Chapter 6

CHARLESTOWN, MASSACHUSETTS

January 2006

___________________

“Learning the basic strategy in blackjack is like learning to float in water; it enables you to survive. But if you want to get somewhere, something additional is required. In this sense, learning to count cards for playing blackjack is analogous to learning to swim.”

Peter A. Griffin, The Theory of Blackjack

It was rush hour along the Metro West corridor of Boston and cars were moving at a snail’s pace in every direction, jockeying to get in position for a quicker ride home. Normally my blood pressure would rise in frustration and I’d call a friend or turn up the music. But today was different. I was in a zone, looking from one license plate to the next:

A36 873. Plus 2.

J88 QJ9. Minus 3.

Then there were other things.

Exit 27A. Even.

Toll $2.75. Plus 2.

BMW 325. Plus 3.

I saw numbers everywhere I turned and those numbers represented something unique. Not a license plate, or an exit, or a toll, or a make, or a model. No, I saw playing cards.

I’ll never forget the scene in Rain Man when Tom Cruise realizes that his autistic brother, played by Dustin Hoffman, can keep track of hundreds of playing cards from multiple decks. First stop: the casino. It’s what people think about if you say “Rain Man.” And “Rain Man” is often what people first think about when they hear “card counter.” But the reality is that it doesn’t take autism or genius to count cards. All it takes is a fairly simple and statistically accurate system for identifying the saturation of “good cards” versus “bad cards” in the decks remaining to be played.

Several good systems exist but with most, the premise is comparable. Counting cards allows the player to know when the cards remaining to be dealt favor the dealer or favor the player. When the player has the advantage, he can wager more. When the house has the edge, the player can wager less. But this is just the beginning—the running count.

The running count was part of the traditional high-low system for counting cards. Although Harvey Dubner introduced it in 1963, it became popularized in the second edition of Ed Thorp’s book, Beat the Dealer, in 1966. The ultimate approach that was derived was simple. Low cards (2, 3, 4, 5, 6) were more beneficial to the dealer, while high cards (10, jack, queen, king, ace) were more beneficial to the player. The middle cards (7, 8, 9) were neutral.

The reason the high cards are beneficial to the player is that if the dealer gets a natural (a two-card 21), the player loses just one bet. But if the player gets a natural, he’s paid $3 for every $2 he bets. The 3-2 payoff is what makes the game worth playing. In recent years many casinos have resorted to paying only 6-5 on naturals. Those games aren’t worth playing.

There are other reasons high cards benefit the player. The dealer must hit until he has 17 or higher, but the player may stand on any total. So if the player knows the shoe is rich in higher cards, he can choose to stand on a bad hand and let the dealer take the risk of busting his hand with a high card. When factoring in double downs and splits, the high-value cards generally provide the player more opportunity to make good hands. On the flip side, small cards are more likely to help a dealer who is forced to always draw to at least 17.

So the high-low count is quite simple in theory. Of the 13 cards in a suit, five are helpful, five are not, and three don’t matter. It was an easy choice for us to use the high-low count, the same count used by MIT teams a decade earlier. When a low card (2-6) is dealt, that’s one less “bad” card left to be played. When a high card (10-ace) is dealt, that’s one less “good” card left to be played. Consequently, if low cards are played, since they are bad for the player, a +1 is assigned to them, meaning there’s now one less bad card in the remaining cards to be played. Conversely, if high cards are played, since they are good for the player, a -1 is assigned, since there is now one less good card remaining. And 7s, 8s, and 9s are less significant, so they’re valued at zero.

The high-low system is a “balanced” count, meaning that if you counted every card in a full deck in a manner consistent with the values stated above, the running count would end at zero. In one deck, 20 cards are valued at +1, 20 cards are valued at -1, and 12 cards are valued at zero. Add them all together and you get zero—a balanced count.

Learning to keep the running count, the total of all cards dealt, was quite a challenge at first. D.A. and I had researched the many alternatives—the Wong Halves, Uston APC, K-O, Hi-Opt I, Hi-Opt II, Omega II, Ten Count, Red Seven, and others—but we determined that the high-low was the easiest to learn given the depth of the advantage gained. Some systems were easier but not as strong. More accurate systems were also available but, with added complexity, they lent themselves to greater error rates. Not to mention, we had a deep respect for Semyon and his MIT teammates. If the high-low was good enough for them, then it was certainly good enough for us. We continued to meet every week for practice. My weekend partying time was often spent dealing shoes to myself, instead. Reading Sports Illustrated every week was regularly replaced with another book on blackjack.

We learned tips and shortcuts to help us. When counting a hand that consisted of both a king and a 3, we wouldn’t first count the king as -1 and the 3 as +1, to achieve our count of zero. Instead, we recognized that a king and could be canceled out. It was known a “meaningful pair” that equaled zero. Two high cards equaled 2, two low cards equaled -2, and a high card and a low card were zero. We didn’t have to process the math one card at a time.

But getting good meant practice. So I practiced with everything my eyes saw. A busy rush hour became the perfect training ground.

The world became my blackjack table.