The problem with games like WinCraps

Do you believe artificial games like WinCraps can truly be used to develop and accurately test your system?

First time poster here. Non-gambler who was talked into testing a friends craps system. Spent the last 5 weeks playing WinCraps 10 solid hours per day (no exaggeration). After this intense study of the game my personal conclusion is that, although impressive, a man made piece of software like WinCraps cannot accurately reproduce the real world. Why...?

Because in WinCraps, after 5 very long weeks of 10 hour days, I never once saw 30+ rolls go by in which WinCraps did not roll a 7. Don't mean to pick on 7. Just using 7 as an example. Even though the computer program attempts to be random, by it's very nature it will also attempt to roll a certain number of 7's within a certain number of rolls to maintain the 7's overall win ratio of 16.67%. But in the real world you may experience anomalies such as a highly unusual lack of 7's or an unusual density of 7's.

I was left with the conclusion that software games like WinCraps are a great way to learn a game but if you truly need to test your system before putting real money on the line, then you might want to roll real, physical dice. I don't think there is any other way. It is possible that you may actually possess a decent system but that the software game you are testing it on is producing artificial results that may or may not happen that way in the real world.

Any thoughts welcome.

Burt

Any computer game is only as good as the random number generator(s) behind it.

How many times have you seen 50 rolls without a seven at a live craps table? I've seen 46 rolls once and never seen anything even close to that figure again. I know it happens, it's just spectacularly rare.

I don't think the computer tries to roll sevens or tries to deliver an even distribution of numbers or anything like that at all. The computer tries to generate two random numbers (most likely decimal fractions that are then multiplied & rounded to deliver two integers between 1 and 6).

I would agree that computer-based craps games are good learning tools at best and no different than any other computer solitaire game at worst.

Quote: TerribleTom

Any computer game is only as good as the random number generator(s) behind it.

How many times have you seen 50 rolls without a seven at a live craps table? I've seen 46 rolls once and never seen anything even close to that figure again. I know it happens, it's just spectacularly rare.

After thinking about it a bit more, I believe the max number of rolls without a 7 that I witnessed on WinCraps was not 50, but something like 30. I guess I was just using 50 as a generic example. Thanks.

Quote: Burt

Even though the computer program attempts to be random, by it's very nature it will also attempt to roll a certain number of 7's within a certain number of rolls to maintain the 7's overall win ratio of 16.67%.

This is not true. When the numbers generated are generated, whatever algorithm generates them is from the system, or possibly a third-party DLL, not WinCraps, and I guarantee you it "has no idea" (of course, it has no idea of anything, but bear with me) that they'll be truncated to an integer 1-6, let alone that this will be used to simulate a crapshoot. There's no way it could be set up to "catch up" on the distribution of wins, and no reason they'd want it to be.

Quote: 24Bingo

This is not true. When the numbers generated are generated, whatever algorithm generates them is from the system, or possibly a third-party DLL, not WinCraps, and I guarantee you it "has no idea" (of course, it has no idea of anything, but bear with me) that they'll be truncated to an integer 1-6, let alone that this will be used to simulate a crapshoot. There's no way it could be set up to "catch up" on the distribution of wins, and no reason they'd want it to be.

I understand what you are saying, but this is not consistent with what I witnessed after playing WinCraps for over 300 hours. I was left with the impression that the software cannot be trusted and does not simulate real life. I'm not saying that I am correct. Only saying that I am doubtful. Thanks.

Quote: Burt

I understand what you are saying, but this is not consistent with what I witnessed after playing WinCraps for over 300 hours. I was left with the impression that the software cannot be trusted and does not simulate real life. I'm not saying that I am correct. Only saying that I am doubtful. Thanks.

How many dice rolls were there in your 300 hours of play? Did you use the Classic version were it rolls the dice for you using the neat auto-roll dice feature?
at 60 rolls per hour that would be 18k
at 120 per hour = 36k

The probability of seeing a run of NO 7s of at least 50 in 18k rolls = 0.280466917983084
0.719533082 = prob of NOT seeing one or those

How often do YOU think that event should happen and HOW did you arrive at that conclusion?

The probability of seeing a run of NO 7s of at least 50 in 36k rolls = 0.482689777574320
0.517310222 = prob of NOT seeing one or those

In the Zumma Craps system tester of 35,097 actual casino dice rolls
it did not get past 50 rolls once.
41 rolls was the longest without a 7
40 happened twice

nice first post
I say WinCraps RNG#1 at least can be trusted as it uses the best random number generator available

Sally

GO
UK!
forgot the photo
roll frequencies for Zumma (includes the 7)

Quote: mustangsally

How often do YOU think that event should happen and HOW did you arrive at that conclusion?

Sally. I am editing my first post to read 30 not 50. I was typing fast and made a mistake. If I remember correctly I think around 35 was the most rolls I witnessed with no 7.

I was using WinCraps Pro and I rolled manually. How many rolls? About 1 per second. Easily hundreds of thousands of rolls. My wrist and hand are killing me. 10 hours x 7 days x 5 weeks = many, many rolls. It was very intense. My friend said he has never seen anyone so focused. Several times I stayed up all night.

My conclusion is that if you truly desire a valid test of your system, you need to roll real dice. There is no other way. So many times I witnessed something like an absence of 7's for an extended period of time and I would say to myself, "This software is now going to generate a series of 7's to catch up." And sure as rain, out trotted a string of 7's. I saw this over and over again. It got to the point of being rather ridiculous. I was using RNG #1.

Quote: Burt

I understand what you are saying, but this is not consistent with what I witnessed after playing WinCraps for over 300 hours.

Oh that's classic. Reject the most simple explanation with some subjective "feel" about perceived randomness. So tell us, how many of your runs (for any length) do you expect for your 300 hours? And what's the number of runs would make you suspicious ?

In short words: You know nothing about randomness. Any traditional pseudo random number generator is likely more random than your worn-out dices.

Quote: Burt

Sally. I am editing my first post to read 30 not 50. I was typing fast and made a mistake. If I remember correctly I think around 35 was the most rolls I witnessed with no 7.

the average number of runs of at least 35 in 35,097 rolls (the Zumma actual dice roll collection) = 9.895007685
just about there, of course that is just an average.

Did you record your dice rolls (WinCraps can do that for you too) or keep track of the rng seed # so your testing can be duplicated?

WinCraps is so easy to program and use and auto roll the dice, IMO, even my 11 year old cousin can even do it perfectly.
It can run millions of dice rolls without changing the rng seed # too without having to do it manually

Quote: Burt

I was using WinCraps Pro and I rolled manually. How many rolls? About 1 per second. Easily hundreds of thousands of rolls. My wrist and hand are killing me. 10 hours x 7 days x 5 weeks = many, many rolls. It was very intense. My friend said he has never seen anyone so focused. Several times I stayed up all night.

that is very difficult to believe.
But any one can believe what they want.
350 hours * 3600 = 1,260,000 dice rolls you did all manually.
not believable at all

Quote: Burt

My conclusion is that if you truly desire a valid test of your system, you need to roll real dice. There is no other way.

then one needs to roll real dice in a real casino. no other dice rolls should count.
is it real or fake like seeing 38DDs

That is why actual dice roll collections are so valuable to those that do not at all trust computer dice rolls.
you should do your testing then against actual casino dice rolls that have been verified and collected.
There are only a few of those collections I think.

You should contact ShuffleMaster/Ballys about getting the dice rolls that were recorded in their Rapid Craps table for about 3 years. That must be millions of them.

Have fun
Sally

Quote: MangoJ

Oh that's classic. Reject the most simple explanation with some subjective "feel" about perceived randomness. So tell us, how many of your runs (for any length) do you expect for your 300 hours? And what's the number of runs would make you suspicious ?

In short words: You know nothing about randomness. Any traditional pseudo random number generator is likely more random than your worn-out dices.

Pretty intense hate there pal. I'm just sharing what I witnessed.

Quote: Burt

Pretty intense hate there pal. I'm just sharing what I witnessed.

Thank you for sharing.
the bottom line is the final results from testing the system you tested.

you may be surprised that actual casino dice rolls and computer generated dice rolls will still show a very close expected value and variance for the same length of play.
is the system you tested a secret?

I would not think it could be something so different or special from the thousands of other craps systems that have already been tested
over the last 100 years or so.

coming next Knockout Craps, Craps Blast and the newest
The Hunter System for Craps
COPYRIGHT © 2013 Silverthorne Publications, Inc.

Sally

Quote: mustangsally

350 hours * 3600 = 1,260,000 dice rolls you did all manually.
not believable at all

Sally

Sally, I'm telling the truth. A relative showed up on my doorstep 5 weeks ago jobless, and penniless. He is a gambling addict claiming to have perfected a craps system. I gave him a couch to sleep on and set out to prove him wrong so that he would hopefully give up and get a job. He taught me how to play craps in a few days and I spent literally weeks manually testing his system and eventually systems that I dreamed up. I was highly motivated because I felt the need to prove to my relative that he cannot win at craps in the long run. I am in no way exaggerating. I did in fact play WinCraps nearly non stop for 5 weeks. Many 10 hour days. Many all night sessions.

Thanks for the advice about real casino rolls. I will look into that. If I choose to continue studying craps I will mostly likely purchase a full sized legal craps table, put it in my garage and count actual dice rolls. In my mind it's the only way to know for sure.

Quote: Burt

Sally, I'm telling the truth.

still I find it hard to believe without any data to see.
what we may remember is not always what did happen
almost one roll per second was exactly how many rolls per hour? how can that be calculated without a very large error?

you may already know WinCraps can record into one file or many files every roll it makes.

I would have thought you might record the rolls to replay them if needed.
Why not go to a casino, watch a table and write down the rolls in a small notebook. I have done this before.
a good start to your own collection.

After getting back to your computer you can then enter the rolls into WinCraps.
I use Excel to produce the proper formatting for this and just import the data but I think any text file could do this also
or just enter the rolls manually too.

But if your mind is set against anything but actual dice rolls on a real craps table
that places you into a corner that only you can get out of

still have fun

were you successful in convincing your relative to not play into the long run?
Sally

Quote: mustangsally

still I find it hard to believe without any data to see.

were you successful in convincing your relative to not play into the long run?

While playing the game I became aware of the ability to record rolls and auto-betting but I did not stop playing long enough to dive down those potentially time consuming rabbit holes.

I was only able to convince my relative that he cannot play his system without eventually martingaling beyond his bank roll or the table limit. He has adjusted his views only so far as to believe he can still become wealthy if he wins X out of 10 games.

Can I ask you a question Sally? You seem to be very experienced. Can you tell me how much money you have made overall in all the time you have played craps?

Quote: Burt

I was only able to convince my relative that he cannot play his system without eventually martingaling beyond his bank roll or the table limit. He has adjusted his views only so far as to believe he can still become wealthy if he wins X out of 10 games.

He is actually correct to a certain degree. Players can and do win with a Marty or any type of betting progression.
The chances of continuing to win while keep playing gets smaller and smaller and smaller
Most Martys can double a bankroll about 40% chance for the very 1st attempt. It is not difficult to then see some one could do it 3 or 4 or 5 times in a row before a large loss.
They would consider themselves a winner with their system over the casino due to the system, money management and discipline and not just to plain old good luck.

Quote: Burt

Can I ask you a question Sally?

yes

Quote: Burt

You seem to be very experienced.

yes I am very, very experienced in many things for being only 24 years young.

Quote: Burt

Can you tell me how much money you have made overall in all the time you have played craps?

I would say after the last and only 5 point Firebet I won, $5 paid $1250 total including my bet back that I happily tipped the dealers with, I am up lifetime at Craps by about $700

I play more Roulette, Baccarat and Video Poker than any of the other games.
I feel this could be leading some where.
I am married too and do not yet have any kids.
Practicing on that part.
fun

Sally

Quote: Burt

While playing the game I became aware of the ability to record rolls and auto-betting but I did not stop playing long enough to dive down those potentially time consuming rabbit holes.

You clicked your mouse over 1 million times in 5 weeks to watch simulated dice roll, and didn't have time to figure out that the software could just do this for you and record the results?

*speechless*

Quote: dwheatley

You clicked your mouse over 1 million times in 5 weeks to watch simulated dice roll, and didn't have time to figure out that the software could just do this for you and record the results?

*speechless*

Isn't that the whole point of the program? I mean...that's what Wincraps DOES...

Quote: rdw4potus

Isn't that the whole point of the program? I mean...that's what Wincraps DOES...

He could have done it in half the time & gotten a full night's sleep, if he'd used WinCraps's automatic features. Spend 2, 3 days learning & testing out WC & then maybe a day running his program & a day analyzing results. Oh, I see. He was in a hurry, didn't want to waste time learning something that would have saved him time.

Quote: dwheatley

You clicked your mouse over 1 million times in 5 weeks to watch simulated dice roll, and didn't have time to figure out that the software could just do this for you and record the results?

*speechless*

I never intended to spend 5 weeks of my life playing craps. It just morphed into something larger and more intense than I imagined. The time went by pretty quickly.

Betting systems don't work. All bets have a have a house edge. You can't add up a bunch of negative numbers to get a positive number.

So instead of teaching your friend some fundamental properties of negative EV games, or learning some simulation or math skills, you became addicted to (Win)Craps instead.

It seems obvious, the next step would be to go into your frieds favourite casino, blast through your own bank savings to finally say to your friend: "see, I told you so".

Quote: Burt

I never once saw 30+ rolls go by in which WinCraps ...[could be anything] ... But in the real world you may experience ...

This kind of thinking is what's leading you down the wrong road. Random events just won't cooperate with preconceived notions. Often in that "real world" the conclusion such thinking leads to is that the game is rigged. So, the computer cheats, or the casino cheats, but it just can't be true that betting systems don't work, because there are so many millionaires out there that made their money that way that we can point to.

If you take anything away from this site, hopefully it's that, as other posters have commented, no betting system works in the long run.

It is pretty amazing how insulting people have been to me on this thread. A few of you should be ashamed of yourselves.

The point of my thread is simply that I am doubtful a human made piece of software like WinCraps is capable of reproducing natural events. Every time I played WinCraps and the 7 did not come up in a long time, sure as rain all those 7's came later in bunch as the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out. Testing with WinCraps is not something I would attempt if I decided to become a serious player. Instead I would invest $4,000 US in a professional craps table and record thousands of real dice rolls while testing my systems.

To those of you who have been rude to me on this thread I can only say that I feel sorry for you. You seem like very unhappy people.

Quote: Burt

You seem like very unhappy people.

Lotsa Bigots too!

In their/my gruff, blunt way we are really trying to help you.

Quote: Burt

The point of my thread is simply that I am doubtful a human made piece of software like WinCraps is capable of reproducing natural events.

The very first people to invent such things questioned it. That concern really has been laid to rest by now, unless you believe scientists and engineers have deceived themselves..

Quote: Burt

the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out.

it doesn't work like that, unless you want to say that real dice work like that. The past does not matter in the sense that the past would be measured for accuracy.

Quote: odiousgambit

it doesn't work like that, unless you want to say that real dice work like that. The past does not matter in the sense that the past would be measured for accuracy.

A flawed human programmer designs a flawed algorithm that's goal is to roll 7 16.67% of the time. Within that context any number of dice combinations can be rolled. But in the end the software must eventually adhere to the 16.67% rule otherwise it would not mimic 'real life' and everyone would complain. So by its very nature it is doubtful that the algorithm is random.

Quote: Burt

A flawed human programmer designs a flawed algorithm that's goal is to roll 7 16.67% of the time. Within that context any number of dice combinations can be rolled. But in the end the software must eventually adhere to the 16.67% rule otherwise it would not mimic 'real life' and everyone would complain. So by its very nature it is doubtful that the algorithm is random.

oh, come on, flawed human programmer? How do you know that?

Quote:

the software must eventually adhere to the 16.67% rule

no, again, the past is not measured. The parameters are 1/6 for each die face, same as a real die, that's it.

One thing I don't understand is why you don't accept the Wincraps results? Aren't you trying to convince this person that his system won't work in the long run?

Quote: Burt

Every time I played WinCraps and the 7 did not come up in a long time, sure as rain all those 7's came later in bunch as the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out.

That's just now how it works. But, for fun. If it was designed to "spit out" 16.67% 7s, how did it get behind and need to "play catch up?"

Quote: Burt

It is pretty amazing how insulting people have been to me on this thread. A few of you should be ashamed of yourselves.

The point of my thread is simply that I am doubtful a human made piece of software like WinCraps is capable of reproducing natural events. Every time I played WinCraps and the 7 did not come up in a long time, sure as rain all those 7's came later in bunch as the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out. Testing with WinCraps is not something I would attempt if I decided to become a serious player. Instead I would invest $4,000 US in a professional craps table and record thousands of real dice rolls while testing my systems.

To those of you who have been rude to me on this thread I can only say that I feel sorry for you. You seem like very unhappy people.

No design necessary. After a large enough set of dice rolls, the average number of 7's rolled will tend to 1/6. That's exactly right (law of large numbers).

Example:

You roll 100 times and get 5 7's:

7 = 5%.

You then roll 100 more times and get 17 7's.

7 = 17% for this sample, overall = 11%.

Over time, the -percentage- of 7's will tend to 1 in 6. The overall number will not get to exactly 1 in 6, and there's no need for any catch up. It's the emergent effect of random numbers.

Quote: Burt

It is pretty amazing how insulting people have been to me on this thread. A few of you should be ashamed of yourselves.

The point of my thread is simply that I am doubtful a human made piece of software like WinCraps is capable of reproducing natural events. Every time I played WinCraps and the 7 did not come up in a long time, sure as rain all those 7's came later in bunch as the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out. Testing with WinCraps is not something I would attempt if I decided to become a serious player. Instead I would invest $4,000 US in a professional craps table and record thousands of real dice rolls while testing my systems.

To those of you who have been rude to me on this thread I can only say that I feel sorry for you. You seem like very unhappy people.

It is not capable of reproducing natural events. That's one of the reasons you can select from multiple random number generators and seed them manually.

OTOH, as Sally pointed out - in the few times that people have recorded the results of natural events the results have been remarkably close to what might be predicted by statistics...

I assure you that WinCraps is not looking at prior results and trying to hit whatever number is needed to balance the distribution. There is no bit of code in the background saying "Oh, shit! It's been 24 rolls since a 7. Better roll a seven next." It simply does not work like that at all.

Go to a live craps table and record the numbers for yourself.

In any case, I assure you that whatever "system" has been devised is a sure loser.

Burt,

I did not program wincraps, so I can't say with certainty how it is programmed. However, I can tell you that what you are describing (catching up) is at least 2-3 times harder to program than just letting the dice do their thing (depending on how complicated their "catch up" is). I would be willing to bet money that wincraps uses the standard methods of RNG to get its rolls and those are, for human purposes, correct. In fact, I would guess a physical, human system would be more likely to produce biased results than wincraps.

You are welcome to buy your own table and record your own results, but I can tell you that the amount of time and effort in doing that will not help you more than a proper wincraps experiment from a mathematical standpoint. Best of luck, I hope your relative can get their gambling addiction under control. If your buying a table will help with that maybe it would be worthwhile.

Quote: TerribleTom

It is not capable of reproducing natural events.

Thank you for confirming my doubts.

Quote: Burt

Thank you for confirming my doubts.

... and you would be unable to tell the difference between two lists, one from WinCraps and one from the Craps Table.

Quote: endermike

You are welcome to buy your own table and record your own results

Thanks. Maybe I will.

Quote: Burt

The point of my thread is simply that I am doubtful a human made piece of software like WinCraps is capable of reproducing natural events.

It turns out that software is readily capable of replicating the distribution of natural events, like rolling a pair of dice or dealing a hand of cards. Not "reproducing", since software that generates dice roll numbers does not actually roll physical dice. Instead, it uses an algorithm to generate two numbers between 1 and 6, inclusive. Similarly, the algorithm inside video poker games does not actually shuffle a physical deck of cards but instead randomizes an array of numbers between 0 and 51, inclusive.

Not only is software capable of replicating the distribution of live events, it's actually required to behave that way in most regulated gaming jurisdictions. Nevada Regulation 14.040(2)(b) says:

Quote: NGCB Reg 14

14.040 Minimum standards for gaming devices. All gaming devices submitted for approval:
2. Must use a random selection process to determine the game outcome of each play of a
game. The random selection process must meet 95 percent confidence limits using a standard
chi-squared test for goodness of fit.
(b) For gaming devices that are representative of live gambling games, the mathematical
probability of a symbol or other element appearing in a game outcome must be equal to the
mathematical probability of that symbol or element occurring in the live gambling game.

http://gaming.nv.gov/modules/showdocument.aspx?documentid=2921

Bottom line: you may have doubts that software can behave equivalently to live gambling events, but the manufacturers, operators, and regulators of software-based casino games don't have similar beliefs. Presumably you don't believe that video poker games would behave any differently (mathematically) if dealt from a hand-shuffled deck of cards as opposed to using the commonplace software inside video poker machines. Why should software-based dice rolls be any different?

Quote: MathExtremist

Presumably you don't believe that video poker games would behave any differently (mathematically) if dealt from a hand-shuffled deck of cards as opposed to using the commonplace software inside video poker machines. Why should software-based dice rolls be any different?

I would never play video poker for real money.

Quote: Burt

I would never play video poker for real money.

Fair enough -- you don't trust technology. I was about to point out the closer analogy of video roulette: Bally has a roulette game that's been in Nevada for years, and there's no meaningful difference between picking a single number from a range of 38 and picking two numbers from a range of 6. It's the same underlying, approvable software. But if you don't believe that any RNG technology can reproduce the distribution of live events, you could prove it for yourself. Take Excel, type in the formula for generating a random number between 1 and 6, and do a copy-down for 100 rows. Then do 100 single dice rolls. You won't be able to tell the difference in distribution.

Quote: Burt

Thank you for confirming my doubts.

Don't draw the wrong conclusion. Just because a computer random number generator is not the same as actually passing dice around a live craps table does not mean that it is unreliable for testing a betting system. The number of times you'll see 30+ rolls at a craps table without a single seven is not large.

For a simple test of a random number generator, create a spreadsheet with a column full of random numbers.

=INT(RAND()*6)+1

Make two columns, add them up if you'd like.

Now start rolling some dice on a craps table and recording the results. Get back to us when the two lists are identical.

When you're done, enlist the help of your gambling-addicted buddy and track the numbers at two tables at the same time. Again, get back to us when the two lists are identical.

That's the thing with random numbers. They're - wait for it - kind of random.

You could track the actual rolls at every table in the world for years on end and when you put them into a database and compare them you are going to find very few sets of data that match for more than a few rolls at a time.

Nature isn't capable of reproducing natural results either. Each set of natural results is eventually unique, and most of them are unique in just one roll.

No matter what, the betting system is a guaranteed loser.

Quote: Burt

I never intended to spend 5 weeks of my life playing craps. It just morphed into something larger and more intense than I imagined. The time went by pretty quickly.

5 weeks is a long time. You must have had some fun doing it.
Time flies when having fun.
It also looks like as you mentioned you tried other systems out too.

You were disappointed at your results because of the distribution of the 7 in your sample size that was not documented.
over 100,000 dice rolls and never did a 7 go 30 rolls without showing and when it did disappear for maybe 25 rolls, here it comes trying to catch up to that 16.67% number as you say.
The first part I really doubt (never seeing the 7 go 30 rolls without showing one time as you claim with no data to show this)
but the second part is believable to me.
I mean, the 7 is the most common number to roll on each and every roll and over any number of rolls verses any other number.


Maybe, just maybe you changed, and did not see you did, the values on the probability page and used a different distribution than a normal 2d6 one that is the default.
If not, then you really should start to repeat your efforts and post your data so others can verify what you are doing.
Words are words are words, you must agree.

Maybe you did find an error in WinCraps Pro, when doing many manually rolled dice rolls.

I also think that you can not have the program auto roll the dice for you unless you pay the registration fee of about $15
In Classic, the program auto-rolls the dice without paying the reg fee but I do not know if it also does auto-betting without the fee.
The WinCraps webpage does not really say and my version came already in a folder.

Looks like Steen could make that more clear to current and future WinCraps users
Sally

I think I have heard enough. Thanks to everyone who gave constructive comments. I appreciate that fact that computer programmers are very smart people and that computers are amazing machines. My best friend is a software engineer. He is no dummy. However, I remain skeptical and if I were to peruse craps seriously I would invest in a professional craps table and roll real dice. WinCraps seemed to do some funny things that I don't think would happen the same way in real life. Maybe it was just my imperfect memory or my imagination. I respect WinCraps for the handy and fun game it is. I did not mean to pick on WinCraps specifically.

Thanks again for all the help,
Burt
Non-gambler
Software skeptic
350 hour WinCraps player

Quote: Burt

I think I have heard enough. Thanks to everyone who gave constructive comments. I appreciate that fact that computer programmers are very smart people and that computers are amazing machines. My best friend is a software engineer. He is no dummy. However, I remain skeptical and if I were to peruse craps seriously I would invest in a professional craps table and roll real dice. WinCraps seemed to do some funny things that I don't think would happen the same way in real life. Maybe it was just my imperfect memory or my imagination. I respect WinCraps for the handy and fun game it is. I did not mean to pick on WinCraps specifically.

Thanks again for all the help,
Burt
Non-gambler
Software skeptic
350 hour WinCraps player

Just remember that no matter where you go - digital or natural, casino or your own private table in the garage - your buddy's betting system is doomed.

Especially if his system relies on 30+ rolls without a seven.

Quote: TerribleTom

The number of times you'll see 30+ rolls at a craps table without a single seven is not large.

for his claimed sample size it is.
he had to do at least 100,000 rolls.
the probability of NOT seeing at least one run of 30 rolls without a 7 in 100k rolls is about 1 in 13,207,037,030,412
yes, 13 trillion

using
Streak Calculator
about 1 in 13,157,894,736,842

Quote: Burt

"Because in WinCraps, after 5 very long weeks of 10 hour days, I never once saw 30+ rolls go by in which WinCraps did not roll a 7"

the average number of such runs over 100k rolls = 70.195153
Not 0 or 1 or 2 or even 1,000
It has it's own distribution

Quote: TerribleTom

No matter what, the betting system is a guaranteed loser.

this can not be correct as currently stated.
You mean make 100 lifetime bets at Craps and everyone will be a loser. Zero chance of winning.
Bet on random rollers at Craps and you will be a loser.
That sounds like a Frank Scoblete statement.
he is famous for those as I know

added: but not as famous, IMO, as
The Hunter System for Craps
COPYRIGHT © 2013 Silverthorne Publications, Inc.
coming soon in it's own betting system thread.
"Be able to beat any online or land-based craps game so easily you’ll have trouble
believing it isn’t a dream until you realize all the winnings are yours and they are good as
gold!"

"Instead of having to learn a number of complex bets, players using this system only have
to learn two bets.
That’s it! Once they learn the Two Key Bets to Craps and then use the Hunter System to
make these bets they automatically beat the game."

Sally
stop it

Quote: mustangsally

Quote: TerribleTom

No matter what, the betting system is a guaranteed loser.

this can not be correct as currently stated.
You mean make 100 lifetime bets at Craps and everyone will be a loser. Zero chance of winning.

Over 1 bet or 100? No. Anyone can place a bet and win - if they couldn't nobody would play. Betting on black, winning and walking away is not a betting system.

It sounds like (from posts above) that the OP has a relative or friend that thinks they can Martingale themselves to glory and that is simply not going to happen.

There are countless articles and posts (on this site, WoO and many others) that demonstrate the shortcomings of betting systems. There is no way to turn a -EV game into a +EV game by betting some certain way.

You might win today and you might win tomorrow but eventually your betting system is going to lose if you keep playing.

Burt: I know you're done w/ this discussion, but I have more to say. It's commendable your concern for your relative. You haven't said anything re: his system, how he manages his bankroll, etc., only that he's addicted. There's a lot more here than perfecting some "winning" system.

I don't believe you need to go to the trouble/expense of buying a craps table to compare results with WC. All you need to do is sit down at a table with a piece of felt or similar cloth in front of you, take a pair of dice, and just roll them out of your hand for a few tumbles, with a pad and pencil by your side and record each roll.

I had played a few games on my computer, wondering the same thing you are. Later, I did as described above, and was amazed at how real dice reflected the computer game. I probably did several hundred rolls on each; but it is amazing how real dice and computer dice were similar. For example, on the computer, I saw some numbers repeating frequently. When I rolled by hand, the same thing happened, not identical numbers, but the patterns were the same. IOW, neither is predictable, according to the math. Why? Well, that's a long story for me & I won't get into it, here.

I will say this: Both computer results and live game results are the same. Does that mean you can take a system on the computer to the table and get the same result? No. The reason is that each is in its own territory, meaning that the sequence of rolls is different. You can't take the same sequence you ran on a computer to a live game. Everything is random, and randomness does not produce certainty. That's why we have DI, which tries to defeat randomness in craps, which I don't ascribe to.

So, why mess w/ WC or similar? Because they give you an idea of what to expect in comparing different systems, which is what I'm doing, right now, using the same RNG seed for each system. I run each five times w/ a different RNG seed to get an idea of the system's performance. Can I take the "best" system to a live table and expect to win consistently? No.

Heed what others are telling you and your relative: Casino craps is set up for the casino to win, after all is said and done. Does that mean you will consistently lose? No. Some players are luckier than others, and that's all it is, luck. Your relative's goal is to enjoy his wins and accept his losses.

And that takes money management and discipline, the two things your relative lacks and which both of you need to work on to help him, instead of trying to find a winning strategy. These are your two biggest issues, right now. He may have to quit gambling, if it's ruining his life. And, you better watch out, too.

Burt, you pose a question about the reliablity of WinCraps' rolls so I feel compelled to respond. I asked Michael Shackleford and he said he wouldn't mind if I comment on WinCraps here. I'm not here to post adds but to participate in the discussion.

To be fair, it seems you really were addressing all computer simulators. Nevertheless, it's a good question and it's good to be sceptical. You may not have realized that most computers use random number generators (RNG's) which do NOT produce numbers from a truly random process. Rather, they're produced from mathematical algorithms intended to simulate random numbers. So the question is, can they reasonably SIMULATE real random rolls? My experience says yes, and as MathExtremist has pointed out, the gaming industry seems to agree, though I wouldn't take that as an endorsement for any and all RNGs. As you might suspect, there are both good and bad RNG's. Better to see for yourself, right?

Since your question mentioned WinCraps, try comparing the output from WinCraps with rolls generated from a known random source such as actual recorded dice rolls or atmospheric noise such as provided at www.random.org As Sonny44 posted, it's surprising how similar computer generated rolls are to real random rolls. Choose whatever metric you like for comparison but I doubt you could tell the difference.

Burt writes:
"... after 5 very long weeks of 10 hour days, I never once saw 30+ rolls go by in which WinCraps did not roll a 7."

Of course anything's possible when it comes to random numbers but it does sound highly improbable as phrased. Are you saying that it could NEVER happen in real life? I'm guessing you either didn't know how to check it, you tried to keep track of it in your mind, or you haven't given us some critical information such as the length of your sessions. Was it one long session of 1.2 million rolls or multiple sessions of 50 rolls? How about sending me some saved game files so I can see your configuration? As MustangSally pointed out, you may have inadvertantly skewed the probability distribution. Then again, maybe WinCraps is out to lunch so let's check it. This is an easy enough metric to check.

The following screen shot shows roll frequency data for the number 7. One was obtained from a sample of 200,000 computer-generated rolls (RNG#1) and the other from a sample of 200,000 real random rolls. The way it works, a counter keeps track of how long it takes for each 7 to appear and records the results in a table. For example, if a "1" appears in blue next to the #49, then there was one time when it took 49 rolls for the 7 to appear (48 non-7's, followed by a 7). As you can see, the overall percentage of 7's in each sample is not far from 16.67%. So the first question is, which sample contains RNG rolls and which contains real random rolls? You have a 50% chance of guessing correctly so if you think you know then tell me why.



Both samples clearly show many times when the 7 did not appear for 30+ rolls, so the second question, do you think this might be an anomaly? Did I just luck-out and happen to find a computer-generated example of this or did I sit here rolling for 5 weeks to find it? Although this is a fairly rare event occuring less than 0.09% of the time (ballpark estimate), the truth is, it happens darn near every time (real or RNG) when enough data has been collected. Try it yourself. Choose RNG#1, record your seed (check Game>Configuration>Misc) and then let me know your seed if you find a sample that should contain times when the 7 doesn't appear for 30+ rolls but doesn't.

It's interesting to see some of the mail I get. Things like:

"I played for 200 rolls and saw too may 7's."
"I played 500 rolls and didn't see enough 7's."
"It always knows when I make a big bet and makes me lose."
"It rolled six 7's in a row which I've never seen in a real casino."
"It plays catch-up to make the numbers balance out."

Now, why would I possibly want users to lose? Wouldn't I be cutting my own throat to write a program that schemes against them? A proof to verify that this is not happening would be to record your rolls without betting, then replay them with bets. If it still makes you lose, then you'll either learn something about probability and expectation or I've written a piece of software that's downright clairvoyant!!

I'm not trying to be flippant. If there's a program flaw or bad RNG then I certainly want to correct it, but should every sample of 200 rolls contain 33.33 sevens? Should I do something to the RNG to ensure that? Well, does every 200 roll sample of casino-recorded rolls contain 33.33 sevens? Of course not, it varies, so shouldn't the RNG be allowed to vary too? Yes, of course and so it does. Try running a series of 200-roll sessions (or whatever length you desire) and record the number of 7's in each. Do this for computer-generated rolls and real rolls. I'll wager that you'll see a lot of variance in both sets of data and that the aggregate percentage of 7's will get closer and closer to 16.67% as the amount of data increases.

The following screen shots show two histograms of the number of 7's found in one thousand 200-roll sessions. Again, one contains computer-generated rolls and the other real random rolls. Can you tell me which one is real? Why?




The acid test for a craps RNG is to run it against a battery of betting systems. If something is biased, abnormal, or predictable then there's going to be some system that can exploit it consistently. The fact is, ANY DEVIATION whatsoever from a random distribution in a simulator or in real life can be exploited. This is particularly true with craps since you have the ability to play either side of the dice.

For example, if you think there are no streaks of 30+ rolls without a 7 then you could play a very simple system to exploit it. Try entering this one line on the auto-bet screen:

If # of rolls since 7 >= 29 Then Bet $10 on Seven EndIf

If you're right, then such a system would prove to be a consistent winner. Try it and get back to me.

The Mersenne Twister is a great and popular RNG (one of several that I use) yet it doesn't pass all mathematical tests for randomness (I'm unaware of any for the PC which do). Even so, I've not had anyone report to me that they found a system which can consistently beat it. I say "consistently" because there are times when any system will win as we should expect.

Sometimes you might feel that dice rolls seem to always oppose you. At other times it might seem that you can do no wrong. This of course is the nature of the beast. Fact is, it's somewhat hard to say for sure that something is random or not because randomness itself encompasses every possibility!

Burt writes:
"... the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out.

Nonsense. I assure you, there's nothing I've entered in the code to tell the rolls to balance out and if I attempted such a thing it would make a real mess of things. The percentage of numbers tend toward the predicted probability all by themselves by virtue of the law of large numbers.

Burt writes:
"Do you believe artificial games like WinCraps can truly be used to develop and accurately test your system?"

Frankly yes, but then it depends on your expectations. Do you believe winning at Sam's Town is any indication of the results you'll get from playing at Bellagio?

Burt writes:
"So many times I witnessed something like an absence of 7's for an extended period of time and I would say to myself, "This software is now going to generate a series of 7's to catch up." And sure as rain, out trotted a string of 7's. I saw this over and over again."

Any gambler can tell you many such stories from real life experiences. I swear, whenever I make a large Lay10 bet is the time the shooter gets hot for 10's. Count on it!

Mind you, I'm not on some crusade to glorify RNGs. It's wise to cast a sceptical eye on questionable output. You can try another RNG or if you just don't trust RNGs then there are alternate methods of generating rolls, so why argue about it? If you had doubts, I'm surprised you didn't venture to try one of these alternate methods. I'm impressed by your tenacity though -- 5 weeks of 10-hour days clicking the mouse has got to be a record.

Steen

I usually expect 7craps to comment on anything involving WinCraps, but a quick check of his profile shows that he hasn't logged on in almost three months (1-23-2014).

He has always been an active member, and I have personally found his posts (and his expertise with WinCraps) to be very helpful.

Anyway, hope all is well with him.

Quote: blount2000

I usually expect 7craps to comment ... he hasn't logged on in almost three months

definitely noticeable absence.

Steen has laid out his case so well, there's not much to say

Quote: Steen

As MustangSally pointed out, you may have inadvertantly skewed the probability distribution.

I had to find out because there really is a lot to WinCraps. It has way more features than the WoO JB version and many things in WinCraps could go unnoticed if one does not understand all of it's features as Steen does.

I checked out the skewed probability distribution and the program knows that we are using something different from standard dice.
But you still have to look for it.

I see this in both Classic and Pro

Quote: Steen

Then again, maybe WinCraps is out to lunch so let's check it. This is an easy enough metric to check.

I love going out to lunch.
for any that do not want to believe any simulation from a computer that uses a rng, there are a few collections of actual dice rolls to test systems and run sims against for craps but not as many as just flipping a coin at home.

The OP saying that he would go out and buy a real casino craps table and dice and roll the dice I still do not believe one would go that far unless money is no object.
he still has to roll or have someone roll the dice many many times.
I would think that would be way more difficult, but maybe more fun, than clicking the mouse.
I hope too that casino conditions would apply, keeping newer dice on the table than just one set and rolling both dice to hit the wall.

Thank You Steen!

Sally

Quote: Steen

Burt writes:
"... the algorithm played catch up to spit out the eventual 16.67% figure that is was designed to spit out.

Nonsense. I assure you...
Steen

Like Sally, you are very insensitive and insulting Steen. You can both rant until you are blue in the face. I am no longer listening.

And another one gone, and another one gone. Another one bites the dust.