If you have a 45% chance of winning, is it “yours to lose”?

Nate Silver gives Brazil a 45% chance of winning the World Cup, with only Argentina and Germany having more than a 10% chance. My gut feeling is that that’s a bit high, but I’m no expert.

What I find striking, though, is that the headline says it’s “Brazil’s to lose.” Huh? If we take Silver’s prediction seriously, Brazil has a 55% chance of NOT winning the World Cup. How does that make it “theirs to lose”? Personally, I’d have to think they had something like a 75% chance before I’d say it was “theirs to lose.” I could maybe see an argument for using this language if their chance is simply over 50%, although I wouldn’t use that characterization. But if I think they are probably NOT going to win the World Cup, there’s no way I’d say it’s “theirs to lose”!

This post is by Phil.

 

21 thoughts on “If you have a 45% chance of winning, is it “yours to lose”?

  1. In a literal sense, I believe that the Cup is Spain’s to lose, what with them actually holding the cup at the moment.

    That said – who here saw this number and then went and checked the betting odds and realized Nate Silver disagrees pretty strongly with the bookies? And then thought: crap, do I believe in Bayesian inference, or in back-alley conversations, whispers, game fixers*, witch doctors**, Messi, and the lack of international experience up front for Brazil?

    I still won’t take Brazil at 3:1 odds, despite the apparent insider info we’re getting from Silver’s simulations. Because I don’t think Neymar is up for it. And I don’t think Paulinho, or Oscar really have the mental steal either (Hulk? Well, Hulk smash, but that’s not enough).

    Conclusion: Just because Brazilian players have in fact been known to break the laws of physics*** doesn’t mean they can break the laws of probability. Silver, bookies, whoever – it’s still not “theirs to lose” in any sense.

    USA! USA! USA!

    * http://www.nytimes.com/2014/06/01/sports/soccer/fixed-matches-cast-shadow-over-world-cup.html?_r=0

    **http://www.theguardian.com/football/2014/jun/04/cristiano-ronaldo-portugal-ghana-witch-doctor

    ***http://www.youtube.com/watch?v=fOmqDOi4DH0

  2. I guess the reasoning is that Brazil has a much higher chance than any other team and thus it’s “Brazil’s to lose”. Intuitively I’d say it’s fine to say it (if I believe the model). I do agree, though, that it’s strange on the other hand as they do not have a chance higher than 50%. Maybe I’m just more or less unconsciously affected by my belief that Brazil will win the Cup anyway.

    • Daniel, how far would you take the “plurality of probability means it’s theirs to lose” principle? If they had a 10% chance, but that was still higher than any other single team’s, would it still be “theirs to lose”? I’m not being sarcastic, just asking about the use of this phrase. It doesn’t seem right to ME that if you think a team is unlikely to win, you’d still say it’s “theirs to lose”, but obviously some people think differently!

      • Phil, to be honest I’m not sure. They must at least have a clear advantage compared to all other teams and thus have a *much* higher chance than one would expect if all teams would all be equally likely to win (see also BadYodeler’s comment). I just doesn’t feel right to me to say one team out of 36 is unlikely to win if the chance is higher than 40 % – and that’s of course all I say here. I’m only trying to verbalize my intuitive acceptance of the “it’s theirs to lose” phrase. It’s something about “relative probability to win” which translates into “clear advantage compared to all other teams” which is probably seducing me to accept the “it’s theirs to lose”. Of course, if we take the model literally and just draw randomly from the posterior distribution, Brazil would still not win more often than not win, though.

  3. Re: “But if I think they are probably NOT going to win the World Cup”

    I don’t understand this. If there were only two teams, then yes 45% would mean that they would probably not win it. But since there are 32 teams the ‘prior probability’, i.e. the probability of winning if skill would be uniformly distributed across teams, would be 1/32 ~ 3%. And compared to that, 45% would imply a huge edge, no? Anyway, no team will ever have that kind of an edge in a world class competition, not even Brazil in Brazil.

    • According to the model, Brazil has a 45% chance of winning, which is to say that Not-Brazil has a 55% chance of winning, which is to say that Brazil is more likely to lose than win, which is to say that Brazil probably will not win.

      • But the question is, what does it mean that it’s “Theirs to lose”? Of course, it really doesn’t mean anything statistical, so this may actually be a good example we could use to define it…

        For any given match, Brazil is the favorite; Brazil simply has to win every match (notwithstanding an acceptable round robin loss) they’re expected to win and they’ll win the tournament. That could be a definition of “Theirs to lose”, because they have to be on the unlikely (< 50% chance of winning a tournament from the outset? If that’s the only situation you want to allow “Theirs to lose”, then that’s a different definition.

        • I’m not sure what you’re saying here.

          To me, saying the World Cup is “theirs to lose” implies that they are expected to win. (In fact, I would apply an even stronger standard, I wouldn’t use this phrase unless a loss would be surprising). If they have less than 50% of winning then they are not expected to win.

          If they’re heavily favored in every individual match, I might say “this match is theirs to lose” but I wouldn’t apply that to the whole tournament. Suppose you offer to give me a prize if I make 10 consecutive free throws. I’m favored to make every shot but it is extremely unlikely that I would make ten in a row. I wouldn’t say the prize is ‘mine to lose.’

          The fact that several of you think that something can be ‘yours to lose’ even though you are expected to lose, that’s interesting to me, and shows that even common phrases like this can mean different things to different people.

    • It’s not just about the relative strengths of the teams, but how competitive their group is. The US is in the group of death. Ecuador also has a ‘home time zone’ advantage.

  4. One point I do like is that Nate is making a concrete predition. No fuzziness. We will soon know if he was right or wrong.

    Its somewhat hard to get pundits and academics to make non fuzzy, unqualified predictions.

    • > We will soon know if he was right or wrong.

      And how will we know that?

      No matter the outcome, that seems like it would be a really difficult question to answer.

  5. Pingback: World Cup pseudo-science « Statistical Modeling, Causal Inference, and Social Science Statistical Modeling, Causal Inference, and Social Science

Leave a Reply to Chip Lynch Cancel reply

Your email address will not be published. Required fields are marked *