What is the worst play in blackjack?

Ignoring abnormal plays such as hitting a 19 or 20 or standing on Ace-Ace, the worst play, expectation wise, is standing on 8,8 versus 7, rather than splitting them! You will lose about 70 cents on the dollar each time you make this play. If you stick to basic strategy you won’t have to worry about questions like this!

David Sklansky presented a short chapter on this very subject in his book Getting the Best of It 1)Getting the Best of It is an outstanding expose on the mathematics of gambling, general gambling concepts, sports and horse betting, poker, blackjack, and the other casino games. It is highly recommended. Sklansky writes, “Picking the worst play is not just a trivial exercise. This is because in order to reach the right answer it is necessary that one understand the underlying concepts used in determining the correct basic strategy. These concepts really all come down to one thing: mathematical expectation.

In my book Blackjack: A Professional Reference, I define the term “expectation” to mean the amount (expressed in dollars or percent) that a player should win (or lose) based on the player’s statistical advantage (disadvantage). If you flip a coin and bet \$100 on heads your expectation is exactly zero dollars. This is because you have an even (50%) chance of winning each bet. In the long run your expectation is to break even. If the coin was rigged to positively come up heads once every 100 flips your expectation would now be \$1 (or 1%) each time the coin was tossed.

“…you have the wonderful appendices to the latest edition of Wong’s Professional Blackjack, in which every expectation is catalogued. For me, this has become the definitive source for this kind of study.”

One easy way to figure out how much of a mistake you would be making by deviating from basic strategy is to use Stanford Wong’s Blackjack Count Analyzer (BJCA) software. All you have to do is set up the game rules, turn basic strategy on and enter the game simulator. The software allows you to set up any hand you want. All you have to do is play the hand differently from basic strategy and you will be alerted that you have made a mistake along with the severity (in percent). In the example given above (8,8 vs 7) BJCA reports a 66.2% error in single deck and 64.4% error in multideck.

But what about plays that are very close? Could these plays be used for camouflage by an experienced card counter? The answer is, of course, yes. A good source for this information (at the time) was in Bill Brown’s 190,000,000 Hands of Blackjack2) After originally writing this article for Blackjack Review Magazine, renowned expert Don Schlesinger, author of Blackjack Attack, questioned my choice of Bill Brown’s book as a source of information. Schlesinger commented, “I was a bit surprised to see you quote, of all people, Bill Brown’s 190 Million Hands of Blackjack as a source for your 8,8 v. 7 study. There are so many other places you might have gone. The first person to print this kind of comparison was Braun, in his How to Play Winning Blackjack. Next, you have the wonderful appendices to the latest edition of Wong’s Professional Blackjack, in which every expectation is catalogued. For me, this has become the definitive source for this kind of study. Finally, I hope you haven’t forgotten my ‘Basic Strategy Camouflage: How ‘Dumb‘ Can You Afford to Appear’ article in the September 1993 Blackjack Forum. Ironically, on page 13, you will find a detailed analysis of the exact play you were discussing. But I consider splitting versus the next logical choice of hitting, not standing. Yours is somewhat of a “reach,” in that you compare the optimal play to the *third* best choice, instead of the usual second.” I commented as follows: You are right, Stanford Wong’s appendix in Professional Blackjack is a much better source for determining camouflage plays. which was published in 1990. Brown ran simulations on just about every hand possibility and reports the difference if you deviate from basic strategy. The only problem with the book is that the resulting percentages are not always accurate due to the limited number of trials for each hand possibility. But the close plays do stand out! Here’s an example: 16 vs 10. Always hitting this hand resulted in a -78131 loss versus a loss of -78755 for standing. The difference was +624 by following basic strategy or put another way, standing on 16 vs 10 is about a 0.8% error. As a comparison, BJCA reports an error of 0.2%. It doesn’t matter what the exact number is. All we are interested in is that the play is not a gross error (i.e., greater than 5% or so).

An interesting camouflage play in multideck might be to hit a 12 versus a dealer 4 when the count is neutral and you have a medium sized bet on the table. Both BJCA and Brown suggest this is less than a 1% error. Of course, if the count was negative you would be hitting this hand anyway but your bet would most likely be minimum.

Just remember, anything can happen in the short run. If you make too many errors you will assuredly go broke. Likewise, card counters who use too many camouflage plays run the risk of giving up their edge entirely!