Here’s the score on Super Bowl markets and myths
The Super Bowl might be done for another year, but there’s still time to consider the Super Bowl Indicator (SBI). The SBI began in 1978, when the sportswriter Leonard Koppett wrote a column showing that historical correlations are meaningless.
He pointed out that for all 11 of the Super Bowls played until then, if the winning team was from the old National Football League (before the merger with the American Football League in 1970), the stock market went up for the year; if the winning team was from the AFL, the stock market went down.
The supposedly meaningless indicator worked for the next 21 years, until 1999, when the old AFL Denver Broncos won the Super Bowl, and the stock market rose 20 per cent. It has failed six times in the 40 years since it was first identified. Over the years, myths have grown up around the SBI:
Myth number 1: Data mining
The SBI is often used the way Koppett did in 1978, to show that if you examine enough pairs of variables, some will be highly correlated without causal connection.
While that is true, the SBI doesn’t illustrate it because it continued working after it had been identified.
This is “out of sample performance.” The statistical evidence for the SBI is better than most published findings in the social sciences.
Myth number 2: The failure in 1999 and subsequent years disprove the SBI
In the past 25 years there has been convincing research that sporting outcomes and other seemingly irrelevant factors like whether it’s sunny in New York, affect markets.
However, these are on the order of a country’s stock market falling 0.6 per cent the day after its team is eliminated from the Fifa World Cup.
The SBI effect is an order of magnitude larger. Implausible as it is to see why global markets would be affected by which U.S. city prevailed in one football game, it’s harder to credit a SBI that worked 100 per cent of the time.
That would have to be magic, or some equally fundamental reinterpretation of reality. So the 85 per cent success rate of the SBI since 1979 is easier to accept as a causal effect than a 100 per cent success rate would be.
Myth number 3: The probability of the SBI working by chance is microscopic
One common calculation is if you treat each of the 40 years since identification as a fair coin flip: the chance of getting 34 or more heads out of 40 flips is about 1 in 250,000.
But the stock market goes up about two years out of three, and a team from the old NFL wins the Super Bowl about two years out of three.
Given that, the chance of the SBI occurring by chance is about 1 in 8,000. It’s still an unlikely occurrence, but closer to the realm of everyday events than is sometimes asserted.
Myth number 4: Even if the SBI worked, it wouldn’t help investors
In the 27 years an old-NFL team won the Super Bowl, the S&P 500 Index was up an average of 1 per cent from the beginning of the year to the Super Bowl, and up 11 per cent for the remainder of the year. In the 13 other years, the S&P 500 was down an average of 1 per cent before the game and up only 2 per cent for the rest of the year.
If you had invested in the S&P 500 each of the last 40 years in which an old-NFL team won the Super Bowl, you would have 33 times your initial investment. If you had invested in the other years, you would have broken even.
So the SBI, even if it worked, wouldn’t protect against losses in the long run, but it would allow you to earn bond returns in years when the stock market is net flat.
What non-myths should you learn from the SBI? Correlation isn’t causation, but correlation can continue for a long time. Don’t seize on a single observation that matches your preconceptions to ignore other data.
Be wary of claims that something is highly improbable. Don’t underestimate the investment value of quantitative patterns. Don’t manage your investments with the SBI, but do use it to sharpen your appreciation for statistics.