Why Blackjack Seems to Run in Streaks

2 June 1997

By Alan Krigman

Seasoned blackjack buffs are often bewildered when they get battered in a game. And casino bosses are likewise frequently flabbergasted when dealers take a drubbing. After all, blackjack has low house edge, fluctuation, and skewness. So solid citizens who don't radically alter bet size during a game should drift up and down - with a bias toward the latter. And casinos should enjoy slow but steady income over moderate samples of players.

This is what happens when a craps player makes a single wager, like $6 on the eight, waits until the bet's resolved, then places it again. Or when a roulette player goes on red, spin after spin.

Blackjack doesn't always seem to work that way. More often than can be called coincidence, blackjack appears to run in streaks.

One reason is that blackjack players come up short on single bets, depending on splits and doubles to bring the edge into balance. If hands appropriate for these moves during a session are rare, players generally suffer. If splits and doubles occur frequently but go awry, the effect is a double whammy. And if the gambling gods smile on the advent as well as the resolution of these opportunities, profits can escalate.

But the bankroll swings often encountered in blackjack may have an even more fundamental cause. The low house edge - 0.46 percent in six-deck games with typical house rules - doesn't result because hands have uniformly low margins. Nor because they're clustered in the neighborhood of the average value. On the contrary, the low overall edge is an amalgam of situations heavily weighted toward either the casino or the bettor, with nothing in between. The accompanying table shows that for various dealer upcards, edge extends from 4.13 to 34.17 percent favoring the house and from 5.72 to 23.42 percent favoring the player.

The situation is analogous to a community center which operates a seniors' club and a pre-school day care facility in the same building. The mean age of the participants might be 35 years, although one group is all between 65 and 80 while the other ranges from two to five. No single member of either set is within 30 years of the average. Stand outside at 3:00 pm and watch people leaving for home. Ultimately, you'll get a mixture of ages. But, during any short time span, you could easily encounter a run of geriatrics or juveniles. And, either way, you wouldn't base many decisions on the 35-year average age.

Over a typical blackjack session, a dealer can be expected to draw upcards in proportion to their occurrence in the deck or shoe. That is, one out of 13 times for ace through nine and four out of 13 times for 10-values. During relatively short periods, the dealer may also draw upcards close to the statistically-correct distribution. In these instances, players actually experience an edge approaching the nominal value for the game.

However, short sessions are often characterized by distributions of upcards which don't conform well to the make-up of the deck or shoe. The figures in the table hint that a run of high dealer upcards can severely disadvantage players, while a series of low upcards can ensconce bettors in the catbird seat. The table also suggests why statistics based on a composite value of the edge don't necessarily help anticipate what will happen in any particular session. Sumner A Ingmark, revered among rhymers for rationalization, reveled in reminding readers of this reality thusly:

The rules derived with good statistics,
Should beat those coming from the mystics.
But in close games not oft repeated,
None knows who'll win or be defeated.