Model Predicts That The U.S. Is Going To Dominate SochiS

We're nearly past the Opening Ceremony, so it's time to get serious: Who's going to win the Winter Olympics? Based on an analysis of economic and geographic data from participating countries through Winter Olympic history, Dan and Tim Graettinger have built out a model for medal performance at the games, and it looks good for the U.S.

The brothers found that the strongest predictors of a country's medal performance were geographic size, GDP per capita, the value of the country's exports, and the capital city's latitude, although there was also a distinct home-field advantage. The final model had an adjusted R-squared of 0.55, and it was accurate within three medals for 80 percent of the 2006 and 2010 Winter Olympic participants (which the Graettingers used as test cases). That's okay, but not amazing, considering how many participating countries have very low medal counts.*

How will the model perform in 2014? The model's predictions (shown above) call for 29 U.S. medals, easily beating out also-rans Germany, China, and Russia. It's less optimistic about Jamaica's chances.

*My proprietary model, called "Every country gets one medal," would have been accurate within three medals for 79 percent of participants in 2010.

[Co.Design]