How to calculate Pass/Dont Deviation?

Hi-

I like math but have a weak understanding of statistics. I understand the house edge and the short term possibility of a deviation over the mean to obtain a win (assuming you leave the table).

If you could document the number of PASS line wins (and conversely DON'T line wins) would it be useful? Assumptions are that it is a flat bet, no odds, and you are not concerned with changing the bet amount.

More assumptions- 7,11 are pass line wins. 2,3,12 are pass line losses but 12 is a DON'T PASS push. All I am concerned documenting is if the pass line wins or the don't pass wins.

Question-
1- how would you calculate the deviation. I understand the Standard Deviation Formula but I'm not clear how you would use that to compare Pass and Don't flat bets (no odds).

If the pass line standard deviation is 0.8 (I don't understand that calculation) and the chances of a pass line win at 49.25%. Would you calculate the PASS wins as a 1 and LOSSES as a 0 to calculate ((Xwins - .4925)^2 / total rolls)^.5 and DONT wins with the same formula?

2- When or IF would it be useful? In a 200-300 decision session would knowing the session PASS wins/losses are X outside of the mean one way or the other alter the betting strategy?

I was initially thinking of a way to determine a choppy table with the assumption that the pass/dont ratio would be near 50/50. That got me thinking about the potential value of the deviation over a ratio of wins/losses as an indicator.

Any insight would be helpful.

Thanks,

JC308

you seem to have a certain degree of education, an articulate writer as well, but you have failed to come across the first step in rising above the sea of ignorance that surrounds Craps...

which is to realize the dice have no memory

Quote: odiousgambit

you seem to have a certain degree of education, an articulate writer as well, but you have failed to come across the first step in rising above the sea of ignorance that surrounds Craps...

which is to realize the dice have no memory
link to original post

The other thing is the concept of “session”. While it has a meaning to us, that meaning exists outside of odds and statistics; we use sessions to manage our money and our time at the tables, not to manipulate the odds. In the real world, the “session” started with The Big Bang, and will end with the heat death of the universe.

Without knowing the math, I'm noticing that hot shooters are rare and cold shooters are more frequent. It's easier to come across 10 PL losers aggregate than to come across 10 PL winners aggregate. I'd lower my DP bet and add odds so the 7-11's don't hit me so hard and to lower the HA substantially. As much as I want to bet with the table, I keep getting burned on the PL and after several hours I may be down 50 PL bets. If I bet the DP I'd be far ahead.

I understand the Gambler's Fallacy.

I like numbers and tracking data. The craps table can be hot, cold, or choppy- or all of the above during a a very small Data set in which we call a session. The odds are the odds. One roll has nothing to do with the next. If you have a run of 20 wins, it wasn't luck- it was just a small deviation of the mean of a very large data set which ultimately means nothing. You might have been lucky to be there at that moment however. After all, I'm not at the craps table for several million rolls, just a small snapshot of time.

We should all know the odds statistically over millions of rolls (Mosca). I'm curious about the short snapshot of data when we are at a craps table where a deviation can occur in one direction or another (preferably in my favor).

Data, which is historical in nature, cannot predict the future. But it is interesting. As human beings we form behavior responses based on imperfect inputs which dictate actions moving forward. A childhood event in life can determine a knee jerk response, "I will never do that again", that we adhere years to come into adulthood despite the limited data used to come to that conclusion. Sometimes it is completely subconscious. You don't think about it. You're brain already knows what to do and does it. Don't touch the hot stove. All it takes is once.

As I said, I was working on an indicator for a choppy table. I know it will not forecast the future. But what does it look like from a data standpoint? We know it when we play.

I was using win/loss ratios initially and as the data set grows it obviously gets closer to the mean but a snapshot of twenty or so decisions can vary greatly. If you graph per roll and add +1 for a win and -1 for a loss with 0 as the baseline- it can be all over the place.

So I was wondering about the math for calculating a standard deviation and then you could take a data sample and look at the deviations. If the deviation is a wave (plus or minus) whats the amplitude of the deviations over a given data set? What's it look like over 20 data points, 200, 500, 1000? I would be curious to graph it and see. I know- one data sample won't look like the next. Especially considering the outcome is binary- win or lose. What would a 100 wins in a row look like as a deviation? I don't know but it'd be cool to graph.

So my next question would be- if there are wave deviations, when they do occur, whats the average amplitude? I was told the standard deviation on the pass line without odds is 0.8. I'd like to see that math.

I don't know. I don't know the math to even see for myself. It maybe a mute point and we find out that the data is irrelevant. But I am curious about the math.

You say it's only interesting to look at, like the Mona Lisa or something, but methinks a gambler... well, who cares what I think!

I do have a formula I learned from Ace2 that gives you the standard deviation

SD = SQRT [prob of a win * prob of a loss] * [amnt won including return of wager/bet amount]

so for the pass line that's sqrt [244/495*251/495] * [1+1/1] for one unit bet
= 0.9999000052035608

to check this, the wizard lists SDs at the link https://wizardofodds.com/gambling/house-edge/ ... the pass line has an SD of "1" [edited out mistaken hacker theory, see posts below]

I don't know why the formula works

So if I can only hang out for 400 rolls, what's the best way to play? Ace2 would say PL with odds and probably add a couple of come bets with odds because of the low house edge compared to all other bets on the table. I'd say I've had my best payouts with that strategy and also with a progression on the 5, 6, 8, or 9 that got out of hand with over 9 hits on one number. But either one takes a hot shooter and just when are they going to happen? I may be playing DP and scaring all the bettors away and have a hot roll that everybody missed.

Quote: odiousgambit

to check this, the wizard used to list SDs at the link https://wizardofodds.com/gambling/house-edge/ but I notice now some hacker[?] shortened this list and there is nothing alphabetically past Casino War.
link to original post

(snip) I can see the list continue right down to Wild Hold 'em Fold 'em.
Craps Pass/Come is listed with SD = 1.

Best guess is that your browser isn't displaying something right... maybe scroll inside the subwindow?

Quote: Dieter

Quote: odiousgambit

to check this, the wizard used to list SDs at the link https://wizardofodds.com/gambling/house-edge/ but I notice now some hacker[?] shortened this list and there is nothing alphabetically past Casino War.
link to original post

(snip) I can see the list continue right down to Wild Hold 'em Fold 'em.
Craps Pass/Come is listed with SD = 1.

Best guess is that your browser isn't displaying something right... maybe scroll inside the subwindow?
link to original post

you're right, I think maybe I didn't have to do that before ... I didn't expect it, and the scroll down bar doesn't appear till you mouse-over it

jumped to conclusions, and yes there's confirmation on the SD of "1". The others I have tested also, and get the correct SD with the formula

Quote: JC308

Hi-

I like math but have a weak understanding of statistics. I understand the house edge and the short term possibility of a deviation over the mean to obtain a win (assuming you leave the table).

If you could document the number of PASS line wins (and conversely DON'T line wins) would it be useful? Assumptions are that it is a flat bet, no odds, and you are not concerned with changing the bet amount.

More assumptions- 7,11 are pass line wins. 2,3,12 are pass line losses but 12 is a DON'T PASS push. All I am concerned documenting is if the pass line wins or the don't pass wins.

Question-
1- how would you calculate the deviation. I understand the Standard Deviation Formula but I'm not clear how you would use that to compare Pass and Don't flat bets (no odds).

If the pass line standard deviation is 0.8 (I don't understand that calculation) and the chances of a pass line win at 49.25%. Would you calculate the PASS wins as a 1 and LOSSES as a 0 to calculate ((Xwins - .4925)^2 / total rolls)^.5 and DONT wins with the same formula?

2- When or IF would it be useful? In a 200-300 decision session would knowing the session PASS wins/losses are X outside of the mean one way or the other alter the betting strategy?

I was initially thinking of a way to determine a choppy table with the assumption that the pass/dont ratio would be near 50/50. That got me thinking about the potential value of the deviation over a ratio of wins/losses as an indicator.

Any insight would be helpful.

Thanks,

JC308
link to original post

Yes, a random sequence has frequency components. You can extract the amplitude of these components from a finite series of trials using a discrete cosine transform (DCT) https://en.wikipedia.org/wiki/Discrete_cosine_transform discrete cosine transform.

Extracting the frequency components from a past series of random events has no predictive power. The next series of events will have a DCT whose amplitudes will tend to be similar, but the phases will be completely random. Random phases means no predictive power. Random processes generate 'noise' with a spectrum characteristic of the random process. This is why humans can find patterns in random outcomes.

If you are trying to roll a 6 on one die, you can find the 'wave' with a wavelength of 6 using a DCT of a large number of trials. How does this help you to know that the amplitude of the DCT wave will be similar in the next 100 spins, but you do not know anything about the phase. A 6 is still equally likely on any of the next six rolls.

Mental- very helpful. Thank You.

My original intent here was to try to identify the characteristics of a choppy table during a "session" sized data sample which has led me to deviation and now phase and frequency. Straying into standard deviation, I was curious if the deviation over a session sample size remained within the standard deviation would that be a good indicator? Would it be better than just using win/loss ratios- which over a larger sample size just become the mean and doesn't show us anything more than we already know.

Conversely if the deviation was outside the standard would it indicate a hot or cold table? From a more useful standpoint- can one use this to shift strategies from a choppy "win, lose, win, lose, lose, win, lose, win, win,...." type of play to a more aggressive strategy.

I'd like to learn how to read the table better to shift strategies. If nothing else it is mentally entertaining. Maybe the best thing to learn is when to walk away:)

Quote: JC308

I'd like to learn how to read the table better to shift strategies. If nothing else it is mentally entertaining. Maybe the best thing to learn is when to walk away:)
link to original post

If you can predict future outcomes based on the past outcomes, you have succeeded where so many countless others have failed.

Have fun throwing the dice and cheering with your friends (unless you're on the darkside).

Quote: JC308

[snips]

... the characteristics of a choppy table .... would it indicate a hot or cold table? ... how to read the table better to shift strategies.
link to original post

in spite of your protest when first confronted about it, the language you use indicates you cling to gambling fallacies. You will have to show you have rid yourself of them, or the help you get here from the best sources will cease

Quote: JC308

Mental- very helpful. Thank You.

My original intent here was to try to identify the characteristics of a choppy table during a "session" sized data sample which has led me to deviation and now phase and frequency. Straying into standard deviation, I was curious if the deviation over a session sample size remained within the standard deviation would that be a good indicator? Would it be better than just using win/loss ratios- which over a larger sample size just become the mean and doesn't show us anything more than we already know.

Conversely if the deviation was outside the standard would it indicate a hot or cold table? From a more useful standpoint- can one use this to shift strategies from a choppy "win, lose, win, lose, lose, win, lose, win, win,...." type of play to a more aggressive strategy.

I'd like to learn how to read the table better to shift strategies. If nothing else it is mentally entertaining. Maybe the best thing to learn is when to walk away:)
link to original post

If the trials are independent, then nothing about the DCT, variance, standard deviation, or 'choppiness' has any predictive power about the future. You can look at the past all you want, but I don't believe it will affect whether you win or lose.

Why is the table choppy? Why not say the dice are choppy? Or is it just the two dice that are chosen for each roll that are choppy and the other three dice are not? Or maybe the shooters are choppy? Or maybe the chandelier above the table is choppy?