You are here

World Cup

GD(ij) =0.137 log (pop (i)/pop (j)) + 0.145 log (y (i)/y (j)) + 0.739 log (exp (i)/exp (j))

Submitted by Roanman on Sun, 06/06/2010 - 09:53


The quants strike again.

Simon Kuper and Stefan Szymanski have applied the following forecasting model they developed for their book “Why England Lose” to the 2010 FIFA World Cup Tournament.

GD(ij) =0.137 log (pop (i)/pop (j)) + 0.145 log (y (i)/y (j)) + 0.739 log (exp (i)/exp (j))

Where i and j are the opposing sides.

“GD” is the goal difference between the two sides, “pop” = population, “y” = GDP per head and “exp” = team experience.

South Africa gets an additional 0.657 each game for home team advantage.

The above formula based on the ratio between the opposing team’s populations, GDP per person, experience and home field advantage for the host nation, Kuper and Syzmanski have applied linear regression (Huh?) to show that the outcome of international matches can be accurately predicted 72% of the time. 

Their winner?

Brazil over Serbia.

Click the chart below for a larger, more readable version.

The chart was scanned from a Wired Magazine UK article at my other new favorite site, Chart Porn

The favored matchup by the betting public is England v. Brazil at about 14/1.

The odds for a Serbia v. Brazil final currently stand at about 150/1.

I'm thinking about it for a hundred.

I might put another hundred on Spain, my favorite team, to win.

Why Spain?

They play the beautiful game.



At least now we know why.


Subscribe to RSS - World Cup