New game house edge question/analysis

[SunOfBeach]

I have finished a first attempt, and think that it's accurate... of course, a couple of reviewers would surely be appreciated.

For the record, I think I've "proven" that I was wrong, and that the "casino" is indeed mispricing the highroller option. It's about 1013:1, and they're paying 1200:1 (and I think I actually see how/where they goofed on the analysis, too). Of course, this means that I LOSE my side bet... so much for assuming the casino is always right (and for assuming that a "casino" based in a warehouse in South Boston should be considered a "casino").

I'm having difficulties with using BruceZ's suggestions on copying from excel to here (it's a mess to format)... and I also have no idea how to get this onto my webpage. So, either BruceZ or PairTheBoard (or anyone else) - may I email it to you?

PC

[PairTheBoard]

[ QUOTE ]
I have finished a first attempt, and think that it's accurate... of course, a couple of reviewers would surely be appreciated.

For the record, I think I've "proven" that I was wrong, and that the "casino" is indeed mispricing the highroller option. It's about 1013:1, and they're paying 1200:1 (and I think I actually see how/where they goofed on the analysis, too). Of course, this means that I LOSE my side bet... so much for assuming the casino is always right (and for assuming that a "casino" based in a warehouse in South Boston should be considered a "casino").

I'm having difficulties with using BruceZ's suggestions on copying from excel to here (it's a mess to format)... and I also have no idea how to get this onto my webpage. So, either BruceZ or PairTheBoard (or anyone else) - may I email it to you?

PC

[/ QUOTE ]

BruceZ is definitely the one to email anything that requires either significant work or truely expert analysis. He's also the one to figure how to get the data posted here.

What did you figure the -EV was for the game if there were no bonus? How bad does the "3 tries and you're out" rule hurt your EV?

PairTheBoard

[PairTheBoard]

[ QUOTE ]
I have finished a first attempt, and think that it's accurate... of course, a couple of reviewers would surely be appreciated.

For the record, I think I've "proven" that I was wrong, and that the "casino" is indeed mispricing the highroller option. It's about 1013:1, and they're paying 1200:1 (and I think I actually see how/where they goofed on the analysis, too). Of course, this means that I LOSE my side bet... so much for assuming the casino is always right (and for assuming that a "casino" based in a warehouse in South Boston should be considered a "casino").

I'm having difficulties with using BruceZ's suggestions on copying from excel to here (it's a mess to format)... and I also have no idea how to get this onto my webpage. So, either BruceZ or PairTheBoard (or anyone else) - may I email it to you?

PC

[/ QUOTE ]


You know Sun, if you're correct with those Bonus odds I'm thinking, based on some very rough calculations, you might still be ok with your side bet.

The Bonus would be adding a little less than 20% to your EV on the $1 Bonus Bet. But the -EV on the minimum $10 Come Bet looks pretty bad. Craps with dice is about -1.5% EV if I recall. I'd guess this would be slightly worse due to effects of removal of your point cards. The 20% +EV on the $1 Bonus Bet would just barely make up for that. However, the big difference is the "3 times and your out" rule. You will be working on a point roughly 24/36 of the time - when you don't "roll" 2,3,7,11,12. The "3 times and your out" rule hurts your EV compared to regualar craps when you don't roll your point or 7 for 3 straight times. This happens for the following points roughly,

6,8 : (25/36)^3 = 33%
5,9 : (26/36)^3 = 37%
4,T : (27/36)^3 = 42%

Weighting these by their relative frequency:

10(.33) + 8(.37) + 6(.42) / 24 =
3.3 + 2.96 + 2.52 /24 = 36%

So the 24/36 of the time you get a point you lose automatically 36% of the time. That's a total of about 24% of the time. Of that 24% of the time you would normally win with points

6,8 : 5/11 = 45%
5,9 : 4/10 = 40%
4,T : 3/9 = 33%

Again weighted for frequency:

10(.45) + 8(.4) + 6(.33) / 24 = 40%

So (.40)*(.24) = 9.6% of your wins have been chopped off by the "3 times and your out" rule. Roughly, without figuring effects of removal and with some loose rounding above.


That means if you made 1000 bets you would normaly win about 492. The 3 times and you're out rule reduces this to about 445. You lose 555 for a net loss of 110. If the table minimum is $10 this means for every 1000 wagers of $10 you will lose on average $1,100. Ah, but you are also betting $1 on the bonus 1000 times and on those bets you will win on average about $200. That's a net loss of $900 on every $11,000 wagered. Looks like pretty negative EV. Worse than Roulette.

Better check my work.


PairTheBoard

[SunOfBeach]

I see that I should have studied regular craps a bit here, as I clearly spent alot of time reinventing a circular object used for moving fast... I thought it was a "finding" that {4,10}, {5,9}, and {6,8} were the same :~). Anyhow...

My numbers seem to match roughly with yours - I've found total EV of -.05689 on the come bet, derived as p{7,11} = .22377, p{2,3,12} = -.10956, and then the sum of the conditional probabilities weighted for the 2/3 of the times you go forward to the "3 and out" bet is about -.17...

I was surprised that the high-roller bet was priced at +EV, and didn't even think of the fact that the game itself was still highly -EV...

Would still love to check my numbers with someone, if anyone is interested.

PC