How to Predict the World Cup

With the World Cup in Brazil only a day away, the predictions are pouring in from gamblers, analysts, and even investment banks. All of them are trying to guess who will advance in the tournament, as well as who will win it all. But which model is best?


It’s a question that can’t be answered in the course of a single World Cup. Like any individual bettor, a statistical model can get lucky. Just because it predicts the winners correctly this year doesn’t mean it will next year, or indeed ever again. Rather, a good statistical model picks a high proportion of results year after year using the same formula.

Unfortunately, the World Cup only comes along every four years, and its format isn’t always the same. As a result, it can take decades to determine the quality of a prediction model. And even then, the model may be rendered useless for the tournaments to come.

In this situation, what’s a publicity-hungry forecaster to do? Well, here’s a win-win strategy that the most visible predictors may in fact be using. First you have to know who the favorite is, and the bookmakers are virtually unanimous that it’s Brazil. Next you assign a probability of between 45 and 49 percent to Brazil winning the tournament. This way no team has a greater than 50 percent chance of winning, but Brazil is the clear favorite in the “model”.

If Brazil comes through, the statistically naive among the public will say, “The model was right!” If Brazil doesn’t win, you’ll say, “Well, the chance of Brazil losing was greater than 50 percent,” and the public will nod in agreement. Either way, the model looks like it was correct.

I’m not saying that the folks at FiveThirtyEight and Goldman Sachs simply made up their numbers, but at some point they had to decide whether their models seemed reasonable or needed more tweaks. The “seems reasonable” criterion probably had to do with their prior beliefs about the competition, but it may also have depended on the risks they would have taken by going against the consensus.

The easiest way to reduce risk, as my strategy above shows, is to exploit the public’s ignorance. Most people don’t necessarily understand that one tournament is not enough to judge a model. So if you pick an outsider to win, then the dispersion of possible outcomes is huge: you’re a hero, or a goat. But if you go with the safe play, then you’re guaranteed to confirm your wisdom -- at least in the eyes of the unwise.

Image credit: Shutterstock

​There are two kinds of failure – but only one is honorable

Malcolm Gladwell teaches "Get over yourself and get to work" for Big Think Edge.

Big Think Edge
  • Learn to recognize failure and know the big difference between panicking and choking.
  • At Big Think Edge, Malcolm Gladwell teaches how to check your inner critic and get clear on what failure is.
  • Subscribe to Big Think Edge before we launch on March 30 to get 20% off monthly and annual memberships.
Keep reading Show less

Saying no is hard. These communication tips make it easy.

You can say 'no' to things, and you should. Do it like this.

Videos
  • Give yourself permission to say "no" to things. Saying yes to everything is a fast way to burn out.
  • Learn to say no in a way that keeps the door of opportunity open: No should never be a one-word answer. Say "No, but I could do this instead," or, "No, but let me connect you to someone who can help."
  • If you really want to say yes but can't manage another commitment, try qualifiers like "yes, if," or "yes, after."
Keep reading Show less

Apparently even NASA is wrong about which planet is closest to Earth

Three scientists publish a paper proving that Mercury, not Venus, is the closest planet to Earth.

Strange Maps
  • Earth is the third planet from the Sun, so our closest neighbor must be planet two or four, right?
  • Wrong! Neither Venus nor Mars is the right answer.
  • Three scientists ran the numbers. In this YouTube video, one of them explains why our nearest neighbor is... Mercury!
Keep reading Show less

Why is 18 the age of adulthood if the brain can take 30 years to mature?

Neuroscience research suggests it might be time to rethink our ideas about when exactly a child becomes an adult.

Mind & Brain
  • Research suggests that most human brains take about 25 years to develop, though these rates can vary among men and women, and among individuals.
  • Although the human brain matures in size during adolescence, important developments within the prefrontal cortex and other regions still take pace well into one's 20s.
  • The findings raise complex ethical questions about the way our criminal justice systems punishes criminals in their late teens and early 20s.
Keep reading Show less