Blackjack Leak

[SHT]

I always hear people say that it is correct to split 88 against a 10. Where did this BS start and why has it not died a horrible death?

For a long term blackjack player, this is a losing play. The logic is that 16 will only win against a 10 23% of the time, while an 8 beats a 10 38% of the time. So you are apparently increasing your chances of winning. Through simple math, I will show you that this is BS.

Suppose I will be faced with this hand 500 times in the future. Assume every bet is $10. If I decide to not split the 8s, I should be winning 23 % of the time. I should win 115 of these 500 hands, being down 385 bets which would be $3850. If instead I decided to split the 88s, By doing this I increase the total bets to 1000 and will now win 38 % of the time. I will win 380 of these hands. Being down 620 bets which would be $6200.

I believe this myth was created by the casinos, to make players lose more money to the casino.

[Jabberwocky]

You've had the misfortune of talking to the wrong people. The computer generated basic strategy calls for splitting. Against Ace it says surrender if allowed.

[21forme]

This thread should be moved to Voodoo

[Three]

If you use a Composition Dependent Analyzer of the top as BS is determined (assumes dealer checked for BJ):
(# of Decks, EV of Standing, EV of doubling, EV of hitting, original bet EV of splitting 1 time, original bet EV of splitting 2 times, original bet EV of splitting 3 times, Late surrender):
(8 deck, Stand -53.78%, Double -107.3%, Hit -53.65%, Split once -48.48%, Split twice -47.76%, Split thrice -47.63%, LS -50%)
(6 deck, Stand -53.69%, Double -107.1%, Hit -53.54%, Split once -48.32%, Split twice -47.61%, Split thrice -47.49%, LS -50%)
(2 deck, Stand -52.96%, Double -105.2%, Hit -52.62%, Split once -47.06%, Split twice -46.42%, Split thrice -46.35%, LS -50%)
(1 deck, Stand -51.83%, Double -102.4%, Hit -51.18%, Split once -45.17%, Split twice -44.71%, Split thrice -44.69%, LS -50%)

As you can see splitting has the highest original bet EV (losses the least money on this losing matchup), followed by late surrender, then hitting barely being better than standing for every # of decks used in the examples above. The answer to your question is because splitting loses less money than hitting or standing and less than surrendering in the long run. So splitting is the best choice and any revision would be a mistake.

http://www.bjstrat.net/cgi-bin/cdca.cgi

[SHT]

As you pointed out, the original bet EV is highest when splitting, but you are making an additional bet when splitting, and the new bet will be in a -ev situation. If you want to ignore the second bet, by all means go ahead, just know that splitting 8s is -ev. I think you are too reliant on external sources that you have abandoned your common sense!

[Stealth]

I always hear people say that it is correct to split 88 against a 10. Where did this BS start and why has it not died a horrible death?

For a long term blackjack player, this is a losing play. The logic is that 16 will only win against a 10 23% of the time, while an 8 beats a 10 38% of the time. So you are apparently increasing your chances of winning. Through simple math, I will show you that this is BS.

Suppose I will be faced with this hand 500 times in the future. Assume every bet is $10. If I decide to not split the 8s, I should be winning 23 % of the time. I should win 115 of these 500 hands, being down 385 bets which would be $3850. If instead I decided to split the 88s, By doing this I increase the total bets to 1000 and will now win 38 % of the time. I will win 380 of these hands. Being down 620 bets which would be $6200.

I believe this myth was created by the casinos, to make players lose more money to the casino.

What you are missing is that splitting 8's was never a winning play, it is a play that reduces your losses thereby becoming a more positive play than not splitting. That is, losing less due to splitting is a better play. Suggest you improve the thoroughness of your information and consider that the advice to split was developed over the past few decades by some of the most intelligent mathematicians in the country.

They are not wrong, regardless of what your "back of napkin" math tells you.

[Norm]

SHT, this is known as a defensive split. Yes, you will lose. But, as Stealth says, you will lose less.

[SHT]

What you are missing is that splitting 8's was never a winning play, it is a play that reduces your losses thereby becoming a more positive play than not splitting. That is, losing less due to splitting is a better play. Suggest you improve the thoroughness of your information and consider that the advice to split was developed over the past few decades by some of the most intelligent mathematicians in the country.

They are not wrong, regardless of what your "back of napkin" math tells you.

How does it reduce your losses? I clearly laid it out how you end up losing more on this play. You don't need an "intelligent mathematician" for this no-brainer. How is my math inaccurate or incomplete in anyway?

[Norm]

Your numbers are incorrect. They don't even take into account pushes and doubles after split. They also ignore the fact that if the dealer has an ace, the split is nullified.

[Three]

Maybe I used the wrong term but what I meant was the expected return for your overall bet condensed to the original bet. Basically if you split you expect to lose less in the long run. I showed you the results, gave you a link to the CDCA calculator but you cling to someone thoughts that didn't know what they were talking about. If you split you expect to lose about -48% of your original bet on each entire round played including the splits and if you hit you expect to lose about -54% of your original bet on the round. That is what I meant by original bet EV. I hope I used the term correctly. Now you do put out more money which increases variance. It isn't you win, push or lose but becomes you win both, win one and push one, push both, win one and lose one, lose one and push one, OR lose both. Covariance comes into play since a strong dealer final total or a dealer bust will increase the likelihood of getting the same result for both hands. It seems your confusion is rooted in both you and your advisor not understanding the percentages.

For a long term blackjack player, this is a losing play. The logic is that 16 will only win against a 10 23% of the time, while an 8 beats a 10 38% of the time. So you are apparently increasing your chances of winning. Through simple math, I will show you that this is BS.

Suppose I will be faced with this hand 500 times in the future. Assume every bet is $10. If I decide to not split the 8s, I should be winning 23 % of the time. I should win 115 of these 500 hands, being down 385 bets which would be $3850. If instead I decided to split the 88s, By doing this I increase the total bets to 1000 and will now win 38 % of the time. I will win 380 of these hands. Being down 620 bets which would be $6200.

(# of Decks, EV of Standing, EV of doubling, EV of hitting, original bet EV of splitting 1 time, original bet EV of splitting 2 times, original bet EV of splitting 3 times, Late surrender):
(8 deck, Stand -53.78%, Double -107.3%, Hit -53.65%, Split once -48.48%, Split twice -47.76%, Split thrice -47.63%, LS -50%)
(6 deck, Stand -53.69%, Double -107.1%, Hit -53.54%, Split once -48.32%, Split twice -47.61%, Split thrice -47.49%, LS -50%)
(

You have you winning 23% and losing 77% if you hit but the truth is you win (23-x)% and lose like (77-x)% and push 2x%(about 7%). Negating pushes gives you your EV but if you compare results to a number of bets you must include pushes. If you resolve 500, bets as in your hitting calculation, you push 10x bets for a total of 500+10x bets made to make your result (570 bets made if you use 7% pushes). If you split you will push on a win/loss combo (which happens a lot), and pushing both hands. So x is much greater for a split than a hit (standing on a stiff has x=0) which causes there to be many more bets made to reach 500 non pushed rounds. To equate results they must be for the same number of bets made including pushed bets. You are comparing apples to oranges in your calculations.

You get a win/loss push on your split:
Dealer 17 (12.24%): you bust one of you 8's (24%) and make 18-21 (63%) on the other. Total 1.8%
Dealer 18 (11.51%): You bust or get 17 on one hand (37%) and you make 19-21 on the other (27%). Total 1.15%
Dealer 19 (12.24%): you bust or get 17-18 on one hand (73%) and make 20 or 21 on the other (14%). Total 1.25%
Dealer 20 (37.08%): You bust or get 17-19 (86%) and you make 21 on the other hand (7%). Total 2.23%
Dealer bust (24.12%): You bust on hand (24%) and get 17-21 on the other(76%). Total 4.40%

So not including push/push scenarios we get pushes on split hands 10.23% of the time. 10.23%+ pushes


Push if you hit 16:
Sum of the dealers chances of making a hand times your chances of making the same hand:
(12.24+11.51+12.24+37.08+ 3.77)* .0769 = 76.84*.0769 = 5.9090% pushes

So you are comparing the results for hitting 559 hands when counting pushes to the results of splitting more than 602 hands. Like I said apples to oranges. Did you really think pushing hands with a huge negative expectation had no effect on your edge? If you stand on that stiff 16 you never push so you would be comparing 500 hands. Do you see the problem with your flawed method of thinking. This is why computers are used to figure these things out. It has been done and there are not mistakes. Next time you want to assign credibility to sources before deciding to believe one over another.

By doing this I increase the total bets to 1000 and will now win 38 % of the time. I will win 380 of these hands. Being down 620 bets which would be $6200.

If you win 380 hands and lose 620 you lose a net of 240 hands and would be down $2400. Last time I checked $2400 is a lot less than $3850.

[Jabberwocky]

The facts are irrelevant when you have a need to believe your own theories.

[MJGolf]

Suppose I will be faced with this hand 500 times in the future. Assume every bet is $10. If I decide to not split the 8s, I should be winning 23 % of the time. I should win 115 of these 500 hands, being down 385 bets which would be $3850. If instead I decided to split the 88s, By doing this I increase the total bets to 1000 and will now win 38 % of the time. I will win 380 of these hands. Being down 620 bets which would be $6200.

I believe this myth was created by the casinos, to make players lose more money to the casino.

I think what you might be thinking about also is the "variation" in the loss/win rate. You will lose less if you split as others here have advised and explained. But BS was never created by the casinos. It was performed and confirmed and calculated by mathematicians far smarter than you and I. You are making the individual decision on your own personal experience that I really DON'T want to risk double my bet in a known losing situation; that is not "to lose less" in the long run but to "lose less with my bankroll".

In actuality, even using YOUR OWN MATH; you are supporting the opposite situation.............If you lose $3850 on 500 hands that is still MORE proportionately than losing $6200 (or $3100 twice) on 1000 hands but in 500 hand increments. Your own math reflects that you would lose $750 less per 500 hands. Perhaps that would get you thinking more along the lines of expected EV. Losing less in the long run is still better even if we all know that this is a crap/losing hand with negative expectation over all.