- I try to find things I can share that others aren't doing, and there are quite a few other sites doing the same thing
- One can easily figure out tournament winner probabilities by looking at the odds you could be betting on any given team to win
Plus watching game odds rise and fall as goals are scored and time marches on is just more interesting for me. All that said, I am going to post them because a couple of things have jumped out as interesting. First, the model is pretty bearish on Brazil. The betting markets have them at 24% chance to win, but my model only has them at 17% chance. Remember too, that the model is calibrated using actual betting odds from all the individual pool games. So this isn't me disagreeing with Vegas, so much as it's Vegas disagreeing with itself.
The same goes for my model's assessment of the Netherlands chances. Vegas has them at around 7-8%, I have them at close to double that.
Notes on methodology:
- The odds are based on a simulation of 10,000 World Cups
- Team ratings are calibrated using odds on all the remaining group play games, then adjusted for teams that qualify for the knockout stage using their simulated performance in group play
- Brazil was given a home field advantage for each game in the knockout stage
Your model has Brazil with only a 58% chance of winning its round of 16 game given that it survives the group? This is a flaw. The home field advantage alone, even if playing an equally strong team, would give them a stronger than 58% chance of winning a knockout game. Never mind that they may end up playing a much weaker team, like Chile. If they play Netherlands or Spain they will be 75/25 favorites.
ReplyDeleteLast 5 seasons in the Premier League: home team wins 47%, away team 27%, draw 26%.
ReplyDelete