In 2005, Craig Bennett went shopping in a California fish market. He saw a large salmon that looked good, so he bought it. Like most fish in a market, this one was quite dead. However, it soon would have a curious afterlife that would affect the world of neuroscience, statistics, and nearly everything related to numbers, including blackjack.

Bennett was a postdoctoral researcher in the Department of Psychology at the University of California in Santa Barbara. He took the dead fish back to his lab, put it in an fMRI scanner, and did various brain scans on the deceased animal. Why? Well… he was practicing with the equipment. His previous experiments involved scanning a pumpkin and a dead bird. However, this time something amazing happened...

According to a paper that Bennett later wrote (in somewhat flowery scientific prose), "The salmon was shown a series of photographs depicting human individuals in social situations… The salmon was asked to determine what emotion the individual in the photo must have been experiencing."

Apparently, the deceased creature had some opinions, because they showed up in the brain scan. A dead fish bounced the meters!

Was this evidence of an aquatic afterlife, a miracle of paranormal proportions, a Frankenstein fish? Hardly.

As Bennett says in his published presentation, "Can we conclude from this data that the salmon is engaging in the perspective-taking task? Certainly not. What we can determine is that random noise… may yield spurious results if multiple comparisons are not controlled for."

Or to put it in less scientific terms, strange stuff can happen by accident, especially when you examine only a very small sample of data. During the salmon scan, environmental background noise randomly spiked the meters. It was just a coincidence that the spikes occurred during the questions. Regardless of what we can fancifully imagine, the salmon wasn’t "thinking" about the pictures.

Bennett published the study to demonstrate that people sometimes make incorrect judgments because they base decisions on statistical samples that are too small or poorly measured. His presentation was aimed at the neuroscience community, but it has gone viral in a global way. Statisticians all over the world have read and recognized the fundamental implications of his research.

In fact, the lesson of the dead salmon’s brain scan stretches all the way into the world of gambling and blackjack.

Players often see "trends," in patterns of wins and losses, and they incorrectly extrapolate significance from random luck. It’s human nature. We see pictures in the clouds. We hear the ocean in seashells. We look at stars and imagine constellations. Some patterns do have significance, but many trends are simply statistical noise.

For example, suppose we take a dead salmon and slap it into a blackjack seat at the Bellagio. Then we play five hands of blackjack. What’s the probability that the fish will play perfect basic strategy?

If you say 0%, then you’re wrong.

Actually, the fish has at worst 1% and at best, 18% chance of playing perfectly by doing nothing and standing every time. Or to put it another way, nearly one out of five dead fish will play five hands perfectly (depending on the dealer’s upcard). Does that mean a dead fish can play basic strategy? Clearly, no. Five perfect plays would be a coincidence.

Now let’s switch it around. Be honest… If you sit down at a blackjack table and lose five hands in a row, you might think something unusual was happening. Another two losses would cause some players to scratch their heads and leave the table, or maybe curse the dealer for bringing bad luck. However, seven consecutive losses isn’t so unusual. It happens about 1.2% of the time. Play long enough, and it becomes inevitable.

So random streaks of wins and losses generally don’t indicate predictable trends; they’re just coincidences, noises that spike the meters.

The great irony is that sometimes we miss or ignore real patterns and trends because we’re looking for imagined phenomena.

Can a pattern be repeated and tested? Yes. Counting cards is a good example. If you take tens out of the deck, the dealer is likely to win more hands. This is irrefutable math. In these situations, the past definitely affects the future. Therefore, if you follow the "trend" using a counting system, and you change bets accordingly, you’ll win more and lose less over time.

Other measurable easy-to-miss "phenomena" are the various rules for doubling. If you cannot double down on soft hands, or double after splits, you win less than you would otherwise.

These examples of trends may seem moronically obvious to experienced players, and they are obvious when you read this article in a calm moment away from the table. Nevertheless, bad streaks sometimes skew perceptions. It’s tough to concentrate on basic strategy and counting when you’re being hammered with multiple losses. It’s tempting to think, "Maybe there is a pattern in these results."

So okay… Setting aside the well-known conditions of counting and game rules, what would be a legitimate measurable trend?

Fifteen consecutive losses happens about 1 in 11,000 trials. If you see streaks like that, or longer, more than once or twice in 180 hours of play (about 45 to 90 typical sessions), then that’s a trend to study.

Likewise, winning 13 or less hands out of 60 (losing net 34 bets) happens about 1 in 15,000 trials.

If you have bad streaks like these regularly (repeatedly every session, or every third or fourth session), then something is wrong. If you’re following a sound strategy and repeatedly having bad luck like that, then it’s time to call the gaming commission because somebody is cheating.

But if you lose seven consecutive bets, or even ten, or you lose 20 out of 60 decisions (losing net 20 bets, happens about 2.8% of the time), then don’t sweat it. Stick to strategy and play the favorable games. Don’t try to read the tea leaves. Be patient and the streak will end. Don’t imagine that the dealer has a secret advantage beyond the edge calculated in basic strategy.

Really, his luck is only about as good as a dead fish.

**********

(c) copyright 2011 Basil Nestor