Being Contrarian

Have you ever played the guessing game where you are challenged to estimate how many jelly beans are in the jar? Closest to the actual number wins. I used to estimate how many jelly beans were in each row and then count how many layers there were. It’s no surprise I don’t remember ever winning one of these contests. Without fail, the range of individual guesses is large. Seems like everybody else isn’t very good at estimating the jelly beans either!

Well, there’s a curious phenomenon that occurs in games like the jar guessing game: while individual guesses are rubbish, the average of these guesses (provided they’re independent) is quite accurate!

There is wisdom of the crowd!

The best forecasters at the Good Judgment Project take advantage of the multiplicative effect of combining multiple, well-calibrated forecasts in serving consequential predictions to their corporate clients and even the Department of Defense. And it’s the reason my NFL algorithm aggregates various published NFL predictions to hone my number.

Let’s transition to the NFL landscape.

Hypothetically, if you were to ask a decent sample of people to predict the final score of a game, averaging their scores would result in a more accurate prediction. I presume this is part of the value add of the popular prediction apps like Fox’s Super6 and NBC’s Sports Predictor. In the Super6 app, for example, entrants are asked to select a winning side of the game and the final score margin range (e.g., “Cardinals win by 7-10 points”). By dangling a prize for getting the most selections correct, Fox is able to get thousands of (mostly) independent estimates, that is to crowdsource a final score distribution that, in principle, they could then use to inform that game’s line in their revenue-generating, Fox Bet app.

But there’s a common pitfall that casual bettors fall into when trying to gain the wisdom of the crowd. Wagering against the spread is a binary exercise. Yes, no. Go, no go. There’s usually no room for nuance (ignoring alternative lines and teasers). Bettors are asked to take a side. Win, loss. Over, under. Favorite, underdog. Some might say following the “wisdom” in these scenarios is to bet the side that is getting more action. Others might argue following the money is the wiser approach.

That’s not the approach I take though. I like to consider myself a contrarian. What does it mean to be contrarian? In plain terms, it is simply betting against the crowd. Leo Dworsky, the late fund manager who ran Fidelity’s Contrafund, observed that it was at the inflection points where the crowd had it wrong. He looked for “crescendos of euphoria and pessimism” where he could bet the other side. Each week in the NFL offers ample opportunity to apply the same philosophy.

One of the most successful against the spread signals I use each week is to discern the “sharp” side of each line. The sharp side, as I’ve defined it, is the side where the line has moved opposite the majority of public tickets. For example, if 75% of the public is taking the Packers but the line has moved from the opening number of Packers -7 to closing at -5.5, I can reasonably deduce that respected money is backing their opponent.

Another effective signal is to bet against the overzealous crowd outright. I’ve empirically found that a public ticket count ratio of 5:1 is the “crescendo of euphoria” to use Mr. Dworsky’s phrase. On the other end of the spectrum, I am training myself to be attracted to the pessimism surrounding the most malodorous of teams as these teams can be tough to put your money behind, but are rich with value.

Diamonds in the rough as the Cave of Wonders sand lion says in Aladdin.