Go KOREA !!!! Yeah !!!!

KOREA !! KOREA !!!ps. Sorry for another 'very alike' message~~ But... the Korean team has certainly shocked the world !!!!Hope we beat Spain today !!

It will be a wonderful weekend since England swept away!! ...I just couldn't help it. WARNING- to all members of these forums in London, please avoid to wear sort of green or Yellow clothing. Football fans sneaking out from the pubs with madness and could assault anything that can be connected to Brazil!

Thanks Aaron,The best odds available from various sources at the end of April allowed the following allocation:Team Price Bet------------------------France 4/1 21.38Argentina 9/2 19.43Italy 6 15.27Brazil 7.5 12.57Spain 10 9.72England 12 8.22Portugal 14 7.13Germany 16 6.29 ---- 100 ----each team therefore returning 6.88K.Using the 'mini Sharpe Ratio' calculation in Paul Wilmott's 'Optimality in Racing' wefind the mean return is 4.75 and the second moment 245.15, so SR = 0.3186. (In fact the World Cupis an almost an example of the problem set at the end of Paul's spreadsheet [by which I mean thecorrect allocation is to take the same profit on each outcome - not quite arbitrage because the oddsdo add to > 1]. See attachment - which I may have altered slightly with Sunday's updates in case thenumbers don't quite match up.) Using the second Gambler's Ruin approximation of Ethier/Khoshnevisan for a 5% chance of ruin we calculate bank_units_required = 126 and so the 100 stakes corresponds to 79.2% of the bank.Surprisingly three seeded teams depart the tournament in the group stage, leaving 11 outsidersavailable at the following best odds:Team Price Bet------------------------Paraguay 66 0.083Mexico 25 0.219USA 100 0.055Ireland 80 0.069Korea 66 0.083Denmark 33 0.166Belgium 100 0.055Sweden 22 0.249Senegal 40 0.137Japan 33 0.166Turkey 50 0.110 ----- 1.393 -----each team capable of returning 5.49K. Note that the 'sum of the best odds' is only 104% so it is betterwe think to take those rather than bet on the spreads.We could hedge for 1.393K and close our position now at 5.49K to the good. However we keepour faith in the major nations and hold out for a decimation of the underdogs at the firstelimination stage. Korea upset Italy but otherwise the seeds do well leaving the followingquarter final hedging option:Team Price Bet------------------------USA 22 0.266Korea 40 0.146Senegal 20 0.292Turkey 20 0.292 ----- 0.997 -----each team returning 5.88K.We could now close out for 5.88K. However if Spain keep their nerve we will arrive at theposition originally envisaged for the semi-final: 3 seeds and one underdog. We imagine a final hedge:Turkey or Senegal, 10/1, hedge 0.62, closing at 6.2K.Oh, oh. Korea 5 Spain 3: looks like we are now going to need two semi final hedges!An elimination tournament is surprisingly difficult to analyse and will need a full model of judgementprobabilities and market estimations, continuously updated, to work out with any degree of accuracy.This is one of the attractions of Dutching: as you said in a previous post a simple model works bestwhen the experts are floundering. Here we only the vague assessment 'most outsiders will lose, most favourites win',rather than attempt even the most cursory of models.----Soc.

You britons, . . . It was a wonderful game, the two Rs - Rivaldo and Ronaldinho - did wellGod save the spice girls !Pele

We fly the girls out and this happens! Apparently they'd packed enough till after the final game.Feigning injury became an art form in that game.

You britons, . . . It was a wonderful game, the two Rs - Rivaldo and Ronaldinho - did wellGod save the spice girls !Pele >>Actually I thought it was a very dull game, a disappointingly weak performance from England especially when they had the (rather unjustified) numerical advantage. Rivaldo again showing his acting ability - why does a great player have to do this? All three goals coming from bad mistakes - defender gifting the ball to Michael Owen, Beckham not wanting to get his tootsies hurt and David Seaman...oh well.

markfd, i agree with you but in the british perspectiveNext show will be against TurkeyPele

All three goals coming from bad mistakes - defender gifting the ball to Michael Owen, Beckham not wanting to get his tootsies hurt and David Seaman...oh well. >>Most goals come from mistakes.... The first one England scored against Denmark...and the other two if i remember well.... and the penalty against Argentina... and Campbell being isolated on the corner against Sweden...I'm sure if it had been England coming back from 1-0 and winning 2-1 despite - say - Beckham being unfairly sent off at the beginning of the 2nd half, you'd be saying that it was an amazing game and an extraordinary free kick by the genius Beckham... Rivaldo again showing his acting ability - why does a great player have to do this? >>I agree. And he got away with a ridiculous fine against Turkey. The young teams with no experience don't do this sort of cinema and they're being jusdged as naive. The big teams jnow all the tricks and they're being called mature , tactically awaree. As if it was a quality to know how to foul the referee...

Now I understand. With Germany in the top 8 it worked out okay.However, I think this was a idiotic bet (apologies to your friend). The odds table he was working with, converted to probabilities, added up to 1.12. The 0.12 is the bookie's profit, he sells $1.12 of claims for a $1 payout. Paying 0.94 for an 0.84 probability of getting the $1 (using the bookie's implied probabilities divided by 1.12) doesn't seem sensible. Even if he thinks the teams beyond the top eight have less chance than people think, all of them put together must have more than 0.06 probability.However, even with Spain's loss, he's not in terrible shape. He's lucky that Germany and Brazil do not play in the semifinals. But it still seems to me he has about a 0.15 chance of losing.

Now your friend can breathe a little easier. The chance of Turkey beating Brazil and then Germany is 0.05 by my calculation. At this point, I think the bet is about fair again.

Whoops. I redid my model adding in the last week of games, now I find Brazil has a 0.55 probability of beating Turkey and 0.45 of beating Germany. Germany has a 0.67 probability against Turkey. Overall, Germany has 0.60 probability of winning the Cup, Brazil has 0.25 and Turkey 0.15.The model is extremely accurate over the 61 games played to date. 16 ended up at their most likely score from the model, in 33 games one score was exact and the other was off by one goal. Most of the 12 deviant games are understandable, like Germany beating Saudi Arabia 8-0 instead of the predicted 5-0 or Mexico beating Ecuador 2-1 instead of the predicted 1-0. The only surprising results were Portugal beating Poland 4-0 (3-3 tie predicted), Spain and South Korea scoreless during play (2-1 win for Spain predicted), England tying Sweden 1-1 when it should have won 2-0 and the United States losing to Poland 3-1 when it should have won 4-2.

Aaron,'All of them put together must have more than 0.06 probability.'If everyone thought the bookmakers odds were always accurate then there would be no betting!'Now your friend can breathe a little easier.'He has relaxed after cashing out with 2 semifinal hedges.'The model is extremely accurate over the 61 games played to date.'A. Notwithstanding that it is backfitted, would it be possible to compute the chances of 0, 1 or 2 teams outside the top 8 favourites I nominated reaching the semi-final stage.B. Hope you described the model completely in your initial post - the payouts for correct scores are pretty good. I think you might be onto a winner if you appled it to UK league football! Can we all use it with appropriate acknowledgement :-)Thanks, as always.-----Soc.

Okay, I reran the World Cup playing all games according to my Poisson model with backfit parameters. I didn't simulate, I calculated the probability of each outcome (up to 12 goals per side). According to that there was 0.07 probability of having four of the eight top pre-tournament teams in the semi-finals, 0.14 of having three, 0.32 of having two, 0.32 of having one and 0.15 of having none.A more reasonable answer is to redo the computation assuming your friend had identified the eight top tournament performers. That is, a priori the favorite teams that left early probably had better chances of winning than their results show, and the lesser-regarded teams that hung around probably had worse chances of winning. This should eliminate the backfit bias. In that case the probabilities are 0.18 of four, 0.31 of three, 0.32 of two, 0.15 of one and 0.06 of zero.The actual answer is probably somewhere between the two. The first gives no weight to the pretournament expectations, the second regards it as certain.I think the real bad news in the tournament for your friend's strategy was not the disappointing showings of France, Agentina, Italy and Portugal. It was the depth of talent. Turkey, the US, Ireland, Costa Rica, Belgium, Russia, Mexico, South Africa, Poland, Nigeria, Senegal and South Korea posted the same average performance in my model as the top eight teams.In other words, the volatility was higher than your friend expected. He ended up doing better than average given that volatility. Perhaps that is due to some "championship" factor not found in goals scored; maybe the Germany's and Brazil's are more likely to win the Cup than the South Korea's and Turkey's; more than you would expect from game-by-game results. Or maybe he was lucky.

The friend was right. He just had to wait for the finals to be vindicated.I say that not just because Brazil and Germany made it to the finals, but because these two teams stood out among the field for the entire tournament. Germany has the best defense and second-best offense, Brazil the best offense and the eighth best defense, no team was close to these two.Maybe it's true that a good outsider can play as well or better than the average favorite, but the top favorites will rise in the end.