Return to Player - Critiques Requested.

4 of a kind

Repeated violations of forum rule 1.16 - troll
Joined
Mar 11, 2009
Location
New York
it is awesome that this information is being presented to the players. I find it interesting that only 15% of the players use the help. Maybe having a button that the players could select while playing the games would increase the awareness that this information is available to them.

On the same note, 3 Dice has been reporting similar information to the players for quite sometime. It is definitely the right approach for on-line casinos to take when showing transparency to the players.

In the states, I haven't encountered a land-based casino that reports this information to the players. To be fair, I haven't gone up and asked the attendants or casinos either.

Interesting read.

You could find this information out with land-based casinos in the states by visiting the gaming commissions web sites, and review their monthly reports like these two I posted from the NJCCC for Atlantic City Casinos here...https://www.casinomeister.com/forums/threads/online-rtp-settings.45680/

Or, you could subscribe to "Casino Player Magazine" and read a monthly printout of the RTP's like I posted here... https://www.casinomeister.com/forums/threads/b-m-payouts.36710/
 

KasinoKing

WebMeister & Slotaholic..
webmeister
PABnonaccred
CAG
MM
Joined
Aug 25, 2004
Location
Bexhill on sea, England
I would like to modify the link that is contained in my first post of this thread. However, I cannot see how to do that. (I don't see any sort of Edit button associated with the post.)

Can anyone help me out here? How do I modify the first post so that the link doesn't go dead?
Posts can only be edited by members up to 2 hours after they are first made.
If you want something changed now, PM a mod (CM, Maxd, Simmo or Webzcas), or use the "Report a Post" function to request your change.

KK
 

thelawnet

Dormant account
Joined
Apr 4, 2005
Location
UK
I would add a column for standard deviation and a brief description to note that if game A has a standard deviation of 1 and game B has a standard deviation of 4 then, in the long run, at the same stake size, game B will have swings - up or down - four times as large as game A.

I would also note that the house edge for the skill-based games is the theoretical minimum house edge assuming perfect play.
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
KasinoKing - OK, I'll try a PM. Thanks.

thelawnet - Re: standard deviation. We decided against doing that because to do it properly, for all types of games, would really blow up the size of this page. We do give a nod to "variance" at the top of the page.

Re: Theoretical Minimum House Edge. Good idea, and done.

Chris
 

Tengil

Senior Member
Joined
May 4, 2006
Location
Finland
KasinoKing - OK, I'll try a PM. Thanks.

thelawnet - Re: standard deviation. We decided against doing that because to do it properly, for all types of games, would really blow up the size of this page. We do give a nod to "variance" at the top of the page.

Re: Theoretical Minimum House Edge. Good idea, and done.

Chris

Would you still give that information if one asks?
 

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
KasinoKing - OK, I'll try a PM. Thanks.

thelawnet - Re: standard deviation. We decided against doing that because to do it properly, for all types of games, would really blow up the size of this page. We do give a nod to "variance" at the top of the page.

Well variance = standard deviation squared, so if you list variance values, standard deviation values are square root of these.

As for Game Group 3, which includes games with bet raises, it might be helpful if you would write "House edge relative to total amount bet" (ie. Element of risk) versus "House edge relative to initial bet" (which is the standard definition of house edge). Now you employ terms "Theoretical Minimum House Edge" and "Theoretical Maximum RTP" which don't express this difference and these terms might be confusing to some (ie. why are the latter values lower than the first ones).
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
Spiderlegz,

The short answer to your question - yes. I'm not trying to hide anything here, I'm trying to disclose things, and trying hard to do so in a meaningful and helpful manner.

A somewhat longer answer - I am of the opinion that someone who truly understands a particular subject is able to explain that subject to any audience such that they leave with a greater understanding of the subject than when they entered. (Personally, I've always found examples and analogies extremely helpful for me when I am in the audience, so I tend to use them when I am a speaker.)

Our objective in creating this Theoretical RTP page was to present both the raw data (tables of numbers) for the subject, but also (and more importantly) present a clear, accurate, and hopefully understandable explanation of what the data means to a wide audience.

There are about a dozen or so technical terms already included in the current page, which is a heck of a lot.

There are also a huge number of words in this current file, which I'm aware is a little self defeating. That is, the members of the audience which would benefit most from a clear and accurate description of the subject are initially intimidated by the visual size of the page.

So, before including Standard Deviation values, I would need to satisfy myself that I completely understand exactly what these values mean relative to any and all appropriate aspects of our various games. That would allow me to create, or edit, a description of the subject such that it provides helpful and understandable information to a wide audience.

And finally, we would also need to either gather, or calculate, these values for all of our games. I confess that I don't have them immediately at hand.

I'm sure, given your previous posts, that you prefer my short answer. (If I were the reader, I would too.)

Chris
 
Last edited:

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
Jufo,

In the current page, our use of "Theoretical Minimum House Edge" is the standard definition of house edge; that is, "expected win/loss relative to initial bet only" (assuming statistically perfect game play).

Our use of "Theoretical Maximum RTP" is 100 - "expected win/loss relative to total bet"; that is, 100 - Element of Risk (again, assuming statistically perfect play).

I'll spend some time thinking about words I can add or modify for the final text of this Game Group 3 description that will make this clearer.

Chris

Edit: I just made a few changes that I think do a better job at clarifying what is being reported in the data table, and why.
 
Last edited:

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
Jufo,

In the current page, our use of "Theoretical Minimum House Edge" is the standard definition of house edge; that is, "expected win/loss relative to initial bet only" (assuming statistically perfect game play).

Our use of "Theoretical Maximum RTP" is 100 - "expected win/loss relative to total bet"; that is, 100 - Element of Risk (again, assuming statistically perfect play).

I'll spend some time thinking about words I can add or modify for the final text of this Game Group 3 description that will make this clearer.

Chris

Edit: I just made a few changes that I think do a better job at clarifying what is being reported in the data table, and why.

Ok I looked at the description and it is clearer now. After thinking it more, wording "Theoretical Maximum RTP" makes sense if you think of it as the ratio: total amount of money player puts in / total amount of money casino pays out, which is the same as "element of risk" (NOT house edge). I have seen even professionals confuse the two.

As for whether variances should be included, there was an interesting thread at Wizard of Odds' own discussion forum about
You do not have permission to view link Log in or register now.
It turned out it was not just the house edge figure to determine a good bet, but rather the ratio: house edge/variance, so from this perspective house edge is only half of the story, and to make an informed decision what to play on, the gambler should always consider both house edge and variance (or standard deviation).
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
Jufo,

Good. I'm glad that the changes met with your approval. I agree that the previous words were not as clear as they needed to be.


Re: "I have seen even professionals confuse the two."

As part of the research I needed to do to prepare myself for properly editing this article, I also discovered that this was true.

This was one of the reasons that we felt it necessary to address this specific subject clearly, completely and accurately.

All in all, I think we did a pretty good job at stuffing 10 pounds of crap into a 5 pound bag.


Re: "As for whether variances should be included, ..."

For an article whose title is "How to quantify 'a good bet'", I agree that a detailed discussion of variance is an appropriate topic for inclusion in that scope.

However, for an article whose title is "Return To Player (RTP) Percent Summary - All Games", I think our nod to variance regarding its impact on short term RTP is sufficient. As I mention above, I'm concerned that the current size is already self defeating, which is a little frustrating given the effort we have put into it.


Chris
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
Jufo,

After responding to your post in this thread, I followed your WizardOfVegas link and read through the thread over there.

Off the top - We are writing (or trying to write anyway) to a much broader audience than Mike. I'm proceeding under the assumption that any technical term which we use needs to first be explained to some extent.

For example, we would not use the term Standard Deviation without first providing a satisfactory explanation - one that does not require extensive knowledge of Mathematics to understand - of exactly what that term means and how it applies to the subject at hand.

I believe that Mike can safely proceed from a different baseline.


Data which you may find interesting:

The House Edge of our Stud Poker game, at 3.35%, is lower than the "typical" House Edge for this game of 5.22%. Still, with a sample size of 20 million games, we were initially left to wonder: why, in a 20-million game sample, is the Reported RTP for our game, at 97.48, not in better agreement with its Theoretical RTP of (100 - House Edge) = 96.65%?

Indeed, given that 20 million games is a pretty decent sample size to both factor out variance and factor in the higher payouts, we would also expect that the Reported RTP would be significantly less than the Theoretical RTP (say, 1% or so), because Players don't all, or always, play "statistically perfect" games.

Given this discrepancy, we asked ourselves: what is the most important value for the Player, the Reported RTP or the Theoretical RTP? Our answer: the Reported RTP, because it is the same for all games, and because it answers the Player-critical question - "How much money has gone into, or come out of, my pocket?"

We put forward that Theoretical RTP is of value to the Player only if it:

1. provides an accurate "apples to apples" comparison of different games.

2. provides an accurate prediction for the question "How much money can I expect to go into, or come out of, my pocket?"

You can see, I believe, that we repeatedly hammered these points (ad nauseum) throughout the article.

One of the reasons that we have done all of this is in reaction to the current reporting requirements, and in anticipation of future reporting requirements, for controlled gaming jurisdictions. However, if even the "Professionals" within our industry haven't, or can't, come to agreement on this, then what hope is there for the Player (or the lowly software provider)?

I anticipate the need, at some point in the future, of having to "go to battle" over this issue. I thus want to make damn sure that we are well armed.

To complete the Stud Poker data above: With a Reported RTP of 97.48%, and an Element of Risk of 1.64%, (100 - Element of Risk) = 98.36%, leaving an additional (and expected) 0.88% arising from deviations from "statistically perfect" game play.

Our conclusion: If we are doing all of this reporting not so that Professional can talk to Professional, but rather so that Casino can talk to Player, then a definition of Theoretical RTP as (100 - House Edge) is useless (or at least not as helpful) as a definition of (100 - Element of Risk) for this group of games.

(Squire, fetch me my armor that I may prepare for the battles to come.)

Chris
 

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
Hi, I posted the WizardOfVegas link more as an off-topic note that low house edge doesn't always mean a good a bet for the player which can be seen from the example in the first post of a game where you wager $1 and always get $0.995 back, making it impossible to get ahead. So I was just trying to make a point that the variance is an important factor for the player because only with variance the player can hope to get ahead, regardless of house edge.

To complete the Stud Poker data above: With a Reported RTP of 97.48%, and an Element of Risk of 1.64%, (100 - Element of Risk) = 98.36%, leaving an additional (and expected) 0.88% arising from deviations from "statistically perfect" game play.

Yes those figures sound plausible. It's good that you realized the reason for the RTP 97.48% which seemed to contradict the theoretical house edge (96.65% return relative to initial bet), and you saw that it was just down to different definitions of house edge vs. element of risk, and not a payout mistake in the programming of the game.

Our conclusion: If we are doing all of this reporting not so that Professional can talk to Professional, but rather so that Casino can talk to Player, then a definition of Theoretical RTP as (100 - House Edge) is useless (or at least not as helpful) as a definition of (100 - Element of Risk) for this group of games.

Yes I'd agree.

(Squire, fetch me my armor that I may prepare for the battles to come.)

I don't disagree with what you posted so I am not sure if there is going to be any battle over this ;)
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
Jufo,

I think I understand what you were getting at with the link to Mike's "Socratic Method" thread: The fun of gambling is the hope/expectation that today’s action is going to intercept the distribution curve at the higher end, the winning end, of the X axis. If there is no distribution curve (Mike's "guaranteed to lose a penny, but only a penny" game), then there is no hope, so what's the point?

I saw the discussion in Mike's thread as about how variance addresses the fun of gambling. That is, only variance can provide you with the possibility of having some days come out winners.

I think we both agree that other days are going to be losers, and in total the House Edge (however it might be defined) is ultimately against us. So, even though a million games against a House Edge of 0.1% may leave us at the same point as Mike's "guaranteed to lose a penny, but only a penny" game, we had a hell of a lot more fun getting there.


Where I'm coming from: if you read through the posts on Mike's thread, the audience's understanding of a whole lot of game-specific or statistics-specific terms is simply a given. The audience baseline is a great deal higher than I think Galewind's baseline can be/should be/needs to be for our Help system.

(Ask a Casino Player: "What's your car's MPG?" versus "What's your game's RTP?" They may not know their car's MPG, but I'd bet they at least know what you're talking about.)

I figure that Galewind needs to start with the basics and work our way up from there. And we both agree that even something as basic as understanding a Casino's RTP Audit results is a bit of a climb.

As to my "Squire, fetch me my armor ..." comment: that was a tongue-in-cheek reference to my acceptance speech for Casinomeister's Forum Highlight of the Year for 2009.

And the battles to which I refer are not with you my good Jufo, but rather with the Regulatory Agencies and Certification Laboratories that are beginning to grow as the online gambling world changes. I have already had a conversation with one Certification Lab in which they informed me that Theoretical RTP for Stud Poker is (100 - House Edge). When I raised the issues which you and I have discussed here, the conversation on their side went dead.

I'm concerned that I might be forced to do something that doesn't make any sense simply to gain "compliance", which really chafes my shorts.

Chris
 

Tengil

Senior Member
Joined
May 4, 2006
Location
Finland
Jufo,

I think I understand what you were getting at with the link to Mike's "Socratic Method" thread: The fun of gambling is the hope/expectation that today’s action is going to intercept the distribution curve at the higher end, the winning end, of the X axis. If there is no distribution curve (Mike's "guaranteed to lose a penny, but only a penny" game), then there is no hope, so what's the point?

I saw the discussion in Mike's thread as about how variance addresses the fun of gambling. That is, only variance can provide you with the possibility of having some days come out winners.

I think we both agree that other days are going to be losers, and in total the House Edge (however it might be defined) is ultimately against us. So, even though a million games against a House Edge of 0.1% may leave us at the same point as Mike's "guaranteed to lose a penny, but only a penny" game, we had a hell of a lot more fun getting there.


Where I'm coming from: if you read through the posts on Mike's thread, the audience's understanding of a whole lot of game-specific or statistics-specific terms is simply a given. The audience baseline is a great deal higher than I think Galewind's baseline can be/should be/needs to be for our Help system.

(Ask a Casino Player: "What's your car's MPG?" versus "What's your game's RTP?" They may not know their car's MPG, but I'd bet they at least know what you're talking about.)

I figure that Galewind needs to start with the basics and work our way up from there. And we both agree that even something as basic as understanding a Casino's RTP Audit results is a bit of a climb.

As to my "Squire, fetch me my armor ..." comment: that was a tongue-in-cheek reference to my acceptance speech for Casinomeister's Forum Highlight of the Year for 2009.

And the battles to which I refer are not with you my good Jufo, but rather with the Regulatory Agencies and Certification Laboratories that are beginning to grow as the online gambling world changes. I have already had a conversation with one Certification Lab in which they informed me that Theoretical RTP for Stud Poker is (100 - House Edge). When I raised the issues which you and I have discussed here, the conversation on their side went dead.


I'm concerned that I might be forced to do something that doesn't make any sense simply to gain "compliance", which really chafes my shorts.

Chris

Thats quite facepalm worthy.

Another problem with those agencies etc is that they really dont prove if a game is random in their monthly audits.
A game with a return of ~99,5% (HA being ~0,5%) over a large sample doesnt really prove its random. Who knows what kind of streaks have occurred?
 

Nifty29

Dormant account
Joined
Jun 20, 2001
Location
Turn right, then right. then right again
AFAIK, the auditing process used by online casinos involves the casino submitting a whole bunch of raw data, and the auditing firm signing off that it all adds up (if I'm wrong I stand corrected).

I must admit that this has always bothered me.

It seems that a lot of players think that the auditors visit the casino offices and run software fairness tests and go through all their books with a fine toothed comb.....whereas the reality is quite different.

It's almost like a rubber stamp IMO.

What's your view Chris?
 

Nate

Paleo Meister (means really, really old)
webmeister
CAG
Joined
Feb 14, 2007
Location
Cyberspace
Nifty,

I would like to add that I do not believe it is the case with ALL Casinos. I brought up this subject a while ago and Pat Harrison from 32Red was kind enough to clarify this without me even asking for it.

I quote Mr. Harrison:

"eCOGRA visit our premises on an annual basis and they have full and unrestricted access to any data that they need so as to assess our performance against their eGAP standards.

Our financial accounts, reports and overall business are also audited on an annual basis (London stock Exchange requirements) by an appointed audit business. Again, this is on site and they also have unfettered access to all relevant data"

The question I have is - Can ALL software providers claim the same.. or is it a case of submitting raw data (who knows if they choose what they send) to the auditors for analysis?

Nate
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
spiderlegz - I don't know what "facepalm" means.

Also, see below.

Nifty29 - there are 2 different types of audits that are included in the previous discussion.

The first is the typical Monthly Audit, which usually results in a report certifying RTP values for various game groups.

The second is a System Audit. This is more in line with what eCOGRA does, and what agencies like the Isle of Man Gambling Supervision Commission, the Alderney Gambling Control Commission, and the UK Gambling Commission are putting in place.

Both your question and spiderlegz question deal with this first type of audit, the Monthly RTP Report.

So, I will give you what I've got there. Please Note: This is what I've got, what Galewind does. I don't have a clue what other people/companies do (and it is none of my business either).

First of all, let me preface all of what follows with this: I am absolutely, unshakably committed to ensuring that the Player, and the Casino, receive not one penny more, and not one penny less, than what the game's Theoretical (Maximum) RTP provides. Not one penny more, not one penny less.


On the 3rd of each month, at about 6AM, an application "wakes up" on the Casino's database server. It starts executing what are called "queries" into the Casino database to get game data.

For example, the query for Keno for the month of August 2011 looks like this:

SELECT createdate, playeridno, betamount, gameresult, pickcount, catchcount, random1, random2, random3, random4, random5, random6, random7, random8, random9, random10, random11, random12, random13, random14, random15, random16, random17, random18, random19, random20, pick1, pick2, pick3, pick4, pick5, pick6, pick7, pick8, pick9, pick10 FROM kenogames WHERE createdate BETWEEN 40756 AND 40786.9999999999 ORDER BY createdate ASC

This query pulls all of the Keno game for August out of the database. The application then writes this data to a text file using commas to separate all of the game fields, and line breaks to separate all of the game rows.

The application creates one of these text files for each of the games in the Casino, and then goes back to sleep for another month.

Galewind then ZIPs up all of the files into a big pile, and FTPs the pile over to Certified Fair Gambling (CFG).

CFG unZIPs the pile, and starts running pre-defined "analysis packages" against each of the game files.

I can't speak to the details of exactly what these analysis packages do. To ensure integrity of the analysis, CFG doesn't want me to know, so that I can't manipulate the data to pass their tests. However, I do know that, for Keno as an example, the analysis include at a minimum a statistical analysis of the ball distribution across all of the 20 random fields, and across all 10 of the pick fields.

I can also speak to the professional working relationship that I have developed with CFG. When CFG and Galewind first started working together back in early 2009, we spent a lot of time running tests, dumping data into Excel worksheets, analyzing the results, talking about other tests, different tests we might run to check for various aspects of bias and correlation: step tests, group tests, run tests.

Suffice it to say, we just did a shit load of tests, and generated a shit load of Excel worksheets, to set everything up at the beginning.

Certified Fair Gambling Details

(In this linked post you will note the statement "CFG ran its standard suite of tests, as well as some additional tests by specific request of Galewind Software.")

Also as part of this analysis, CFG sums the bets for a game group, sums the win/loss, and calculates the game group's Reported RTP.

Finally, they issue a PDF report which Galewind publishes.

CFG brings to my attention anything which requires further analysis. Technically, this is usually a result with what is known as a "p-value" that is less than 0.05, or a result which falls outside of a 3-sigma (3 Standard Deviations) expectation.

Every time this has happened, the specific problem "factors out" when a larger sample size is analyzed.

In addition, CFG has combined these monthly files from Galewind into:

- a 6 month analysis.
- a 12 month analysis.
- an 18 month analysis.
- a 24 month analysis.

So, for Galewind and Certified Fair Gambling, this has been a LOT of work, and continues to be a lot of work. There is no rubber stamping going on here.
 

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
spiderlegz - I don't know what "facepalm" means.

Facepalm means this:

You do not have permission to view link Log in or register now.


Spiderlegz was rightly facepalming the fact that if Regulatory Agencies and Certification Laboratories don't even understand the basic difference between house edge and element of risk (like you said), how in the world can they ever be given the responsibility of doing audits if they cannot distuingish a crucial factor like this? The casino could easily cheat 2% from every wager and it would display completely expected RTP in the audit because of this oversight.

First of all, let me preface all of what follows with this: I am absolutely, unshakably committed to ensuring that the Player, and the Casino, receive not one penny more, and not one penny less, than what the game's Theoretical (Maximum) RTP provides. Not one penny more, not one penny less.

Who defines what the game's Thoretical (Maximum) RTP is if the game involves strategy decisions? Does the RTP realize itself from the game's rules and the laws of probabilities (the correct way), or is it a manually set in stone value (the cheating way)? What if the player discovers some way to get a higher RTP than you calculated yourself, by for example finding out something your overlooked? Would he obtain this higher RTP or would your game start to cheat in order to put him back to the RTP you set?

Ladbrokes customer support agent claimed to me once that their Blackjack games are set to deliver 97.28% return, no matter how well the player plays. Well, we all know that with basic strategy the payout is at least 2% higher than that, so I don't know if the support was just clueless or if they are in fact cheating.

[...]

CFG brings to my attention anything which requires further analysis. Technically, this is usually a result with what is known as a "p-value" that is less than 0.05, or a result which falls outside of a 3-sigma (3 Standard Deviations) expectation.

Every time this has happened, the specific problem "factors out" when a larger sample size is analyzed.

In addition, CFG has combined these monthly files from Galewind into:

- a 6 month analysis.
- a 12 month analysis.
- an 18 month analysis.
- a 24 month analysis.

So, for Galewind and Certified Fair Gambling, this has been a LOT of work, and continues to be a lot of work. There is no rubber stamping going on here.

Based on the details given it does seem that you give a high priority to audits and do them rigorously. The concern is that we don't know at what level of rigour other casino software providers do these audits.
 

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
Another problem with those agencies etc is that they really dont prove if a game is random in their monthly audits.
A game with a return of ~99,5% (HA being ~0,5%) over a large sample doesnt really prove its random. Who knows what kind of streaks have occurred?

Yes, think about a game where you bankroll always goes like this:

Expired Image

You would reach the expected 99.5% return infinite number of times and couldn't complain about the returns but the rigged downward variance would mean that you could never get ahead from the session's starting balance by playing this game.
 

Nifty29

Dormant account
Joined
Jun 20, 2001
Location
Turn right, then right. then right again
Yes, think about a game where you bankroll always goes like this:

Expired Image

You would reach the expected 99.5% return infinite number of times and couldn't complain about the returns but the rigged downward variance would mean that you could never get ahead from the session's starting balance by playing this game.

Thanks Jufo. Interesting stuff.

What is rigged downward variance? It's a new one on me.
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
I would feel comfortable if, at this point, we see where we stand relative to where we were.

Re: "facepalm". Ah. I'd have to agree with that one. Another response might be "You're freakin' kiddin' me?", and said, not with a snarl, but rather a laugh. That is, ultimately Picard removes his hand from his face and gets back to the business of dealing with it. (If he didn't, then each episode would just be the crew sitting around the Ready Room, drinking coffee, and bitching about the damn Borg.)

Re: "Another problem with those agencies etc is that they really dont prove if a game is random in their monthly audits. A game with a return of ~99,5% (HA being ~0,5%) over a large sample doesnt really prove its random. Who knows what kind of streaks have occurred?"

I think we've seen that Galewind's Monthly RTP Audit is more concerned with the statistical analysis of the game results (to prove that the games are random) than with the calculation of the Financial RTP. (I use the expression Financial RTP here to differential it from the importance of RTP as a factor in the statistical analysis. That is, RTP as a game performance value is also a part of the statistics.)

I can say with accuracy that 98% of the time that Galewind and Certified Fair Gambling spend each month on these audits has to do with statistics rather than financials. (Good example: of the 36 fields of data included in our monthly Keno analysis, only 2 have anything to do with financial reporting. 32 of the remaining 34 are used solely for statistical game analysis.)

Re: "AFAIK, the auditing process used by online casinos involves the casino submitting a whole bunch of raw data, and the auditing firm signing off that it all adds up (if I'm wrong I stand corrected)."

We have seen that, for Galewind's Monthly RTP Audits, this is essentially true. It all hangs on the understanding, or definition, of "... it all adds up ..." For Galewind, and I again use Keno as an example, "all adds up" includes answering (among many others) the question: "For the first ball dropped, is the number of balls with the value of 1 (and/or 2, 3, 4, etc.) approximately equal to 1/80th of the total number of rounds?" And the second ball dropped? And the third? Etc. Etc.

Re: "It seems that a lot of players think that the auditors visit the casino offices and run software fairness tests and go through all their books with a fine toothed comb....."

I've mentioned that there are 2 different types of Audits: 1.) a Monthly RTP Audit, and 2.) a System Audit.

The comment above more accurately describes a System Audit. The first System Audit is usually excruciatingly detailed, and can span several weeks (or even months). You need to "pass" a System Audit in order to become "accredited".

You could view a passing grade for the System Audit as a B, rather than an A. That is, you don't have to do everything perfectly, but you do have to do everything to a standards-defined satisfactory level.

If your first System Audit does not result in a passing grade, then you have to (essentially) study some more and re-take the test. You do this until you pass, at which point you become accredited.

The eCOGRA and financial audits to which Nate referred are further examples of System Audits.

As mentioned above, the first System Audit is extensive, and covers the full scope of Casino operations. Typically, there are follow up System Audits once a year. These once-a-year System Audits are less extreme than the first, qualifying, System Audit.

Please Note: The topic of this thread, and the point of greatest interest to me, is concerned with Return to Player Percentages. So, although I have given a nod to the issue of System Audits, I'm only going to focus on questions/answers/discussion/debate as it relates to Monthly RTP Audits in any further posts.

OK, that I think brings us up to speed. My next post will address the more recent points of discussion.

Chris
 

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
Thanks Jufo. Interesting stuff.

What is rigged downward variance? It's a new one on me.

It's not an official term. I meant it to describe an imagined situation where the game would force the player to lose at the start for a short while, then switch the results to normal. The player would observe normal return percentages, because most of the time the results would be fair, but he would be disadvantaged by the losses at the start, making it unlikely to never get ahead. This is what I meant by "rigged downward variance".

I wish to point out that I am not in way saying that something like this exists or is even likely to exist. This was just a pure speculation of the various ways how the results could be tampered while the audits might still show that everything is okay. Therefore the audits would need to include a much more detailed statistical analysis than just checking that the game's payout percentage is ok.
 

jetset

RIP Brian
Joined
Feb 22, 2001
Location
Earth
If memory serves me well, I seem to recall that eCOGRA also operates an ongoing outcomes-based checking system similar to that described above (called the Total Gaming Transactions Review or TGTR I think). That's in addition to the accreditation inspections and annual system reviews it requires its operators to undergo.
 

binary128

Dormant account
Joined
May 18, 2009
Location
Vancouver, BC
To address the latest posts:

Re: "... the fact that if Regulatory Agencies and Certification Laboratories don't even understand ..."

If this were true, then the problem would be severe. What data has been offered to suggest, or conclude, that this is true?

"... a conversation with one Certification Lab in which they informed me that Theoretical RTP for Stud Poker is (100 - House Edge)."

So, a quantity of 1 Certification Lab is the data in hand.

My response to a sample size of 1 was: "I anticipate the need, at some point in the future, ..." That is, if it rained today and washed the camp supplies into the river, I "anticipate the need" to move the camp higher up the ravine in case it rains again.

I infer from your comment, in which you included, and pluralized, both Agencies and Laboratories (that is, you expanded the data count significantly higher than 1), that since it rained today it will certainly rain tomorrow, as well as every other day for the foreseeable future.

Please note: I used the phrase "I infer" rather than "You imply". I'm not a mind reader; I cannot speak to what, if anything, you imply. I can, however, accurately speak to what I infer, and from that speculate (I'm allowed to speculate) as to what others may infer.

However, I think we would all agree: A data point of one doesn’t support a conclusion, but rather reinforces the need for further data. To expand on my own analogy - it calls for a "prepare for battle" rather than a "send in the troops" response.


Re: "The casino could easily cheat 2% from every wager and it would display completely expected RTP in the audit because of this oversight."

This would depend completely on:

1. How the value of Theoretical (Maximum) RTP was calculated.
2. How the analysis of the Reported RTP for these games was performed.

If Theoretical (Maximum) RTP was defined as (100 - House Edge), AND

- the Reported RTP was calculated as "expected win/loss relative to total bet", then you would be correct.
- the Reported RTP was calculated as "expected win/loss relative to initial bet only", then you would be wrong.

In the original article, and throughout this thread, I have put forward that:

1. the Reported RTP is the critical value for a Casino to use in communicating with a Player.

2. the Reported RTP for all games should use the same simple formula - (Total Money Bet / Total Money Returned).

3. in order for there to be agreement between Reported RTP and Theoretical (Maximum) RTP for games in which the total bet does not equal the initial bet, their Theoretical (Maximum) RTP needs to be defined, not as (100 - House Edge), but rather as (100 - Element of Risk).

In this way we (Galewind? Software suppliers? The Casino industry?) achieve our stated objectives of providing to the Players an accurate "apples to apples" game comparison, and also an accurate answer to their question: "How much money can I expect to go into/come out of my pocket?"

As I mentioned in the first post of this thread, several people at Galewind worked on and off for several months creating the document for which I was, and am, asking feedback. Specific words were chosen with care, sentences were created and edited with precision, and paragraphs were laid out with attention to presenting a logical data flow. Later points were made working with previous declarations. The document slowly builds to its conclusion. (Well, to be accurate, this is what I/we tried really hard to do.)

Allow me to expand on one of my previous statements: "I am absolutely, unshakably committed to ensuring that the Player, and the Casino, receive not one penny more, and not one penny less, than what the game's Theoretical (Maximum) RTP provides. Not one penny more, not one penny less."

Our reference article goes to great lengths to define (not describe, but define) how we calculate our Reported RTP for each game. It continues with a detailed definition of Theoretical (Maximum) RTP, and publishes precise values for each of our games. What my statement above refers to is my personal commitment to the situation of:

If a Player were to sit down at one of my Pontoon tables, and were to play 1 million "statistically perfect" games, then the result should be/must be/will be that they did not lose 1 penny less than 0.32% of their total bet.

And before you shred me on that, I understand and acknowledge that, given variance, they may indeed lose more than 0.32% of their total bet. Given variance, they may lose less than 0.32%. But I think you get my point.

This speaks to something that I said in a thread that was branched off this one:

"Both Galewind and my current deployment are perfectly content with the House Edge for a profit. There are no "Robots" exclusions, or "Pattern Bettor" exclusions. Players are not blocked if they are on a winning streak, regardless of the extent of that streak (and there have been some real doozies). The "marketing strategy" is very simple: high returns. In short, it is exactly what it appears to be: if you win, you win, and you get paid."

This post has already gotten too long. I have some further comments on your previous post, but I'll put those into yet another post.

Chris
 
Last edited:

Jufo

Three-toed sloth
Joined
Sep 21, 2007
Location
Finland
I recall that when a player discovered that the double up game at English Harbour was not functioning correctly (ie. RTP < 100%) there had been an audit there at that time but it hadn't picked on the malfunctioning game. So it required a smart player to pick up what the audit failed to do. What are your thoughts about this binary128? How likely is it that the audit could miss a malfunctioning (either accidentally or deliberately) game?
 
Top