Craps Simulator
Thread Rating:
November 15th, 2019 at 12:27:02 AM
permalink
I have used a craps simulator and the pass line wins far more often than the don't pass side. I can see each number rolled, so it doesn't just spit out the words "Pass" or "Don't Pass".
How does the house make a 1.4% edge on the pass line if it is winning so many times more frequently than the Don't Pass side?
Maybe I have just seen a very, very long Pass line win streak, or does the Pass line always win more often than the Don't Pass side and the house makes it odds some other way?
How does the house make a 1.4% edge on the pass line if it is winning so many times more frequently than the Don't Pass side?
Maybe I have just seen a very, very long Pass line win streak, or does the Pass line always win more often than the Don't Pass side and the house makes it odds some other way?
November 15th, 2019 at 3:21:08 AM
permalink
Quote: NewtoTownMaybe I have just seen a very, very long Pass line win streak, or does the Pass line always win more often than the Don't Pass side and the house makes it odds some other way?
You saw a long Pass Line win streak.
The Pass Line, and the Don’t Pass, each win 49.3% of the winning hands. The DP has a slightly lower House Edge due to the Push on the 12.
Last edited by: Tanko on Nov 15, 2019
November 15th, 2019 at 5:08:40 AM
permalink
The Pass either Wins or Loses, whereas the Don't Pass Wins, Loses or Standsoff. So the Don't Pass wins less often, loses whenever the Pass Wins, and standoffs on 12.
P W: 49.293% P L: 50.707%
DP W: 47.929% DP T : 2.778% DP L: 49.293%
If you ignored the standoffs DP wins 49.299% and loses 50.701%, so has a slightly lower House Edge.
P W: 49.293% P L: 50.707%
DP W: 47.929% DP T : 2.778% DP L: 49.293%
If you ignored the standoffs DP wins 49.299% and loses 50.701%, so has a slightly lower House Edge.
November 15th, 2019 at 7:17:27 AM
permalink
you did not define 'far more often'Quote: NewtoTownI have used a craps simulator and the pass line wins far more often than the don't pass side.
I have also used many craps simulators but
using the "'Perfect 1980" one could see over 1,980,000 comeout rolls (pass line decisions) the pass line is expected to win
976,000 times (976/1980 - win probability)
while the Dpass wins 949,000 times (949/1980 - win probability)
for a difference of 27,000 times in favor of the Pass line.
that could be seen as 'far more often' (some call that the frequency)
as to your question on the house edge, below are a few tables to use for the calculations
Pass line house edge = 28/1980 (7/495) about: 0.014141414
Dpass line house edge = 27/1980 (3/320) about: 0.013636364
Pass Line
| Event | ways | pct |
|---|---|---|
| comeout win | 440 | 22.22% |
| comeout loss | 220 | 11.11% |
| win on 6 | 125 | 6.31% |
| loss on 6 | 150 | 7.58% |
| win on 8 | 125 | 6.31% |
| loss on 8 | 150 | 7.58% |
| win on 5 | 88 | 4.44% |
| loss on 5 | 132 | 6.67% |
| win on 9 | 88 | 4.44% |
| loss on 9 | 132 | 6.67% |
| win on 4 | 55 | 2.78% |
| loss on 4 | 110 | 5.56% |
| win on 10 | 55 | 2.78% |
| loss on 10 | 110 | 5.56% |
| total | 1980 | 100.00% |
DPass
| Event | ways | pct |
|---|---|---|
| comeout win | 165 | 8.33% |
| comeout loss | 440 | 22.22% |
| push | 55 | 2.78% |
| win on 6 | 150 | 7.58% |
| loss on 6 | 125 | 6.31% |
| win on 8 | 150 | 7.58% |
| loss on 8 | 125 | 6.31% |
| win on 5 | 132 | 6.67% |
| loss on 5 | 88 | 4.44% |
| win on 9 | 132 | 6.67% |
| loss on 9 | 88 | 4.44% |
| win on 4 | 110 | 5.56% |
| loss on 4 | 55 | 2.78% |
| win on 10 | 110 | 5.56% |
| loss on 10 | 55 | 2.78% |
| total | 1980 | 100.00% |
as pointed out the DPass Standoff (PUSH) is the difference and it does happen often enough for the house to offer the bet (been around since the early 1900s)
winsome johnny (not Win some johnny)
November 15th, 2019 at 9:55:11 AM
permalink
When 10 out of 10 shooters don't make their point, it's best to bet DP with triple odds. They may be sharp with come-out 7 winners, but not making the point is where you'll get the advantage, especially if they are sharp with the point 7-out.
If I'm alone at the table, maybe I'll set the dice for a 5 or a 9 on the come-out, and set for a 7 after the point is established.
If I'm alone at the table, maybe I'll set the dice for a 5 or a 9 on the come-out, and set for a 7 after the point is established.
November 15th, 2019 at 3:12:07 PM
permalink
I know there is no 'proper' answer, but what is the absolute longest number of P/DP decisions combined where the Pass Line loss rate should fall back in line with the expected stats? 500 decisions? 1,000, 5,000?
I was thinking there must be an upper limit at some point.
In other words, after how many total decisions should it be before I start to see a 1.41% edge for the house (or in the case of the current simulator, at 500 decisions and climbing I am still way in favor of the pass line.
I was thinking there must be an upper limit at some point.
In other words, after how many total decisions should it be before I start to see a 1.41% edge for the house (or in the case of the current simulator, at 500 decisions and climbing I am still way in favor of the pass line.
November 15th, 2019 at 4:17:56 PM
permalink
You'll probably hit your peak upside around 1,000 decisions, and you'll have a hard time not being negative by the time you get to 5,000 decisions.
If you play for several hours a day, try to quit ahead well before you get to 5,000 decisions. It only gets worse the longer you go on. Like start each day with a simple buy-in and have win goals like double or triple your money unless you're on a hot streak. For tax reasons, my gambling diary works better if I quit the day ahead. Losses may not be tax deductible, so I try to limit those.
I've seen graphs where simple PL & DP results just show a negligible downward slope that adds up over time, but when you add odds bets you can seriously affect your balance, be it positive or negative. Some people like to play a couple come bets with odds and a long roll can make up for a couple days of losses.
If you play for several hours a day, try to quit ahead well before you get to 5,000 decisions. It only gets worse the longer you go on. Like start each day with a simple buy-in and have win goals like double or triple your money unless you're on a hot streak. For tax reasons, my gambling diary works better if I quit the day ahead. Losses may not be tax deductible, so I try to limit those.
I've seen graphs where simple PL & DP results just show a negligible downward slope that adds up over time, but when you add odds bets you can seriously affect your balance, be it positive or negative. Some people like to play a couple come bets with odds and a long roll can make up for a couple days of losses.
November 15th, 2019 at 5:19:05 PM
permalink
In the case you gave where there were 27,000 more pass line decisions, had a player been betting $1000 per hand, would that player have won $27,000,000 over the 1,980,000 Come Out Rolls? Or, are you saying DP wins is excluding Pushes on the DP side, so in essence there would still have been a loss on the P side corresponding roughly to the house edge of 1.41% (had one been flat betting the whole time).
My scenario is after almost 2,200 decisions I am at a Pass ratio of 170 to DP ratio of 143 (170 wins per 143 losses, had I bet only the P side). This seems like a lot of wins on the pass side for a very long time period... although I get it, in the scheme of things, it could/should reverse back to where the Pass side is at 49.28 wins (P) to 50.72 losses (DP).
I was wondering if this number of winning pass decisions (as a ratio of wins to losses) is atypical for a string of 2,200 decisions??
Hoping you/someone has been playing long enough to have seen this or know the math behind if this is possible, or maybe if the simulator is skewed toward Pass Line decisions...
My scenario is after almost 2,200 decisions I am at a Pass ratio of 170 to DP ratio of 143 (170 wins per 143 losses, had I bet only the P side). This seems like a lot of wins on the pass side for a very long time period... although I get it, in the scheme of things, it could/should reverse back to where the Pass side is at 49.28 wins (P) to 50.72 losses (DP).
I was wondering if this number of winning pass decisions (as a ratio of wins to losses) is atypical for a string of 2,200 decisions??
Hoping you/someone has been playing long enough to have seen this or know the math behind if this is possible, or maybe if the simulator is skewed toward Pass Line decisions...
November 15th, 2019 at 5:22:32 PM
permalink
I think you may have just answered my next post... which is that by 5000 decisions is where at the very latest, I should likely see a turnaround. Thank you... I had missed this reply before I posted a follow up to 7craps's reply.
November 15th, 2019 at 6:42:13 PM
permalink
February 14th, 2010 at 12:26:06 AM
goatcabin
https://wizardofvegas.com/forum/off-topic/general/972-the-hoax-that-is-the-1-41-house-advantage-on-pass-line-bets/4/
1980, of which 976 are winners, 1004 losers, 784 are seven-outs, so the probability of sevening out on any given bet is 784/1980 = .396
Down 28 units after 1980 decisions. YMMV
These are decisions on your bet, not the number of times the dice are thrown.
goatcabin
https://wizardofvegas.com/forum/off-topic/general/972-the-hoax-that-is-the-1-41-house-advantage-on-pass-line-bets/4/
1980, of which 976 are winners, 1004 losers, 784 are seven-outs, so the probability of sevening out on any given bet is 784/1980 = .396
Down 28 units after 1980 decisions. YMMV
These are decisions on your bet, not the number of times the dice are thrown.
Last edited by: ChumpChange on Nov 15, 2019
November 15th, 2019 at 8:45:31 PM
permalink
It's perfectly possible that after 5000 Pass line bets you might be ahead, in fact about 1 time in 40.
Let's use your number of 5000 outcomes (rather than not rolls) betting $1, and use (i) Average = Np (ii) STdDev = SQRT(Npq).
Your average number of wins is 5000*(976/1980) = 2464.6465, which means you lose 2535.3535 making a loss of $70.7071.
This is weird as one Standard Deviation is SQRT ( 5000 * 976/1980 * 1004/1980 ) = SQRT ( 1249.75 ) = 35.3518.
This means that you have to be almost exactly more than 2 SDs lucky, i.e. about 2.5% chance of being ahead after 5000 outcomes ( https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule ).
Let's use your number of 5000 outcomes (rather than not rolls) betting $1, and use (i) Average = Np (ii) STdDev = SQRT(Npq).
Your average number of wins is 5000*(976/1980) = 2464.6465, which means you lose 2535.3535 making a loss of $70.7071.
This is weird as one Standard Deviation is SQRT ( 5000 * 976/1980 * 1004/1980 ) = SQRT ( 1249.75 ) = 35.3518.
This means that you have to be almost exactly more than 2 SDs lucky, i.e. about 2.5% chance of being ahead after 5000 outcomes ( https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule ).
November 16th, 2019 at 8:49:33 AM
permalink
Quote: ChumpChangeFebruary 14th, 2010 at 12:26:06 AM
goatcabin
https://wizardofvegas.com/forum/off-topic/general/972-the-hoax-that-is-the-1-41-house-advantage-on-pass-line-bets/4/
1980, of which 976 are winners, 1004 losers, 784 are seven-outs, so the probability of sevening out on any given bet is 784/1980 = .396
Down 28 units after 1980 decisions. YMMV
These are decisions on your bet, not the number of times the dice are thrown.
Thanks for bringing us back to a classic WoV thread. Met the OP in Biloxi some time later, he was serious, not a troll. Of course he was seriously wrong.
When a rock is thrown into a pack of dogs, the one that yells the loudest is the one who got hit.
November 16th, 2019 at 11:23:42 AM
permalink
I think you mixed up some dataQuote: charliepatrickIt's perfectly possible that after 5000 Pass line bets you might be ahead, in fact about 1 time in 40.
This means that you have to be almost exactly more than 2 SDs lucky, i.e. about 2.5% chance of being ahead after 5000 outcomes ( https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule ).
I did not look to see what exactly you calculated
for 5k resolved same bet pass line bets there is a 15.5% probability of showing a profit after 5k wagers.
easily done by many over a lifetime of bets
this is actually very easy to calculate using a Markov chain
here is some data
5k bets resolved (same bet)
loss:0.83791069
even:0.00684363
ahead:0.15524568
10k bets resolved (same bet)
loss:0.91987749
even:0.00293518
ahead:0.07718733
20k bets resolved (same bet)
loss:0.97686668
even:0.00076354
ahead:0.02236978
100k bets resolved (same bet)
loss:0.99999607
even:0.00000011
ahead:0.0000037 (about 1 in 270,271)
winsome johnny (not Win some johnny)
November 16th, 2019 at 2:00:07 PM
permalink
Sorry my maths above was out by a factor of two (not enough beer)!Quote: charliepatrick...Let's use your number of 5000 outcomes (rather than not rolls) betting $1, and use (i) Average = Np (ii) STdDev = SQRT(Npq).
Your average number of wins is 5000*(976/1980) = 2464.6465, which means...
This is weird as one Standard Deviation is SQRT ( 5000 * 976/1980 * 1004/1980 ) = SQRT ( 1249.75 ) = 35.3518....
I ran a simulation (100000 runs of 5000) and agreed with your numbers:-
Ahead 15.548%, Same 0.645% Behind 83.807%.
My mistake was to double count the difference, the average number of wins for 5000 outcomes is 2464.65. It only has to get to 2500 or better to show a profit. This is only ONE Standard Deviation out. So ignoring the chances of being spot on one end of the tail is (100% - 68.27%) / 2 = 15.86% which closely matches Ahead and half the Same.
So it might well be true that a Craps player who doesn't play very much is still ahead.
November 17th, 2019 at 8:47:16 AM
permalink
and for those that do play a lot, they can still show a profit from just one bet. depends on how many bets they actually make and resolve.Quote: charliepatrickSo it might well be true that a Craps player who doesn't play very much is still ahead.
I doubt one can find a craps player that makes just one bet, the same all the time.
Take a Place 6 bettor only.
the probability of showing a net loss
after 5k resolved wagers is 0.83833106
a net profit probability of 0.16166894
Most will make a Place 6 and Place 8
the probability of showing a net loss
after 5k resolved wagers is 0.82875728
a net profit probability of 0.17124272
some may form a conclusion on this while not seeing the whole picture (or understanding the picture).
it can take many resolved wagers until the expectation plays out
(ev dominates standard deviation)
winsome johnny (not Win some johnny)
