Basic strategy for these rules please

[dacium]

I just haven't seen any calcs that specifically say split once that all.
My sim is matching everything else except this so I am trying to find out why.

 

[zengrifter]

I just haven't seen any calcs that specifically say split once that all.
My sim is matching everything else except this so I am trying to find out why.

What sim program are you using? zg

 

[MGP]

I just haven't seen any calcs that specifically say split once that all.
My sim is matching everything else except this so I am trying to find out why.

I'm sorry but that's not what you were doing. You asked a question, it was answered and then you said everyone was wrong:

I disagree. Does anyone have any actual sources for tables that are worked out for only split once rule. because there don't seem to be any.

The following is not a question but an answer to your own question that again implies you are right and everyone else is wrong:

Yes thats what I meant. If you can only split once then best thing to do with 88 was hit, next best was split, worst was stand. If you can split multiple times then best was split 88, next was hit, last was stand.

You haven't wanted to listen from the beginning:

I did simulations of 100 million hands and definatly lose alot more standing on 88 than splitting it against a 10 card if I could only split once. From what I could see the main point of spliting is that 16 is such a bad number and there is a chance you can get a 3 to make 11 and double. But with only spliting once the chance of get another 8 again seems to nullify this. Unfortuently I left the results at work but I'll post them tomorrow. It was VERY insignificant but enough to change 88 from split to hit against a ten.

and your very first post:

I have done some math/simulations and it has a big effect for example its not werth splitting 8's against a 10 if you can't split more than once (as far as my simulations/calcs showed).

Just because you haven't seen it on the internet somewhere doesn't mean you are right and no one else has any idea what they're talking about. It's just not clear to me why you even asked the question when you've already determined that you know the answer and that no one else does based on a whole 100M hands.

A 100M hand sim is nothing for strategy decisions. CA when done right is exact.

That being said, in case you are actually interested in learning - here are the exact EV's:

8,8 vs 10 8D D9 NS
SPL1: -0.484873
Hit: -0.536485
Stand: -0.537752

And that's not an insignificant difference and it's not that close.

Sorry for the attitude but it's just reflection of yours.

Sincerely,
MGP

 

[dacium]

I am really sorry if I came off bad because that is not what I mean and I am thankful that many people have replied. My english is not 100%. Please forgive me and I hope you can reply again

I mean to say please can you post link to where you obtained these numbers and if the generator has source code. That is problem I had, people say it is so but I have no way to look it up. I have seen many different numbers for 8,8 vs 10 ranging from -0.48 to -0.57, all apparently for 8 deck, split once, double 9-11, stand 17, no surrender. But obviously only 1 is correct.

Also there seems to be some significant difference in what people do to calculate the EV. Some go ev= (wins - losses) / total games, while some do ev = (wins-losses)/(total games - draws)

 

[MGP]

I mean to say please can you post link to where you obtained these numbers and if the generator has source code. That is problem I had, people say it is so but I have no way to look it up. I have seen many different numbers for 8,8 vs 10 ranging from -0.48 to -0.57, all apparently for 8 deck, split once, double 9-11, stand 17, no surrender. But obviously only 1 is correct.

I use my own program that I wrote. You can use Eric Farmer's program. His splitting values use a CDZ- strategy and are exact for SPL1. His values for SPL2-3 are close estimates. It is open source and free and can be found here:

(Dead link: http://home.comcast.net/~erfarmer137/blackjack/)

Also there seems to be some significant difference in what people do to calculate the EV. Some go ev= (wins - losses) / total games, while some do ev = (wins-losses)/(total games - draws)

EV for specific plays is by convention given as simply win-loss. So if p(pwin)=.5 and p(lose)=0.3 and p(push)=0.2 then EV = 0.5-0.3=0.2.

If it's for a doubled bet then with the same probs it would be 2*(0.5-0.3)=0.4

Good luck,
MGP

 

[dacium]

Thanks!

I have decided to rewrite my program. It was just simulating every type of hand, but this is pointless when I can actually do a recursive algorithm that actually carries through the probabilities and works out the mathematics exactally. I already have it partly going! Hopefully I can achieve the correct results this time. I did find a bug in the old one where the program would be hitting soft 19 in some wrong cases which probably explains why 88 varied so much because much higher chance of getting hands to soft 19.

 

[dacium]

OK Ive finally found out what was going on with the splitting of the 8's. I was using rules of you loose original and busted bets against a blackjack, while the standard rule is only original bets (since america has card peaking). The busted bet loss makes splitting a bad idea because if the dealer ends up with blackjack you could loose both bets.

 

[MGP]

OK Ive finally found out what was going on with the splitting of the 8's. I was using rules of you loose original and busted bets against a blackjack, while the standard rule is only original bets (since america has card peaking). The busted bet loss makes splitting a bad idea because if the dealer ends up with blackjack you could loose both bets.

The strategy for 8,8 vs 10 BBO is still to Split - even with SPL1. Here's a post on BJ Math where Cacarulo simmed the results and I fixed my CA to get the correct values:

http://www.bjmath.com/bin-cgi/bjcomputer.pl?read=2562 (Archive copy)

So I'm sorry but there's still something wrong with your sim. Even with NDAS the strategy is still to split.

Note: BBO is Australian rules Busted Bets plus 1 which is commonly called original and busted bets. I use OBBO as the terminology for a literal original and bused bets. Norm and Cacarulo like BB+1 for the Australian rules but I'm stubborn and think BBO is more consistent with other BJ abbreviations since it doesn't have a "+" sign.

 

[dacium]

Its weird that they call it 'australian' I am from australia and none of the casinos I have played do you loose busted bets. If you bust on a split and the dealer is showing A or T, they stack the cards and chips infront of your square and you only loose original bet to blackjack.

And yeah I have basically figured out my whole split code is wrong! oh well