Predicting the Medal Count

Economists love to make predictions about the Summer Olympics, but the Winter Games generally attract less attention. One economist, however, does have some predictions for this year’s Games. Daniel Johnson analyzes country population, income per capita, climate, political structure, and host-nation advantage to generate medal count predictions — his “model demonstrated 94% accuracy between predicted and actual national medal counts” for the last five Olympics. At this year’s games, Johnson expects Canada to win the medal game with 27 medals, and the United States to tie Norway for second. Johnson’s model also predicts that the Nordic countries will make big gains over previous years, while Germany and China will slip in the medal count. (HT: Josh Levin and Kiril Rusev.) [%comments]


In Ancient Greece, as with the Cold War, the medal count was used as an indicator of the superiority of one political system over another. Today, "the count" is an anachronism, particularly with the multiple dual citizenship or immigration to increase probability of getting on a country's team. It is even more so with the coaches as very much invested in the capitalistic system. China is about the lone hold out with a national push and policy toward every increasing their medal count.

The nationalistic medal count reminds me of a line from a John Christopher novel, "The Guardians" commenting on the fostering of team loyalty as a means to control (and express) pent up emotions/frustrations.


Susan, you say that "China is about the lone hold out with a national push and policy toward every increasing their medal count."

This is inaccurate. As has been widely reported, Canada poured hundreds of millions of dollars into their Own The Podium program, and the US, by far the best capitalized Olympic Team, spends millions more in an effort to capture gold, silver, and bronze for its athletes. True, the USOC does not receive government funding, but any notion that the Chinese are somehow alone in their nationalistic desire to get on the podium.

Andreas Kitzing

Nice stats... However, from a German point of view, let's hope that Johnsons forecast regarding the German medal decline is wrong. ;)
Best wishes from Hamburg, Germany,

Rodney Fort

I hope Freakonomics will return to this prediction to see how it actually works out AFTER the Vancouver games are over.

Too often, soothsayers get their day in the sun, but then are never called to task for their actual merit.


Are there any prediction markets for the number of medals? As in, where many people can make their predictions? It would be interesting to see how these perform in comparison. Although 94% accuracy is hard to beat.


You should compare with Nate Silver's analysis at his blog He's projecting Canada and USA tied (!) right now (he updates his projections as things happen), with Germany a close third and Norway a big gap behind them.


@6: Ditto. Nate's the man where sports are concerned...actually where politics are concerned too.

Economists Do It With Models

Is it bad that all I can think about is that I wonder what the R-squared of the regression that this guy is running is??


YouReputation Prediction for
Medal Counts for the Vancouver Olympics
(based on viral entropy)

Michael Peters

538 does a good job of predictions that are adjusted day-by-day as more data comes in:

Rudiger in Jersey

I think Germany will lead overall medal count followed by America, Russia and Canada. Canada would be most happy with a medal in Men's Hockey, but home ice is only 2-4 medals over a typical Olympics.

Kevin C.

I wonder what Johnson's analysis would have predicted for Canada in 1988...


My only comment is that in the informal scoring one finds in the newspapers they ought to follow the lead of track and field; viz., 5 points for first (Gold), 3 points for second (Silver), and 1 point for third (Bronze).


I thought this article (and the research paper itself) were both very interesting. Now, a little more than halfway through the games, the actual medal counts look very different from Professor Johnson's predictions (the US leads with 25 medals while Canada languishes in fifth place with just 10). It would be good to have a follow-up article discussing what went wrong and how one might develop a better model. I'm also wondering how they calculate that his model is 94% accurate...



How does this prof's model explain South Korean's high medal count compared to neighboring powerhouses such as Japan and China? Per Daniel Johnson's methods(country population, income per capita, climate, political structure, and host-nation advantage) both Japan and China should handily beat Korea in medal counts. So far, this has not been the case. At all.

How does Johnson explain this strange phenomenon? I certainly can't, even as a Korean myself...


Well, it seems like the prof missed the actual medal count by more than in the past: 26 predicted vs 37 actual for the US, 20 vs. 30 actual for Germany, 25 vs. 16 actual for Austria, 23 vs. 15 actual for Russia. More accurate for Canada and Norway: 27 vs. 26 actual, 25 vs 23 actual... what was it that threw his predictions off? The weather, perhaps?

arunabha guha

alas, our dependence on models (particularly from economists) lead us so astray!