HOW TO BEAT ATLANTIC CITY BLACKJACK
\n Copyright 1991, Michael Hall<\/p>\n
—————-> Part 1: The Basics
\n Part 2: About the Strategy Charts
\n Part 3: Postscript Strategy Charts (LONG)<\/p>\n
Help for the novice blackjack player
\n====================================
\nThe basic idea of the game is to get a total less than 21 that is
\nhigher than the dealer OR to not bust (go over 21) when the dealer
\nbusts. With basic strategy, you reduce the house edge to about -.45% in
\nAtlantic City (or -.40% where late surrender is offered); it is the
\n*best* way to play, unless you are counting cards. All hands are dealt
\nface up in Atlantic City; don’t touch the cards. A “soft” total means
\nyou have an ace and can use it as 11 without going over 21; hard means
\nyou aren’t counting an ace as 11 in your total.<\/p>\n
Insurance is a side bet for up to half of your original bet. It can
\nonly be placed at the start of a round when the dealer has an ace
\nshowing. A basic strategy player should never take insurance.
\nInsurance pays 2-1 only if the dealer has blackjack.<\/p>\n
With early surrender, you can give up half your bet to avoid playing
\nyour first two cards; late surrender is the same, except you still
\nlose you whole bet if the dealer has blackjack. To surrender, just say
\n“surrender.” <\/p>\n
Splitting can be done only on your first two cards in Atlantic City.
\nYou push out a bet equal to your original, the dealer splits the cards
\napart and deals a card to the first one, which you play normally
\nexcept that you can’t resplit, and then the dealer deals a card to the
\nsecond one, which again you play normally without resplitting.<\/p>\n
Doubling can be done on any two cards. You push out a bet equal to
\nyour original, and you will receive exactly one more card.<\/p>\n
Standing versus hitting is the most common and important decision. To
\nhit you tap the table or draw your fingers towards you. Standing is
\nindicated by a waving motion parallel to the table.<\/p>\n
About Atlantic City
\n===================
\nAll Atlantic City casinos use the same rules, except when they get
\nspecial permission from the Gaming Commission to try something else.
\nAtlantic City rules are no resplitting, split aces get only one card
\neach, double down allowed after split, dealer stands on soft 17,
\nblackjack pays 3 to 2, insurance pays 2 to 1, and 4, 6, or 8 decks
\nare used, but you will find only 8 decks for less than $25 minimums.
\nUntil recently the absolute lowest minimums were $5, but now the Taj
\nMahal offers $3 tables during the day on weekdays. Late surrender was
\nfound only at the Claridge until recently, when Trump Plaza announced
\nthat it is offering it too. It is unlikely that early surrender will
\never be offered again, because the casinos lost so much money when it
\nwas offered that the Gaming Commission declared a state financial
\ncrisis (or some such) in order to get rid of it and protect this
\nsource of New Jersey tax revenue.<\/p>\n
Help for the aspiring card counter
\n==================================
\nI recommend Stanford Wong’s book, “Professional Blackjack” as a
\nreference on the High-Low counting system; it is finally out as a
\npaperback after 9 years of being a hardback. I also recommend Humble’s
\n“The World’s Best Blackjack Book”, which focuses on the Hi-Opt I
\ncounting system but which has lots of general information that any card
\ncounter should know, though the authors of this book are a little too
\nparanoid about getting cheated. Hi-Opt I and High-Low counts are very
\nsimilar, but I feel that High-Low is marginally better for most
\nplayers. More advanced counts do exist (using more numbers than -1, 0,
\nand +1), but they offer very little theoretical gain coupled with an
\nincreased chance for errors. Most professional card counters use
\nHigh-Low or Hi-Opt I. An additional reference containing useful tables
\nof information is “Fundamentals of Blackjack” by Chambliss & Rogenski.
\nFor example, they give a table that shows the effects of various rules
\non basic strategy expectation for 1, 2, 4, 6, or 8 decks.<\/p>\n
Here is how you do the High-Low count. Initialize running count to
\nzero at start. Add one for each 2, 3, 4, 5, or 6 you see and subtract
\none for each 10 or Ace you see. Divide running count by estimated
\nnumber of unseen decks to get true count used in the strategy
\nadjustment table. The strategy adjustment table is just a minor
\nrefinement; you get most of the benefit of counting from bet size
\nvariation, and you should do fine if you avoid strategy adjustments at
\nfirst.<\/p>\n
The Kelly Criterion is a betting heuristic that minimizes your chance
\nof going broke while maximizing your long-run profits, and for
\nAtlantic City, this heuristic dictates that you should bet
\napproximately (TC*0.5 – 0.5)*.0077*BR, where TC is the True Count and
\nBR is your BankRoll (i.e., how much money you’ve got on you.)<\/p>\n
On games with large numbers of decks, it is absolutely imperative that
\nyou abandon the table when the count goes negative. How negative?
\nThat’s a personal decision and depends on your betting spread (difference
\nbetween your lowest bet and your highest), but I would advise leaving
\neight deckers when the count hits -1.<\/p>\n
You should only take insurance if the TC is above +3 (more precisely,
\n+2.8 for four decks, +3.0 for six decks, and +3.1 for eight decks).
\nDon’t be swayed by what cards you have (i.e., don’t fall into the
\ninsure-your-blackjack trap); it’s a side bet, so only the count
\nmatters.<\/p>\n
The maximum edge that most card counters claim to attain in practice
\nis about 1.5%. In Atlantic City, you will need about a 1-8 spread
\n(i.e., highest bet is eight times your lowest) to grind out any profit
\nat all. My simulations show a .5% advantage (ratio of winnings to
\ntotal amount bet) for a 1-8 betting spread, 7 players, -1 to +10 strategy
\nadjustments, and abandoning counts of -1 or worse. If late surrender
\nis available, the edge improves to .66%.<\/p>\n
As far as risk goes, a 500 unit bankroll (e.g. $2500 for $5 minimums)
\nhas a 81.5% chance of doubling before going broke. If late surrender
\nis available, this improves to 89.3%. You are risking quite a bit
\nto win how much? 0.9 units is the average win per 100 hands; 1.3 units
\nwith late surrender. So you could make about $5 an hour or so if you
\nare willing to have more than a 10% chance of losing $2500 before
\ndoubling it.<\/p>\n
If you want to make money at blackjack, either join a blackjack team or
\nplay the single or double deckers in Vegas.<\/p>\n
HOW TO BEAT ATLANTIC CITY BLACKJACK
\n Copyright 1991, Michael Hall<\/p>\n
Part 1: The Basics
\n—————-> Part 2: About the Strategy Charts
\n Part 3: Postscript Strategy Charts (LONG)<\/p>\n
Description
\n===========
\nThis article describes basic and High-Low strategy tables for Atlantic
\nCity rules with four or more decks. The strategy information was taken
\nfrom Stanford Wong’s book, “Professional Blackjack”. The tables tell
\nyou the mathematically best play given a certain circumstance –
\nwhether to surrender, split, double-down, hit, or stand.<\/p>\n
Rationale
\n=========
\nI made these tables for myself, because I was unsatisfied with
\nany I could find in published books. I am very satisfied with the
\nresult, so I thought I would share it with y’all. You may wish to
\nmodify the tables for your particular situation (different counting
\nsystem, different casino rules, etc.) If so, you’ll need to get the
\ntroff source from me, or else you can use the “Do Your Own Strategy”
\nblank chart that is included.<\/p>\n
Caveats
\n=======
\nI do not guarantee that these tables are correct. If you find any
\nmistakes, or have any suggestions, please let me know, and I will
\nrepost if necessary. Also, note that Wong computed his numbers for 4
\ndecks, and he assumes 4 decks = 6 decks = 8 decks for purposes of
\nstrategy adjustments. If anyone has High-Low strategy numbers that have
\n*proved* to be more accurate for 6 or 8 decks, then let me know.<\/p>\n
How to print the tables
\n=======================
\nIn a subsequent article, you’ll find the Postscript gobble-dee-gook
\nthat hopefully can be understood by your printer. However, it’s
\nuuencoded and compressed. Save the article to a file. “uudecode”
\nthe file. “uncompress” the resulting file, high-low.ps.Z (“uudecode”
\nand “uncompress” are UNIX programs that you hopefully have. There is
\nno need to strip out the article header before running uudecode. If
\neverything works, then you should wind up with a file named high-low.ps
\nthat has “%!PS-Adobe-1.0” as its first line.) Send high-low.ps to a
\nprinter that understands Postscript. (This includes the popular
\nApple Laserwriter II printer and many others.) There will be a few
\nsemi-blank pages, because the original text formatter, troff, is
\nbrain-damaged. What you want are the pages with the tables for
\n“High-Low”, “Basic Strategy”, and “Do Your Own Strategy”.<\/p>\n
How to use the Do Your Own Strategy table
\n=========================================
\nUse the Do Your Own Strategy table for memory recall practice or to
\ndevise a table with a different set of strategy adjustment numbers,
\nperhaps for a counting system other than High-Low.<\/p>\n
How to read the Basic Strategy table
\n====================================
\nCross index your hand with the dealer’s face-up card. If there is an
\n“X”, it means “yes, do the corresponding decision” – conversely,
\na blank means “no, *don’t* do the corresponding decision.” Read from
\nthe bottom up. First see if you should surrender (if this option is
\navailable), then split, then double, then stand. If nothing applies,
\nthen hit. <\/p>\n
For example, suppose you have two 8’s, and the dealer has a 10
\nshowing. If you are playing at the Claridge (or Trump Plaza), you
\nfirst see if you should late surrender, but cross indexing 8-8 with 10
\nunder late surrender shows that you should not. You then check splitting
\n– the table shows that you always split 8’s, since there are X’s all the way
\nacross. However, if you split 8’s and get another hand of 8’s, then
\nyou cannot resplit. You then look up to see if you should double – of
\ncourse not – and then you look up to see if you should stand; 8-8
\nversus 10 is blank, so you don’t stand and instead you take a hit.<\/p>\n
How to read the High-Low Strategy table
\n=======================================
\nCross index as with the basic strategy table. Follow the basic
\nstrategy, except in these cases:<\/p>\n
1) If there is a positive number in the box and the true count is greater
\n than it, it means “override basic strategy, so yes, do the corresponding
\n action.” <\/p>\n
2) If there is a negative number in the box and the true count is lower
\n than it means “override basic strategy, so no, *don’t* do the corresponding
\n action.”<\/p>\n
To conform to the above and to avoid confusion, zeros are noted as
\npositive or negative. The somewhat counterintuitive use of a “Stand”
\ndecision as opposed to a “Hit” decision is again to conform to the
\nabove and to avoid confusion in the long run.<\/p>\n
This all sounds complicated, but it’s simple once you get used to it.<\/p>\n
For example, using the previous example, you would deviate from basic
\nstrategy and surrender 8-8 against 10 if the running count were
\npositive (greater than +0). You would always split 8’s, but you would
\ndeviate from basic strategy and stand on hard 16 when the running
\ncount were positive.<\/p>\n
How to highlight the High-Low Strategy table
\n============================================
\nI highly recommended that you use a highlighting pen to indicate
\nbasic strategy on the High-Low Strategy tables. Overlay your High-Low
\nprinted page on top of your Basic Strategy page. Press down so you can
\nthe X’s through the High-Low page. Highlight everywhere an X shows
\nthrough. Note that there is a basic strategy X everywhere there is a
\nnegative High-Low Strategy number, and there is a basic strategy blank
\neverywhere there is a positive High-Low Strategy number (this would
\nnot be true for some ranges of counts larger than -1 to +6.)<\/p>\n
Still confused?
\n===============
\nYou can send e-mail to hall@rocky.bellcore.com if you have any
\nquestions on these charts.<\/p>\n
