know the fixed RTP of each individual seed, and track all of the spins, then you have no way of knowing when you are at an advantage.

For instance, let's go back to Wizard's fruit machine. If we assume that only one seed is loaded into the machine at a time and is allowed to play all the way out, (and this is really a best case scenario for someone hoping to track it) then you could watch 2,000 spins with returns as follows:

One Cherry 400/2000 = 800
Two Cherries 30/2000 = 150
Three Cherries 2/2000 = 20
Three Oranges 7/2000 = 105
Three Bells 10/2000 = 200
Three Plums 2/2000 = 100
Three Bars = 0
Three Globes = 0

Okay, so the overall RTP in this 2,000 spin sample was 1375/2000 = .6875 or 68.75%

A single starting seed would have 8000 * .965 = 7720 in pays and 8000 in bets

If we subtract the 2,000 spins from both sides, we are left with 6,000 in bets and 6345 in pays, based on what you have seen. The remaining RTP, based on what you have seen is 105.75%, which is clearly awesome. Everyone likes trading 6,000 for 6,345, and actually, since we have this information, we know to quit ANYTIME we are more than 345 units ahead!

Okay, so you decide to play out the remaining 6k in bets, however, the previous day two people came in and played 800 total spins with the following combined results:

One Cherry: 190/800 = 380
Two Cherries: 20/800 = 100
Three Cherries: 1/800 = 10
Three Oranges: 7/800 = 105
Three Bells: 4/800 = 80
Three Plums: 0/800 = 0
Three Bars: 1/800 = 100
Three Globes 1/800 = 500

That's 1275 in pays v. 800 in plays, which brings us to 5200 remaining in seed plays and 5070 in seed pays, and the RTP on the rest of the seed is back down to .975 or 97.5%, which is still better than it started, but the player is at a disadvantage, and unbeknownst to the player, the best pays remaining are 7/9 Three Plums.

CONCLUSION

These things could theoretically be exploited, but a ton of information is needed, the more information the better, but at a minimum:

1.) What is the fixed RTP, per seed?
2.) How many seeds did the game start with, or how many seeds are left?
3.) Since the beginning of the current seed(s), how much has been bet and how much has been returned?

The probability distribution of results would also be nice to know, but strictly speaking, isn't necessary as long as you know how much the pool started with in terms of RTP and how many total bets that reflects, (i.e. House Edge) as well as the results of every spin.

If there is a jurisdictional minimum RTP on these things, then you could always assume that, but you'll not know where you are in a seed on anything except for a brand new machine, and even then, you may not know how many seeds have been loaded at a time.

Simply put, if you watch someone play one of these machines at a lesser return than jurisdictional minimum, or a return less than the base House Edge, (if known), but all that will really tell you is if the House Edge is better or worse than when you walked in.

If you have a barkeeper friend, however, who is in charge of seeding the machines, collusion could be possible just based on knowledge they would have of the machine's take (during a seed, if loaded one at a time) vs. the machine's pays for that seed.

Sadly, Mission, this couldn't be more wrong as to how fruit machines work. To me, compensators are even simpler than that, but then i've been using them for 17 years :) I'll try and explain it in a seperate post shortly, but there are no seeds - although there ARE sharks who learn how the games play (their profiles).

Quote: Wizard

I've been doing some more digging and found some patent numbers that address these "compensated" slots. They are GB2185612 and GB2087618. Based on my layman's reading of the latter, it seems the way these fruit machines work is they keep a running record of the return over the past x spins. Then it adjusts the probability of certain bonus features of the game according to whether the game needs to pay out more or less to get closer to its desired return percentage. It seems to me that there isn't just three settings (tight/average/loose) but a sliding scale according to the probability of the bonus feature(s).

That is about as specific as I can get. GB2185612 was not very helpful and I think what I did get of GB2087618 was not the full thing. American patents tend to run hundreds of pages, by comparison. If anyone wants to try to dig deeper, I used this site.

There may be patents to do with compensators, but they were never enforced... and they are so wooly as to be unenforcable anyway - plus patents in the UK have to be inventions to stand up - for example, you can't patent a game design, feature, reel spin, etc... In the UK, 99% of all US slot patents would be completely unenforceable!

Here's, hopefully, a layman's guide to compensators - i'm going to keep it as simple as possible and skirt around the complex pub fruit machines for the moment, which have multiple compensators....

Why Compensate?
So what do we mean by compensated and why do manufacturers use them? Well, put simply it refers to the ability of the machine to control the experience the player receives by keeping track of money in and money out. Where a random game has the same chance on every spin of showing the player the same outcome (be it a progressive jackpot, a losing spin, or 3 bonus symbols) a compensated game can look at how much money it has stored and make a decision as to whether to give the player a specific outcome or not.
The second, and possibly the most important reason for using compensators, is that the volatility is controlled. This means that for an operator, cash-in-box (the take at the end of the week) is much more stable than for a random slot, hence why pub machines (which the landlord takes a proportion of) are compensated.
In the UK most compensated games run somewhere between 72% and 90%, the lower percentages being in pubs and the higher ones in arcades.

How Do Compensators Control RTP?
Compensators almost always have a top and a bottom which is commonly what we refer to as the range; on current £100 jackpot Category C games this is normally around £300, but this varies greatly my manufacturer and on an even more granular level – by game. As a player plays a compensated game they are adding money from their stake to the compensator(s), and when they win money from the game, the money comes off the compensator(s). This addition and subtraction of money from the compensator(s) is what keeps the machine hovering on or around its RTP. If we assume that the game is going to start throwing money at you whether you like it or not when the compensator hits its highest threshold (the top we referred to earlier) and will not let you win anything if it drops below its lowest threshold (the bottom), then the worst case scenario for the game is that the game will owe £300 at any one time. This means that after 100,000 games at £1 per play, the RTP will be at worst 0.3% off 80%. This is perfectly acceptable under UK Law with regards to compensated games.

How Do the Developers Come Up With the Maths
In my experience, the control of compensated games is done purely and simple in the developers head. Excel simulations are in the most part irrelevant – I occasionally used it to work out the averages for features, but other than that it’s down to trial and error and getting the feel of the game right. This is because as the game moves through the levels of the compensator (see below) you want the feel of the game to change, and as the feel changes, the decisions that you want the game to make, and the factors you want it to take in to account in order to make those decisions, change. Standard Deviation is not required, and in fact I hadn’t even heard of it until I started to do slot machines.

Single Compensator Game – Simple Game Example
Let’s start with a simple compensated game – one with only one compensator, running at 80%, with a stake of £1. Assume the game has 3 reels, each of 16 symbols and in essence looks like a simple slot. At the start of a game, when the player presses the start button, the game would take the £1 from the credit meter and add 80p on to the compensator. Therefore after the first game, the compensator now has 80p available for the player to win. The game then chooses 3 reel positions at random. It checks the outcome of the reel spin, and if it has enough money stored up to pay the outcome to the player, it allows the spin. If it doesn’t, it throws the result away and repicks. It repeats this process until an outcome is picked that it can afford.

So, let’s assume that our compensator is not just used for a simple YES/NO style decision. Let’s assume that the compensator is used to alter the chance of making a decision based on the amount of money it has saved up so that the more money the game has on it, the more chance it has of awarding the higher prizes. Now, let’s assume that our simple game has a range of just £100. We now divide this up in to levels as follows:
Level 0 – Less than £0
Level 1 - £0 to £19.99
Level 2 - £20 to £39.99
Level 3 - £40 to £59.99
Level 4 - £60 to £79.99
Level 5 - £80 to £99.99
Level 6 - £100+

Where the compensator value starts from clear down/reset varied by game and programmer in my time of doing compensated games – but I tended to start the game with at least some money on it for the players to win - £20 in the lifetime of the game is nothing, so let’s assume we start with £20 on the compensator.
At the end of every spin, the compensator would look at the value on it and set the relative level. So with £20 on the game, the compensator would be said to be in Level 2.
Here’s the table we are going to use to make our decision as to whether to allow the outcome of the spin or not. Assume the game does this as a decision process:
START GAME -> PICK OUTCOME -> PASS CHANCE OF OUTCOME ->YES -> SHOW RESULT
-> NO -> PICK AGAIN
You may notice a few odd things as you look through the table, which I will explain below….
Fruit Win Chance
Prize Lvl 0 Lvl 1 Lvl 2 Lvl 3 Lvl 4 Lvl 5 Lvl 6
No Win - 100 100 99 98 96 93 90
Cherry £2 0 100 100 90 80 60 70
Grape £5 0 50 75 100 100 90 80
Melon £10 0 20 40 50 65 80 90
Plum £25 0 0 5 25 50 90 100
Red 7 £50 0 0 0 15 40 60 80
Bell £100 0 0 0 0 0 0 50
*all values are chances out of 100… i.e. to allow a Melon in Level 1 is a 20 out of 100 chance
One of the odd things you may have noticed is the in Level 2, I allow a £25 win on a 5% chance, even though the compensator could have between £20 and £39.99. Allowing compensators to go “negative” is, again, personal choice but as the bands don’t sit perfectly with the wins available and never will, it’s quite normal to allow the game to overpay – you’ll notice in Level 0, there is 0 chance for all wins, so it will quickly get back to Level 1 and in to a positive position.
The other thing you might notice is that the chance of the low wins goes down in the higher levels, as does the chance of allowing a losing spin. This means that the game is now positively curating your experience – the more money it has available, the more it is trying to give it to you. This is very common on compensated games. Some old fruit machines used to just spin in a Jackpot when the compensator got too full!
Let’s take the example in the simple game earlier. The player presses start, the machine picks a random set of reel positions which is going to award £10, which is 3 Melons.
It then uses the compensator level to reference the table above – so it will pass a 40 in 100 chance to determine if it can allow the spin. If it fails, it will pick another set of reel positions.

Here's, hopefully, a layman's guide to compensators - i'm going to keep it as simple as possible and skirt around the complex pub fruit machines for the moment, which have multiple compensators....

Why Compensate?
So what do we mean by compensated and why do manufacturers use them? Well, put simply it refers to the ability of the machine to control the experience the player receives by keeping track of money in and money out. Where a random game has the same chance on every spin of showing the player the same outcome (be it a progressive jackpot, a losing spin, or 3 bonus symbols) a compensated game can look at how much money it has stored and make a decision as to whether to give the player a specific outcome or not.
The second, and possibly the most important reason for using compensators, is that the volatility is controlled. This means that for an operator, cash-in-box (the take at the end of the week) is much more stable than for a random slot, hence why pub machines (which the landlord takes a proportion of) are compensated.
In the UK most compensated games run somewhere between 72% and 90%, the lower percentages being in pubs and the higher ones in arcades.

How Do Compensators Control RTP?
Compensators almost always have a top and a bottom which is commonly what we refer to as the range; on current £100 jackpot Category C games this is normally around £300, but this varies greatly my manufacturer and on an even more granular level – by game. As a player plays a compensated game they are adding money from their stake to the compensator(s), and when they win money from the game, the money comes off the compensator(s). This addition and subtraction of money from the compensator(s) is what keeps the machine hovering on or around its RTP. If we assume that the game is going to start throwing money at you whether you like it or not when the compensator hits its highest threshold (the top we referred to earlier) and will not let you win anything if it drops below its lowest threshold (the bottom), then the worst case scenario for the game is that the game will owe £300 at any one time. This means that after 100,000 games at £1 per play, the RTP will be at worst 0.3% off 80%. This is perfectly acceptable under UK Law with regards to compensated games.

How Do the Developers Come Up With the Maths
In my experience, the control of compensated games is done purely and simple in the developers head. Excel simulations are in the most part irrelevant – I occasionally used it to work out the averages for features, but other than that it’s down to trial and error and getting the feel of the game right. This is because as the game moves through the levels of the compensator (see below) you want the feel of the game to change, and as the feel changes, the decisions that you want the game to make, and the factors you want it to take in to account in order to make those decisions, change. Standard Deviation is not required, and in fact I hadn’t even heard of it until I started to do slot machines.

Single Compensator Game – Simple Game Example
Let’s start with a simple compensated game – one with only one compensator, running at 80%, with a stake of £1. Assume the game has 3 reels, each of 16 symbols and in essence looks like a simple slot. At the start of a game, when the player presses the start button, the game would take the £1 from the credit meter and add 80p on to the compensator. Therefore after the first game, the compensator now has 80p available for the player to win. The game then chooses 3 reel positions at random. It checks the outcome of the reel spin, and if it has enough money stored up to pay the outcome to the player, it allows the spin. If it doesn’t, it throws the result away and repicks. It repeats this process until an outcome is picked that it can afford.

So, let’s assume that our compensator is not just used for a simple YES/NO style decision. Let’s assume that the compensator is used to alter the chance of making a decision based on the amount of money it has saved up so that the more money the game has on it, the more chance it has of awarding the higher prizes. Now, let’s assume that our simple game has a range of just £100. We now divide this up in to levels as follows:
Level 0 – Less than £0
Level 1 - £0 to £19.99
Level 2 - £20 to £39.99
Level 3 - £40 to £59.99
Level 4 - £60 to £79.99
Level 5 - £80 to £99.99
Level 6 - £100+

Where the compensator value starts from clear down/reset varied by game and programmer in my time of doing compensated games – but I tended to start the game with at least some money on it for the players to win - £20 in the lifetime of the game is nothing, so let’s assume we start with £20 on the compensator.
At the end of every spin, the compensator would look at the value on it and set the relative level. So with £20 on the game, the compensator would be said to be in Level 2.
Assume the game does this as a decision process:
START GAME -> PICK OUTCOME -> PASS CHANCE OF OUTCOME
->YES -> SHOW RESULT
-> NO -> PICK AGAIN
In each of the different compensator modes, the chance of awarding each fruit win changes, so in Level 1 there may be only a 1 in 100000 chance of allowing the Jackpot (£100) but in Level 6, that chance might be 50%. In fact, we might even just decide to force the jackpot on a 5% chance if the game hits level 6 - this is how we "top games out"

Does any / all / some / none of that make any sense so far?

Yes, it is making sense. Very interesting and thank you for contributing.

Games such as those you see in the pubs can, and do, have multiple compensators.
One could be a "main" compensator, which is the one the game makes most of the decisions on - this might add 87% of the 90% the game is runing at
One could be a "jackpot" compensator, which saves up in the background just to make sure a Jackpot is paid out - when these get full the game normally finds a nice way of forcing a jackpot on the player e.g. INVINCIBLE MODE! - this might add 3% of the 90% the game is running at.
There are many, many other uses for compensators - some just control feature entry, but don't control the amount of money you win from the feature.

I've done a game in the past that had 6 compensators on it - it was moving money around in the background all over the place depending on what the player did or didn't do when playing the game - i guess you could refer to this as adaptive logic, but we just called it "control".

For the complex features on UK compensated fruit machines, we almost always use a system called "MAX WIN". This limits the amount of money you can win from any feature to a value picked at the start of the feature and therefore stops the game overpaying (paying more than it has on the compensator) - this doesn't mean you are guaranteed to reach that value, but just that if you do, and you try to carry on, you'll get GAME OVER very soon.

That's a very brief, and very cutdown look in to the world of compensation - any questions gladly welcome! :)

Dave

This is great stuff, thank you! I would like to discuss me hiring you to write a formal article on this topic. Please send me a Private Message if you're interested.

Trance,

Welcome to the forum, and thanks so much for the wealth of information!

Thanks for the info, great stuff!

No worries guys. Like i said, this is all the tip of the iceberg for compensated games :)
But hopefully you get some of the principle ideas

Quote: Trancemonkey

No worries guys. Like i said, this is all the tip of the iceberg for compensated games :)
But hopefully you get some of the principle ideas

I'd still like to pick your brain about this further and am willing to pay for your time. My goal is to make my own Compensated game and explain to my readers how I did it.

I've just picked up on the contributions to this thread since my last posting a while back. Very interesting - particularly the info re the "top" and "bottom" limits pre-programmed for "wins/losses" (over/under?) relevent to the RTP for the total amount paid in by players.

A question though for the new expert witness - do large "jackpot" payouts ever result in the running balance falling out well below the pre-programmed "bottom", and if so what happens to bring the balance back into the "zone"??

I shall re-write the page on my own site, and unless the Wizard has any objections put a link into this thread.

Cheers.

In the vast majority of compensated/controlled games, the "Large jackpot payouts" would either be stored up for seperately, or the machine would only pay them out once it had stored up enough money on the main compensator. As the prospect of someone losing enough money to pay out a Jackpot plus one or more repeats is almost impossible now due to the £100 jackpot level, most games store up the money for paying large wins (or repeated large wins) seperately.

As games are not allowed to force an enriched period on to players, the games have to be clever about how they gift the money to the player(s) once the "Jackpot" compensator is full. I can only speak for me here, and I used to do different things on different games, but one3 way of doing this would be as follows:

Assume a price of play of £1 and an RTP of 90%
Main Compensator - add 87p every game
Jackpot Compensator - add 3p every game
The Jackpot Compensator HAS to save up a minimum of £200 before it has the chance of triggering
Once it gets there, it has a 1 in 1,000 chance every spin of dumping all the money in it on the the Main Compensator.
If the Jackpot Compensator reaches £300, it would automatically dumped all the money on to the Main Compensator.
This would then cause the Main Compensator to have a large amount of money on it.
This would mean that the next time the player got in to the feature game, the machine could afford to pay a Jackpot + repeats
Once the player achieved the feature, the game would go in to "Invincible Mode" or "Go All The Way" where the logo may be illuminated in Red, or flash, or some other tell.
If the player doesn't do anything stupid, and achieves the Jackpot (and the game would help AS MUCH AS POSSIBLE to get there), then the repeat chances would be 50% up to a maximum of 2 repeats.
If the machine didn't pass the chance of the repeat, the money would still be on the main compensator, which means there is now MORE chance of it picking higher wins every time the player enters the feature, and may even pass the chance of another "Invincible Mode".

When deciding whether to give repeats or not, I would always look at the amount of money on the compensator AS WELL AS the chance of awarding the repeat before allowing the repeat to happen, eg:

MAIN COMP - £75.20
MAIN COMP LEVEL - 4
REPEAT WIN VALUE = £60
REPEAT WIN CHANCE = 50%
In this case, if the 50% chance was passed, I would allow the repeat through

MAIN COMP - £59.20
MAIN COMP LEVEL - 4
REPEAT WIN VALUE = £60
REPEAT WIN CHANCE = 50%
In this case, if the 50% chance was passed, I would NOT allow the repeat through

Does that all make sense?

TranceMonkey