The John De Lange Sophisticated Prediction Contest - WC 2010
in [World Football]
|
From: milivella on 11 Jun 2010 14:42 futbolmetrix: > Deadline for entries: 15 minutes before kickoff of South Africa - > Mexico. > > Enjoy (and let me know if there are any bugs), and let the most > sophisticated man or woman win! Daniele, would it be easy for you to compute and post the average human entry and the average "computer" (Elo, Castrol, Chance de Gol...) entry? Would it be against the game's ethos? (you wouldn't disclose the single entries, but one could guess where his/her own prediction diverges a lot, and so the results that he/she wants to happen or not to happen...) -- Cheers milivella
From: futbolmetrix on 11 Jun 2010 19:04 On Jun 11, 3:26 pm, Alessandro Riolo <alessandro.ri...(a)gmail.com> wrote: > > Me too, yesterday, and my wife as well, yesterday too. > > > Please confirm reception. > I got all your entries. Here is the final list of entrants (in strict alphabetical order) 7h(a)ch Agnostic Alessandro Riolo Anders T AQB Castrol Football Chance de Gol CHRIS HORYMSKI ELO Futbolmetrix Jesus Petry Jim Goloboy JK Joachim Parsch juanvazquez Jussi Uosukainen Lléo Luiz Mello MH Miguel milivella Myk Olcay Kara Philipp Lucas Voros McCracken Werner Pichler Also, following milivella's request, I have calculated the entries for two additional players: Collective_Simulations (AQB,ELO,Castrol, ChanceDeGol, Voros McCracken) Collective_Humans (everybody else). Collective_Simulations' entry: P_R1 P_R16 P_QF P_SF P_LF P_WF South Africa .7097 .1722 .0733 .0312 .0078 .0057 France .4312 .2846 .1478 .0853 .0285 .0226 Uruguay .4188 .2729 .1587 .0894 .0346 .0255 Mexico .4403 .2399 .1632 .0966 .0326 .0274 Argentina .1921 .3096 .2048 .1589 .0686 .0660 Greece .6268 .2320 .0903 .0373 .0085 .0051 Nigeria .6284 .2289 .0917 .0355 .0110 .0046 South Korea .5527 .2599 .1151 .0534 .0124 .0066 England .0914 .2998 .2279 .1840 .0990 .0979 Slovenia .6653 .2341 .0660 .0282 .0043 .0021 Algeria .8470 .1186 .0259 .0079 .0004 .0003 United States .3962 .3471 .1435 .0778 .0206 .0148 Germany .2695 .2918 .1983 .1307 .0600 .0496 Serbia .4214 .2872 .1441 .0918 .0318 .0237 Ghana .7402 .1731 .0545 .0241 .0041 .0039 Australia .5688 .2482 .1015 .0558 .0167 .0090 Netherlands .1374 .2190 .3100 .1342 .1018 .0976 Denmark .5232 .2379 .1684 .0397 .0181 .0126 Cameroon .6658 .1920 .1106 .0199 .0069 .0048 Japan .6736 .1884 .1060 .0205 .0057 .0059 Italy .1816 .3709 .2728 .0842 .0499 .0405 Slovakia .6228 .2791 .0830 .0115 .0020 .0016 Paraguay .3460 .3899 .1906 .0403 .0193 .0139 New Zealand .8496 .1227 .0244 .0029 .0002 .0003 Brazil .0779 .2600 .2053 .1427 .1295 .1847 Portugal .4275 .3594 .1001 .0570 .0306 .0254 Cote d'Ivoire .5667 .2921 .0734 .0386 .0185 .0107 Korea DPR .9279 .0625 .0063 .0008 .0021 .0004 Spain .0717 .2692 .1752 .1415 .1368 .2056 Switzerland .6904 .2367 .0418 .0179 .0071 .0062 Chile .5026 .3146 .0886 .0462 .0263 .0217 Honduras .7353 .2056 .0370 .0143 .0042 .0036 Collective_Humans' entry: South Africa .5403 .3258 .0945 .0254 .0082 .0059 France .3537 .2309 .2057 .1254 .0472 .0372 Uruguay .5843 .2469 .1179 .0319 .0122 .0069 Mexico .5217 .2369 .1440 .0669 .0177 .0129 Argentina .1106 .2162 .2431 .2035 .1201 .1067 Greece .6307 .2469 .0892 .0166 .0089 .0077 Nigeria .5678 .2836 .0955 .0367 .0099 .0066 South Korea .6910 .2129 .0670 .0196 .0054 .0042 England .1068 .2434 .2521 .2041 .0894 .1043 Slovenia .6510 .2594 .0679 .0129 .0049 .0041 Algeria .7540 .1782 .0515 .0080 .0042 .0042 United States .4883 .3483 .1130 .0325 .0091 .0089 Germany .2138 .2112 .2401 .1572 .0949 .0828 Serbia .4092 .3237 .1667 .0671 .0202 .0132 Ghana .6693 .2263 .0632 .0266 .0079 .0067 Australia .7077 .2097 .0557 .0171 .0049 .0049 Netherlands .1146 .2304 .2841 .1772 .1024 .0914 Denmark .5121 .2861 .1457 .0396 .0083 .0083 Cameroon .5337 .2977 .1062 .0424 .0112 .0089 Japan .8397 .1117 .0317 .0096 .0038 .0036 Italy .1463 .2892 .3189 .1292 .0608 .0557 Slovakia .4835 .3618 .1211 .0217 .0062 .0059 Paraguay .4451 .3761 .1314 .0302 .0084 .0089 New Zealand .9252 .0473 .0133 .0069 .0040 .0034 Brazil .1071 .2265 .2139 .1529 .1533 .1462 Portugal .3749 .3779 .1255 .0591 .0320 .0307 Cote d'Ivoire .5838 .2678 .0810 .0411 .0151 .0113 Korea DPR .9342 .0417 .0116 .0059 .0034 .0033 Spain .0761 .2313 .2065 .1873 .1096 .1893 Switzerland .6432 .2822 .0471 .0159 .0060 .0057 Chile .4993 .3923 .0693 .0237 .0078 .0075 Honduras .7815 .1803 .0263 .0061 .0030 .0029 Anybody wants to volunteer to find the largest discrepancies? D
From: milivella on 12 Jun 2010 02:56 futbolmetrix: > I got all your entries. Here is the final list of entrants (in strict > alphabetical order) Thanks for organizing! > Also, following milivella's request, I have calculated the entries for > two additional players: > > Collective_Simulations (AQB,ELO,Castrol, ChanceDeGol, Voros McCracken) > Collective_Humans (everybody else). Thanks a lot! > Anybody wants to volunteer to find the largest discrepancies? I, of course. Absolute differences (chances_according_to_humans minus chances_according_to_simulations): South Africa: - group: -0.17 - 2nd round: +0.15 Uruguay: - group: +0.17 [I.e. we the humans think that South Africa is less likely to go out in the group, and we put the chances that we take from SA to Uruguay; but all our hope in SA is limited to the group stage, since we agree with simulations about the chances they have to reach the quarters.] South Korea: - group: +0.14 Australia: - group: +0.14 Cameroon: - group: -0.13 - 2nd round: +0.11 Japan: - group: +0.17 [I.e. we tend to find Asian teams as "minnows" more than simulations do.] Slovakia: - group: -0.14 Largest differences in later rounds: - quarters: France +0.06, Japan -0.08 - semifinals: Spain +0.05, Uruguay -0.06 - losing finalist: Argentina +0.05, Spain -0.03 - winner: Argentina +0.04, Brazil -0.04 But, since the sophisticated score is given by natural logarithms, the following computation is maybe more meaningful: LN (chances according to humans) - LN (chances according to simulations) Please tell me if I'm wrong. Anyway, here are the largest differences: Uruguay: - semifinal: -1.03 - losing finalist: -1.04 - winner: -1.31 South Korea: - semifinal: -1.00 Algeria: - losing finalist: +2.35 - winner: +2.64 [who predicted Algeria in final for sure?! ;) ] Australia: - semifinal: -1.18 - losing finalist: -1.23 Japan: - quarters: -1.21 Slovakia: - losing finalist: +1.13 - winner: +1.31 New Zealand: - losing finalist: +3.00 - winner: +2.43 [oh yes, an Algeria vs. New Zealand final...] Korea DPR: - semifinal: +2.00 - winner: +2.11 Highest differences in first two rounds: - groups: Uruguay +0.33 (Brazil +0.31), Argentina -0.55 - round of 16: South Africa +0.64, New Zealand -0.95 Summary: we don't like Uruguay and Asian teams (except North Korea), we like Slovakia. (or: simulations like Uruguay etc.) Even if we know that Argentina is managed by Maradona and simulations strictly speaking don't, we give more chances to La Albiceleste. -- Cheers milivella
From: milivella on 12 Jun 2010 03:40 futbolmetrix: > Collective_Humans' entry: > > South Africa .5403 .3258 .0945 .0254 .0082 .0059 > France .3537 .2309 .2057 .1254 .0472 .0372 > Uruguay .5843 .2469 .1179 .0319 .0122 .0069 > Mexico .5217 .2369 .1440 .0669 .0177 .0129 > > Argentina .1106 .2162 .2431 .2035 .1201 .1067 > Greece .6307 .2469 .0892 .0166 .0089 .0077 > Nigeria .5678 .2836 .0955 .0367 .0099 .0066 > South Korea .6910 .2129 .0670 .0196 .0054 .0042 > > England .1068 .2434 .2521 .2041 .0894 .1043 > Slovenia .6510 .2594 .0679 .0129 .0049 .0041 > Algeria .7540 .1782 .0515 .0080 .0042 .0042 > United States .4883 .3483 .1130 .0325 .0091 .0089 > > Germany .2138 .2112 .2401 .1572 .0949 .0828 > Serbia .4092 .3237 .1667 .0671 .0202 .0132 > Ghana .6693 .2263 .0632 .0266 .0079 .0067 > Australia .7077 .2097 .0557 .0171 .0049 .0049 > > Netherlands .1146 .2304 .2841 .1772 .1024 .0914 > Denmark .5121 .2861 .1457 .0396 .0083 .0083 > Cameroon .5337 .2977 .1062 .0424 .0112 .0089 > Japan .8397 .1117 .0317 .0096 .0038 .0036 > > Italy .1463 .2892 .3189 .1292 .0608 .0557 > Slovakia .4835 .3618 .1211 .0217 .0062 .0059 > Paraguay .4451 .3761 .1314 .0302 .0084 .0089 > New Zealand .9252 .0473 .0133 .0069 .0040 .0034 > > Brazil .1071 .2265 .2139 .1529 .1533 .1462 > Portugal .3749 .3779 .1255 .0591 .0320 .0307 > Cote d'Ivoire .5838 .2678 .0810 .0411 .0151 .0113 > Korea DPR .9342 .0417 .0116 .0059 .0034 .0033 > > Spain .0761 .2313 .2065 .1873 .1096 .1893 > Switzerland .6432 .2822 .0471 .0159 .0060 .0057 > Chile .4993 .3923 .0693 .0237 .0078 .0075 > Honduras .7815 .1803 .0263 .0061 .0030 .0029 Another experiment that you can run using these data is the answering the following question: what picks should have RSS collective mind have done in the unsophisticated contest? http://groups.google.com/group/rec.sport.soccer/browse_thread/thread/2a7e7b2b20d96f69/ Here is what I did in order to answer, using Spain as an example: 1. How many chances do they have (according to RSS mind) to go out at the group stage? 0.0761 2. How many points do you gain/lose if you have picked them and they go out at the group stage? -3 3. Multiply: 0.0761 * -3 = -0.2283 4. Repeat the first 3 steps for each round: 2nd round, quarters, etc. 5. Sum the values you have: -0.2283 -0.4626 -0.2065 +0.1873 +0.2192 +0.5679 = +0.0770. This is how many points you are expected to gain/ lose picking Spain in the unsophisticated contest, since if they'll play 10,000 WCs, 761 times Spain will go home at the group stage and you'll lose 3 points, 2,313 times they'll lose in the 2nd round and you'll lose 2 points, etc., so you'll get a total of 770 points. Now, 12 NTs have an expected positive value: 0.9998 Argentina (3rd most picked in the unsophisticated contest) 0.5960 Serbia (2nd) 0.5316 Germany (1st) 0.4007 Italy (5th) 0.3392 Mexico (5th) 0.2589 Denmark (5th) 0.2399 Slovakia 0.1929 Cameroon 0.1714 Chile (5th) 0.1130 South Africa 0.0895 Nigeria 0.0770 Spain (3rd) (RSS collective mind's picks are pretty consistent in the two contests, even if one is sophisticated and the other one unsophisticated!) --- We can also check what's the most probable last stage reached by the NTs involved in my bet against the rest of world http://groups.google.com/group/rec.sport.soccer/browse_thread/thread/b14d412382b63e48/ (an asterisk marks the predictions that overlap with mine) Brazil round of 16 Spain round of 16 * England round of 16 * Argentina quarters Germany quarters France group * Italy quarters Portugal round of 16 Serbia group * Côte d'Ivoire group * Chile group * Cameroon group * Ghana group * So RSS collective mind would have declined my offer. -- Cheers milivella
From: juanvazquez on 12 Jun 2010 04:37
On 11 jun, 19:04, futbolmetrix <daniele.paser...(a)gmail.com> wrote: > > Also, following milivella's request, I have calculated the entries for > two additional players: > > > Collective_Simulations' entry: > > P_R1 P_R16 P_QF P_SF P_LF P_WF > South Africa .7097 .1722 .0733 .0312 .0078 .0057 > France .4312 .2846 .1478 .0853 .0285 .0226 > Uruguay .4188 .2729 .1587 .0894 .0346 .0255 > Mexico .4403 .2399 .1632 .0966 .0326 .0274 > > Argentina .1921 .3096 .2048 .1589 .0686 .0660 > Greece .6268 .2320 .0903 .0373 .0085 .0051 > Nigeria .6284 .2289 .0917 .0355 .0110 .0046 > South Korea .5527 .2599 .1151 .0534 .0124 .0066 > > England .0914 .2998 .2279 .1840 .0990 .0979 > Slovenia .6653 .2341 .0660 .0282 .0043 .0021 > Algeria .8470 .1186 .0259 .0079 .0004 .0003 > United States .3962 .3471 .1435 .0778 .0206 .0148 > > Germany .2695 .2918 .1983 .1307 .0600 .0496 > Serbia .4214 .2872 .1441 .0918 .0318 .0237 > Ghana .7402 .1731 .0545 .0241 .0041 .0039 > Australia .5688 .2482 .1015 .0558 .0167 .0090 > > Netherlands .1374 .2190 .3100 .1342 .1018 .0976 > Denmark .5232 .2379 .1684 .0397 .0181 .0126 > Cameroon .6658 .1920 .1106 .0199 .0069 .0048 > Japan .6736 .1884 .1060 .0205 .0057 .0059 > > Italy .1816 .3709 .2728 .0842 .0499 .0405 > Slovakia .6228 .2791 .0830 .0115 .0020 .0016 > Paraguay .3460 .3899 .1906 .0403 .0193 .0139 > New Zealand .8496 .1227 .0244 .0029 .0002 .0003 > > Brazil .0779 .2600 .2053 .1427 .1295 .1847 > Portugal .4275 .3594 .1001 .0570 .0306 .0254 > Cote d'Ivoire .5667 .2921 .0734 .0386 .0185 .0107 > Korea DPR .9279 .0625 .0063 .0008 .0021 .0004 > > Spain .0717 .2692 .1752 .1415 .1368 .2056 > Switzerland .6904 .2367 .0418 .0179 .0071 .0062 > Chile .5026 .3146 .0886 .0462 .0263 .0217 > Honduras .7353 .2056 .0370 .0143 .0042 .0036 A question out of curiosity (I have my neuron a bit lazy today and it's 4 am): Do these figures have internal consistency ?, that is, if you fill them into the sheet they make sense and all the controls are right? How do you translate say ELO into the sheet? Do you have a "system"? Can ELO (or any other simulation) be translated with different sets of numbers? Excuse my ignorance. I usually have difficulties translating my predictions into the sheet and always have to compromise several numbers, which are not dictated by my wishes, but by the need to make the prediction consistent. JV |