Who will advance through the knockout stages of the World Cup, and go all the win to be crowned 2014 FIFA World Cup champions?
The previous article analyzed the group stages of the 2014 World Cup. Using a computer simulation that generated estimated goal amounts based on each team’s Elo Rating (ER) as of May 18, 2014, the outcomes of each match were determined by random numbers from a Poisson distribution. The tournament was simulated half a million times.
The value of the simulation is not determining what teams are most likely to advance. The use of seeding in World Cup groups, as well as common sense, reveals that much. What the simulations provide are percentage likelihoods of each outcome. If you want to know how likely it is that Brazil will finish fourth in their group (0.11% without home field advantage (HFA), then a simulation is the best way to do so.
Of course, simulations reflect the assumptions of the author. This study has two main assumptions:
1. That ER from http://www.eloratings.net is a valid measure of the relative strength of each team
2. That the estimated goals algorithm used to feed the Poisson generator is reasonable.
Home Field Advantage
As described in the previous article, whether or not to include home field advantage (HFA) for Brazil was a difficult decision. After looking at each set of results and the dramatically different outcomes they predict, it was decided to use the results with HFA included. Since only Brazil benefits from this inclusion, the obvious difference is that including HFA increases Brazil’s chances of winning the 2014 World Cup.
The Round of 16
The sixteen teams advancing to the elimination rounds of the tournament do not have an equal chance to win the tournament. If strong groups cluster together in the original draw, then the teams advancing from the group stages will have more difficult Round of 16 (R16) matches. Nowhere is this more evident than in Group A and Group B.
With HFA included, Brazil wins their group 96% of the time. This means that either Spain or the Netherlands will likely face Brazil in the R16. 41% of the time this will be the Netherlands, 25% of the time it will be Spain. This difficult round is reflected in their odds of advancing to the quarterfinals. Spain fails to make the quarterfinals 27% of the time. The Netherlands fail to advance from the R16 63% of the time they advance from the group stages. It is amazing that the team ranked fifth in the world by ER has only a 24% chance overall of advancing to the quarterfinals. It is also worth mentioning that Chile has only a 10% chance of making the quarterfinals despite being ranked tenth in the world. Surprisingly, Nigeria has a better chance of making the quarterfinals than Chile.
There are three teams with very favorable draws. Most obvious is Germany. Although in the second strongest group, they are still forecast to dominate and finish first. They then play the second best team in the worst group of the tournament. This will be either Russia or Belgium. Germany has an 84% chance to advance to the quarterfinals, just barely behind Brazil’s percentage. Portugal also benefits from the group match up and has a 46% chance to reach the quarterfinals.
The winner of Group D has a nice advantage. Colombia will probably win Group C and the second place team in Group C will be somewhat weak. England takes this advantage and turns in a 43% chance of quarterfinal action. Uruguay and Italy, after the brutal competition in Group D, also have a bit of a boost in the round of 16 for the same reason as England.
France, after their fortunate Group stage draw, also get a fortunate R16 draw. They will likely play the second place team in Group F, one of Bosnia, Nigeria, or Iran. France has a 40% chance to advance. Not bad for a team with an 1872 ER.
There are a couple of other slight oddities. If the USA can survive their group stage, which is unlikely, they do have a reasonable chance of passing the R16 (20%). Forgotten in Group A is Croatia, with a decent chance of advancing to the R16. However, if they get there, then Spain or Netherlands greets them with a 6% chance of advancement. South Korea also over-performs and advances to the quarterfinals 5% of the time
Groups A and B have an advantage over Groups C and D in the quarterfinals. The most frequent pairings are Brazil v Colombia and Spain v England. It is not surprising that Brazil’s odds of advancing barely decline between the R16 and the quarter finals, from 85% to 80%. Four out of five times Brazil is expected to make the semifinals! Spain also does well, advancing 54% of the time. It looks as though England’s usual tournament lament of making the knockout phase only to lose in the quarterfinals will come true again. There does not appear to be any easy path through the quarterfinals this tournament. England makes the semifinals only 10% of the time.
The other groups are more mixed head to head. The most frequent pairings are: France v Germany and Argentina v Portugal.
What is most striking are how few oddities appear in the quarterfinals compared to the R16. The attenuation of poorer teams is dramatic as the tournament progresses. On the one hand, this is a good thing. After all, the tournament is supposed to identify the best team in the world. On the other hand, it is fun watching a surprise run by a poorer quality team. It occurs in the simulations less frequently than expected. Only eight teams have a greater than 10% chance of making the semifinals. Brazil (80%), Germany (69%), Spain (54%), Argentina (47%), Portugal (24%), Netherlands (16%), France (14%), England (10%).
The largest discrepancies between odds of advancing to the semifinals and team strength are:
France: Over-performs because of advantages accrued in group and R16
Netherlands: Under-performs because of Brazil in R16
Portugal: Over-performs because of pairing with Group H in R16
The USA has a 7% chance to make the semifinals, slightly higher than Belgium!
Only four teams have greater than 10% odds of advancing to the final. Such is the disparity in team quality as the tournament progresses that all four of those teams are at 20% or greater odds of advancing. They are Brazil (67%), Germany (45.25%), Spain (26.53%) and Argentina (20%). Spain is surprisingly low. The difficult group for Spain and tough potential R16 matches weighs heavy on their odds. Spain is the most likely favorite to surprise to the downside in this World Cup. Of course, most teams would love a one in four chance of playing in the final, so perhaps it is not that dire. It is also worth mentioning the pure domination Brazil has in the simulation when playing with historical levels of HFA. Two out of three times they play in the final despite having a difficult R16 match.
The second tier of teams, in terms advancing to the final, contains Portugal (8%), Netherlands (6%), France (5%), England (3%), Colombia (3%) and Uruguay (3%).
There is no surprise here except the extent of the domination. Brazil wins 55% of the time. This result prompted a thorough check of the simulation to make sure everything was working properly. It was. However, a fair amount of this domination comes from HFA. Take that out and Brazil wins 31% of the time. This is much closer to the betting lines.
The winning percentages run like this: Brazil (55%), Spain (17%), Germany (14%) and Argentina (6%). The current betting lines from skybet are: Brazil (3/1), Argentina (9/2), Germany (11/2) and Spain (11/2). It looks as though the professional lines are giving Argentina a HFA advantage.
All other nations have a very small chance of winning. England is (0.87%, 25/1) and the USA is (0.16%, 250/1). Belgium is still the dark horse in the professional gambling world. Odds at 18/1 but the simulation gives a 0.12% chance of winning!
The most frequent match-ups in the finals are:
Brazil v Spain – 43% of all finals – Brazil wins 66%
Brazil v Argentina – 22% of all finals – Brazil wins 80%
Germany v Spain – 19% of all finals – Spain wins 54%
Brazil v Germany – 13% of all finals – Brazil wins 68%
Germany v Argentina – 12 % of all finals – Germany wins 64%
Brazil v Netherlands – 7% of all finals – Brazil wins 83%
Brazil v Portugal – 6% of all finals – Brazil wins 88%
The third place match is usually an afterthought. However, in the interest of full disclosure, here are the odds of winning third place:
Germany – 33%
Argentina – 14%
Brazil – 11%
Spain – 11%
Over half a million simulations, the teams least likely to win the World Cup are:
Algeria – 0 wins
Cameroon – 1 Win
Australia – 2 Wins
Honduras – 3 Wins
Ghana – 6 Wins
Algeria fans, please adjust your expectations accordingly.
Run the World Cup enough and some funny things happen just through random luck. Unusual match-ups that occurred only a single time in the final are:
Honduras v Mexico
Ivory Coast v Japan
Nigeria v Bosnia
Ghana v Croatia