FIFA World Cup Predictor 2010

[mouser]

Thought I would start a new thread to discuss issues in the FIFA World Cup Predictor 2010 site/page.

We need experts to create some pairwise estimates of team matchup outcomes before the tool can be used for anything useful..

But If you want to get a feel for how this tool for calculating betting advice differs from a tool to predict the winner -- select the "Uniform Ratings" from expert ratings list, click Load, then Start the simulation.

What you should see over time is now every team equally likely to win the tournament.  Now look at the Expected Profit columns.  Red entries are expected profitable bets, the higher the better.  Note that Honduras is the highest expected payoff -- which makes sense since it's the team that has the highest payoff odds and we are assuming for this experiment that all teams are equally likely to win.

[mouser]

Even if you're not interested in gambling or the world cup, I'm hoping some of you may find the discussion of markov chain monte carlo statistics/probability interesting.

There is a page now which can be used to help you get a feel for how the different probability function parameters actually relate to game scores, and i've extended it to show actual sample simulated game scores, so you can get a better feel for how the means and standard deviation parameters map to scores:

https://www.donation...m/wcp/experthelp.php

[mouser]

Just fixed a bug in the new 2010 version where it wasn't recognizing that you were a forum member.

[anodyne]

Have entered my first thoughts on the scores.  Spain ends up doing rather better than the odds-makers think & England worse, though I'm not surprised at the latter; I suspect the reason England is 6.5:1 is the odds are from British bookies who have received a lot of sentimental backing ...

[anodyne]

Some more thoughts ... it would be quite interesting to see what the predictions are for the finalists & the semi-finalists as well as for the eventual winner. How easy would it be to add that?

[mouser]

awesome work anodyne, can't wait to check out your scores.

because of the nature of the simulation, it is certainly calculating the predicted finalists, semi-finalists, and winners as it simulates millions of games, the only question would be how to display this info.. are we talking about just reporting the most likely finalists, semi-finalists, winners, or listing probablities of each team for each of these categories, or trying to look at betting payoffs for these bets?

[anodyne]

because of the nature of the simulation, it is certainly calculating the predicted finalists, semi-finalists, and winners as it simulates millions of games, the only question would be how to display this info.. are we talking about just reporting the most likely finalists, semi-finalists, winners, or listing probablities of each team for each of these categories, or trying to look at betting payoffs for these bets?

-mouser (February 02, 2010, 06:01 PM)

The betting payoff is probably least interesting to me. Perhaps 3 columns in the ongoing display: probability of winning (as per now), probability of reaching the final, probability of reaching the semi-final. The probabilities in the second column of course will add up to 200%, in the third to 400%. But it's just a thought ...

[ricardosf]

What is your prediction on the first opening game of WCP 2010, Shouth Africa vs Mexico? What is your scores prediction?

[mouser]

the simulator doesn't make per-game predictions.. in fact it relies on a user to tell it how every possible per-game match up could result, and from that it calculates how the tournament would end.

[mouser]

Nate Silver (political statistician) has done some recent work on predicting world cup 2010 results:
http://www.fivethirt...-simulated-odds.html

[ljbirns]

Ok  now we are into the round of 16.  Is there any way to know how the predictions have worked so far ?

I had  Italy @  8 %  and France at 3 %  both are done..
I'm OK with Spain  27 %  England 18 %  Brazil 14 % Germany 9 % Netherlands 6 %  Argentina @ 5 % and South Korea 4 %

I know the object was to make a good bet on a sleeper. 
Just curious
Go  USA  !!!!