Side bet bj paid 18:1

[flixo]

hello,

Thanks

[Three]

On average you get a BJ very 21 hands. So on average it is a horrible bet. But with an ace side count this bet is probably vulnerable. Variance will be huge though.

[flixo]

BS

[RS]

If I'm not mistaken, figuring out the advantage at each count (TC) doesn't require simming, it's purely mathematical. I'm assuming a ten vs no ten count & side of aces. As far as determining what your hourly WinRate would be, or if it's even +EV or feasible, I believe could still be determined using math, but for those of us who aren't so smart---probably easier just to sim the game (and for me, that means let someone else do it).

[Dog Hand]

flixo,

I simmed the "BJ Pays 18:1" sidebet in CVData. Since you failed to mention all the rules, here are the assumptions I made:

1. The sidebet pays 18-to-1, not 18-for-1: in other words, if you bet $10 and win, you get $180 plus your original $10 back. If you don't get your original bet back, then we call that "18-FOR-1" instead.
2. The base game is 6D, H17, DA2, DAS, 3:2 BJ, with American hole card rules, so the BS EV is -0.64%.
3. The penetration is 75%.

Using the tags you suggested, A=-4, X=-1, 2-9=+1, I ran a sim to find that the SB is +EV at a TC of +2. I then ran another 400-million-round CVData sim for a heads-up BS player playing-all and flat-betting $25/rd on BJ, and also betting $25/rd on the SB whenever the TC is +2 or more. The results show a win rate of $43.11/100 rds, with an IBA of 1.725%: pretty strong. The SB itself is worth $60.18/100 rds. The player places the SB 95,424,548 times in 400-million rounds, or on 23.85% of the rounds. He wins the SB 5,529,125 times, or just over 5.79% of the time he plays the SB, for a SB EV of 10.091%.

But wait! The player can do better than that.

Your tags do not represent the "Effect of Removal" (EoR) of each of the ranks: they under-value the Ace, and over-value the X.

For this SB, the EoR for each rank can be calculated very simply. If we know the probability of winning the SB, the SB EV is given as just

EV SB = 18*(Win Prob) - 1*(Loss Prob) = 18*(Win Prob) - 1*(1- Win Prob) = 19*(Win Prob) - 1

For the full 6D shoe, the BJ Prob "BJ Full" is calculated as follows:

BJ Full = 2*24*96/(312*311) = 0.04749... = 4.749%.

Thus, for the full shoe, we get:

EV Full = 19*(0.04749...) - 1 = -0.09770... = -9.770%.

Now to calculate the EoR's, we simply redo these calculations after removing one card of the indicated rank. Thus, if we remove one Ace, we get this:

BJ-A = 2*23*96/(311*310) = 0.04580... = 4.580%, so

EV-A = 19*(0.04580...) - 1 = -0.12972... = -12.972%.

By comparing these two EV's, we see that the effect of removing one Ace is to decrease the EV from -9.770% to -12.972%: a change of -3.202%. Thus, the A EoR is -3.202%.

If we repeat the EoR calculations, but this time we remove one X, we get the X EoR = -0.364%. The reason it isn't higher is because we have so many X's in the 6D shoe.

Finally, if we redo the EoR calculations, but this time we remove a 2, we get the 2 EoR = +0.582%. Naturally, the EoR of any other card ranked 3-9 is the same as that of the 2 for this SB, so we now have the complete EoR values:

A EoR = -3.202%
X EoR = -0.364%
2-9 EoR = +0.582%

Using these EoR's, I selected these tags: A = -16, X = -2, 2-9 = +3. This gives a balanced system, with the ratios of the tags close to the ratios of their EoR's.

Finally, using this new set of tags, I reran the CVData sim to determine that the SB is +EV at a TC of +8 & up. I then ran another 400-million-round CVData sim for a heads-up BS player playing-all and flat-betting $25/rd on BJ, and also betting $25/rd on the SB whenever the new TC is +8 or more. The results show a win rate of $46.00/100 rds, with an IBA of 1.840%: even stronger than before. The SB itself is worth $62.99/100 rds. The player places the SB only 86,229,841 times in 400-million rounds, or on 21.55% of the rounds. He wins the SB 5,068,864 times, or just under 5.88% of the time he plays the SB, for a SB EV of 11.688%.

What could improve these results even more?

A major improvement would be to find deeper penetration, as that will produce a larger fraction of advantageous counts to play the SB. A minor improvement would be to generate indexes for these tags, so that the player needn't play BS.

Hope this helps!

Dog Hand

[GeorgeJung]

Dog Hand

Thank you for your post!
Great explanation and very easy to follow.

[RS]

DogHand, say you were side counting aces and -2 for faces and +1 for 2-9's. I'm a little fuzzy on how to go about it, but I'd suspect if you didn't lump the Aces with the other cards (when counting them), you could increase your win rate. Not sure how difficult it'd be to do the math on-the-fly. Essentially, the main count would keep track of deficit or surplus Tens with an ASC, so you could have 100% correlation.

However, (I'm taking your math posted here as being accurate) it looks like making the wager ~22% of the time using the simple count (-4, -1, +1), you're just gonna be chasing pennies if you try to go for a more advanced system. It'd be like trying to master the minute details in VP strategy (when to hold TJA suited over a flush kinda nonsense) on a video poker promo where you get 10% cash back. If it's a good enough game, learn a simple strategy and attack it ASAP---don't waste your time learning something that could get burnt out by someone else.

[ZenMaster_Flash]

The max bet of $25 renders this topic academic.
"Exploitation" measured in loose change is silly.

[flixo]

Thanks Dog Hand can you contact me by e-mail cause I have cvdata and have almost same results but you seem to be better than me at that
[email protected]

[RS]

The max bet of $25 renders this topic academic.
"Exploitation" measured in loose change is silly.

Didn't you say you got 86'd somewhere for spreading something like $25-$75 or $25-100?

[Norm]

Value is what value is to the particular person's circumstances. We need to stop assuming that everyone is the same and that there is "one way."

[ZenMaster_Flash]

Didn't you say you got 86'd somewhere for spreading something like $25-$75 or $25-100?

What do you mean ? What has that to with this topic ?

[RS]

What do you mean ? What has that to with this topic ?

The post of yours I quoted essentially said, "This is purely academic because it's not worth enough", unless you meant something else by "'Exploitation' measured in loose change is silly." If you still can't figure out my point -- I don't know what to say.

Value is what value is to the particular person's circumstances. We need to stop assuming that everyone is the same and that there is "one way."

Agreed.