Fascinating But a Red Herring I Presume
One thing that gets generated is a graph in statistics/bankroll/hyperdrive that shows your bankroll going up and down. It includes color coding for trends.
It's fascinating to see how the 120 roll average forms a Sine Wave. This to a player suggests he can use this information to change his betting patterns. I guess the problem is you can't predict the amplitude or the period of the sine wave, even though you get the impression you could. Fascinated me though.
Posted is one generated with pass line bets taking 3-4-5 odds.
One question I have on WinCraps. I run either 300 roll simulations, or to get a full graph, I will run 525 rolls. If you check the roll totals at the end, they are ALWAYS, really close to expected for each number. Is this really possible? Are dice truly so random as to follow the mathematical expectations even in short roll durations?
It seems to me that the distribution of numbers is a bit too close to expected all the time.
Quote: RaleighCrapsI will run 525 rolls. If you check the roll totals at the end, they are ALWAYS, really close to expected for each number.
you should have variance in your ending bankroll, etc, but the distribution of "how many sevens" etc might be pretty similar.
you want to seed the RNG with every new trial, I assume you know that. The function can be set to auto-seed, but I enjoy seeding it myself with keyboard smashing [so to speak]. That's all in configure/misc.
I'm no whiz at Wincraps, i just now figured out you can get 100,000 results in a few seconds using the hyperdrive [accidental discovery]
Quote: odiousgambitQuote: RaleighCrapsI will run 525 rolls. If you check the roll totals at the end, they are ALWAYS, really close to expected for each number.
you should have variance in your ending bankroll, etc, but the distribution of "how many sevens" etc might be pretty similar.
you want to seed the RNG with every new trial, I assume you know that. The function can be set to auto-seed, but I enjoy seeding it myself with keyboard smashing [so to speak]. That's all in configure/misc.
I'm no whiz at Wincraps, i just now figured out you can get 100,000 results in a few seconds using the hyperdrive [accidental discovery]
RTFM! >:-)
You need to be careful about re-seeding new games. It is so fast that, if your "games" are short, the system clock hasn't advanced and you get the exact same series of rolls. The author explains all this in the Help. I urge both of you to read, read the f----- manual.
Cheers,
Alan Shank
Quote: RaleighCrapsYeah, I had noticed the same thing. It is eerie how similar they look to my day trading stock charts. Add in a moving average line, and the fast stochastic indicators and you would have a trigger point for decision making. LOL . Shows how dumb we can be in day trading, to rely solely on indicators.
One question I have on WinCraps. I run either 300 roll simulations, or to get a full graph, I will run 525 rolls. If you check the roll totals at the end, they are ALWAYS, really close to expected for each number. Is this really possible? Are dice truly so random as to follow the mathematical expectations even in short roll durations?
It seems to me that the distribution of numbers is a bit too close to expected all the time.
Are you talking about ONE 300-roll simulation (or 525) or many games of 300/525 rolls?
I just threw together a little auto-bet file that makes $5 pass bets and stops after 500 rolls, resolving the last bet. It stopped after 504 rolls.
I examined the "Roll Totals" screen. The graph has the same general shape as the expected graph, but there are certainly differences. It doesn't take long for the percentages to "take hold".
You know, you can figure the standard deviation for any outcome for any number of trials. The formula is:
(p(outcome) * (1 - p(outcome) * n)^.5
IOW, the square root of the product of the probability of a given outcome times its complement (1 - p) times the number of trials. Here's an example from my little experiment:
number of rolls: 504
expected number of sevens = .1667 * 504 = 84
actual number of sevens: 88
SD = (.1667 * .8333 * 504)^.5 = 8.367
So, if we did this experiment many times, we would expected most of the time to get between 76 and 92 sevens, only about 5% of the time to get more than 100 or fewer than 67.
I have examined the roll results from WinCraps, the bet-outcome results, etc. etc. and they exhibit the kind of variance that we expect.
You can write your own auto-bet file. The bets are irrelevant; just record the number of sevens in a chip stack for each 500-roll "game", then look at that chip stack in the Games Log. You can see a graph of the distribution, you can see how many games had values in different ranges, etc.etc. Don't take my word for it, check it out!
Cheers,
Alan Shank
I have my current autobet file that waits for a point to be established, and then buys the 4,5,9,10 for $90 each, and Places the 6,8 for $90 each.
If one of the numbers pays on the next roll, chip stack 3 increments one, and my bets all come down.
If a 7 out occurs, chip stack 4 increments one.
If a 2,3,11,or 12 rolls, the bets stay put until there is a win or loss.
I start with 2k bankroll.
In autoplay, I set the roll count to 525.
I then start a new game, then open and run hyperdrive.
It runs 525 rolls, and stops. Hit the Okay.
Now I can see my chip stack totals, and I see how many bets I won (cs#3) and lost (cs#4).
I can see where my bankroll is, and the hyperdrive graphs show me the variations that occurred.
If I still have bankroll, run the hyperdrive again. Repeat until bankroll is 0.
It is wild. Can win and lose 4k in one 525 roll session.
I know you can seed the RNG, and you will get the same roll sequence, so you can test different scheme results with the exact same rolls. That bothers me though. I know a RNG will produce the exact same result with the same seed, but that should be a one roll statement. I am hoping the answer is that when a sequence of rolls runs, that the program records all the rolls that occurred, and then if you plug in that seed again, it reads the previous rolls that were recorded, to give the same data.
OTOH, if the program somehow uses the RNG to determine every roll, then I would suggest that we are not experiencing true randomness, and instead, some algorithm is in play that gives us a distribution of numbers that closely fits the expected.
I am hoping the answer can be found in the FM that you keep telling me to read. :-)
Quote: RaleighCrapsYes, I know I need to RTFM, but I am having too much fun creating autobet files to run my hair brained schemes I keep dreaming up.
I have my current autobet file that waits for a point to be established, and then buys the 4,5,9,10 for $90 each, and Places the 6,8 for $90 each.
If one of the numbers pays on the next roll, chip stack 3 increments one, and my bets all come down.
If a 7 out occurs, chip stack 4 increments one.
If a 2,3,11,or 12 rolls, the bets stay put until there is a win or loss.
I start with 2k bankroll.
In autoplay, I set the roll count to 525.
I then start a new game, then open and run hyperdrive.
It runs 525 rolls, and stops. Hit the Okay.
Now I can see my chip stack totals, and I see how many bets I won (cs#3) and lost (cs#4).
I can see where my bankroll is, and the hyperdrive graphs show me the variations that occurred.
If I still have bankroll, run the hyperdrive again. Repeat until bankroll is 0.
It is wild. Can win and lose 4k in one 525 roll session.
Here's a suggestion: put the roll limit in your auto-bet file. Do not use auto-play. I sent you a whole bunch of auto-bet files that run 10,000 sessions of 200 rolls, always resolving the last bet, so they come out averaging 202 rolls or s.t. Make sure you start out with an empty Games Log file. It will take a while to run 10,000 games of 525+ rolls; go have a cup of coffee. Then, you can see the distribution of results of all the sessions.
Quote: RaleighCrapsI know you can seed the RNG, and you will get the same roll sequence, so you can test different scheme results with the exact same rolls. That bothers me though. I know a RNG will produce the exact same result with the same seed, but that should be a one roll statement. I am hoping the answer is that when a sequence of rolls runs, that the program records all the rolls that occurred, and then if you plug in that seed again, it reads the previous rolls that were recorded, to give the same data.
No, no, no, the program does not record the roll sequence unless to tell it to.
Quote: RaleighCrapsOTOH, if the program somehow uses the RNG to determine every roll, then I would suggest that we are not experiencing true randomness, and instead, some algorithm is in play that gives us a distribution of numbers that closely fits the expected.
I am hoping the answer can be found in the FM that you keep telling me to read. :-)
Yes, that's exactly it. Every RNG uses some algorithm to generate its results, and starts with a seed number. There are lots of different algorithms. For a given algorithm and seed, the exact same sequence of numbers will result. I suggest you read the WinCraps Help on the RNG. Also, go to Wikipedia and look up "random number generators". The results are called "pseudo-random numbers", because they are determined by the seed and algorithm, but good ones display all the traits of true random numbers, generated by natural quantum phenomena such as radioactive decay.
Many people believe that simulators keep track of the numbers generated and "force" the distribution into the expected shape, but that is totally false, at least relative to WinCraps or my own programs.
There are a couple of books available that are just actual dice rolls recorded at casinos, but they do not have enough rolls to really run a significant experiment.
Cheers,
Alan Shank
Quote: goatcabinI urge both of you to read, read the f----- manual.
Cheers,
Alan Shank
Quote: The f----- ManualA note of caution ... If you're running an Auto-Bet script which plays multiple games (i.e. contains the reset table command) and any of those games are very short
It seems this is a caution for assuming a programmed re-seed is done correctly, since it uses a clock to generate a seed. I see three conditions: auto-bet script with a reset command, multiple games, short games. Definition of short = "approximately 8 rolls"
I assume you could run into the same problem starting a new game quickly period *if* you rely on the software to re-seed. I do not. [*bristle*]
Actually, thanks for pointing out a possible source of error. RNG #3 should be used with caution also for long trials it says.
edit: thanks also for the distraction, I'm supposed to be doing my taxes. I think I might run into the alternative minimum tax this time which is *really* going to make me cuss and bristle.
Quote: goatcabin
Here's a suggestion: put the roll limit in your auto-bet file. Do not use auto-play. I sent you a whole bunch of auto-bet files that run 10,000 sessions of 200 rolls, always resolving the last bet, so they come out averaging 202 rolls or s.t. Make sure you start out with an empty Games Log file. It will take a while to run 10,000 games of 525+ rolls; go have a cup of coffee. Then, you can see the distribution of results of all the sessions.
Cheers,
Alan Shank
Alan,
I did take the code from your autobet files, so I could run the autobet sessions. Unfortunately, I struggled to understand the distribution of the results. I don't doubt that all the information I would ever want is in there, I just didn't understand it. So I have stuck with the autoplay and hyperdrive, for now. It shows me 525 rolls in 3 secs time. My wave forms are very crisp, it is fairly easy to see exactly what happened, and most importantly, I can see how many times I had point, 7 out scenarios happen 3 or 4 times in a row.
I plan on RTFM, but I think I need to bone up on SD and some other stuff first. I'm afraid I will miss a bunch of stuff as I read the manual, because I won't have any idea what it is conveying.
I want to tweak my current autobet to do Put bets of varying amounts, so the payout is the same, no matter what number is run. Once I have that running, then it is RTFM time.
that could be in the manual [oh no!] but all I see is this assertion: "The pool of numbers that each RNG (except #3) can produce is so large that it can span any number of games you could hope to play in your lifetime."
even #3 is a list a million numbers long
I doubt it qualifies as RTFM, but you will be happy to note (or perhaps not....<;-)) I read the section on HyperDrive and then looked up Standard Deviation in the help as well. I was very surprised. There is quite a bit of information in there, and it was written so I could mostly understand it. I think it is written well enough, that once I really start to study it, as opposed to just read it, I will be able to comprehend what Standard Deviation really means, and how I interpret the numbers.
If the rest of the manual is as complete as this topic, it will be worth every minute invested.
Quote: RaleighCrapsAlan,
I doubt it qualifies as RTFM, but you will be happy to note (or perhaps not....<;-)) I read the section on HyperDrive and then looked up Standard Deviation in the help as well. I was very surprised. There is quite a bit of information in there, and it was written so I could mostly understand it. I think it is written well enough, that once I really start to study it, as opposed to just read it, I will be able to comprehend what Standard Deviation really means, and how I interpret the numbers.
If the rest of the manual is as complete as this topic, it will be worth every minute invested.
The Help in WinCraps is probably better than any book about Craps I have ever read.
Cheers,
Alan Shank
