Saturday, 2 August 2014

The Guessing Game: Soccermeters Revisited

With the dust from the 2014 FIFA World Cup having settled the opportunity to revisit the 'soccermeters' published in the phoney war before the World Cup presents itself. These tools appeared as office timewasters in the weeks before the tournament with a mission to keep fans entertained before the big event got under way.  How did these now forgotten tools get on at predicting the outcome of the tournament? offered this soccermeter which eliminated teams from the competition on the basis of various footballing and non-footballing metrics. Footballing metrics seem roughly to fare better, 'tournament history' having predicting Brazil as winner, but Germany getting eliminated in the semis - although Italy were expected to face Brazil in the final. Slate also told us that Brazil would win on odds, although they would eliminate Germany in the semis and face Argentina in the final.  Germany's 7-1 victory over Brazil certainly stunned the betting markets. FIFA ranking proved a completely inaccurate predictor, placing Spain as winners, though at least they were expected to face Germany.  Under this model, however, Argentina would have been eliminated in the quarters and the Netherlands dropped out in the group stages.  

The Women's Soccer metric showed, perhaps, how far behind the 'established' nations are in women's football, with the United States predicted to win, although beating Germany in the final - England and Japan being expected to reach the semis! Economic indictors also favoured the US, placing the nation top in terms of wealth (on GDP per capita terms) and population.  

With wealth placing Australia as runners up, unsurprisingly neither factor was particularly decisive, though the wealth metric did see Germany and the Netherlands reach the semis. If UN Security Council membership was used, England would triumph.  Political importance doesn't improve footballing performance then.  Much to the likely chagrin of Greek, and southern European observers more generally, Germany does triumph under one economic metric - Standard and Poor rating.  Perhaps this puts a new complexion on the images of Angela Merkel celebrating with her nation's triumphant footballers.  And, if we were to believe that correlation leads to causation, adds new urgency to the need for Britain to get its national debt under control...

The Economist's Daily Chart on June 4th offered another attempt to predict the tournament.  They provided a 'probability circle' along with a straight bar graph of probability of winning the tournament, working on the premise that experience matters. The data was based on the outcome of every FIFA game since 1993 adjusted for ranking when the teams played each other and home advantage. Brazil were heavily favoured under this particularly soccernomic methodology, with a formidable 21% chance of lifting the cup. The Germans had only a 7.5% chance of winning the cup, and also only a 7.5% chance of getting past Brazil in the semis.  The Economist's prediction would suggest that Germany managed to play above the expected level, although it would seem more likely that they over-rated home advantage.  The use of historical data did also not fare well, perhaps suggesting they also over-estimated the role of path dependence - Spain were given a 7.6% chance of winning, marginally better than Germany, while Portgual were rated fourth on 6.9%.  England were perhaps more realistically rated at 3%, and Argentina at 6.4%, but the Netherlands were considered complete outsiders at 2.3%.  

As the outcome of a football match depends heavily on the players on the day, along with a dose of luck, trying to use the types of data that were used by the aforementioned tools to predict football matches seems in some ways to be a fool's errand. But it is an entertaining topic and we applaud these two attempts at soccer meters which have added greatly to the excitement and enjoyment of the 2014 FIFA World Cup. 

In conclusion - if you want to predict the tournament then it would surely be wiser to use the data that betting companies draw upon when calculating their odds, or better still to find out how Paul the Octopus did it in 2010! 

No comments:

Post a Comment