Medal predictions: Did we get it right?

Never afraid of a challenge, before the start of the London 2012 Games we issued predictions for the total medal count for the top 20 countries. They were based on a mathematical model that took account of a country's GDP and population, its performance in 2008 and the home advantage bestowed on Great Britain and also China, who hosted the Games in 2008 (see Mapping the medals).

So how did we do? The first thing to notice is that, despite team GB's gold rush, it actually performed worse in terms of total medals won that we had predicted, ending up in fourth place rather than in second — despite the home advantage, the likes of Russia and China are hard to beat. Hungary makes a surprising appearance, entering in 14th place while we didn't have it down as making the top twenty at all. New Zealand came in 18th, unpredicted by us, and Iran also just about squeezed in, tying with Jamaica in 20th place. And Kenya, which we had down to come 18th, narrowly missed the top twenty, coming in as 21st. Other than that all but three countries from our original list stayed within three ranks of our predictions. Japan and the Netherlands both performed better than we had predicted, Japan coming in as 6th rather than 11th and the Netherlands at 11th rather than 16th. Belarus gave a disappointing performance compared to our mathematical benchmark, coming in as 16th rather than 11th. Overall, 19 of our top 20 predicted countries finished in the top 20 (top 22 actually as 3 countries were tied on 20th).

Where could we improve?

We failed to note that country populations and GDP nearly always rise between Olympics, whilst the number of medals available in 2008 and 2012 were roughly the same.

This means that to use the equation relating population and GDP to medal count derived from 2008 we should have scaled the 2012 country data relative to the whole world's population and GDP. Imagine your population rose between 2008 and 2012 — you might think that you would have a greater chance of winning a medal. But if the whole world's population grew, then your chances wouldn't have increased. We need to use population (and GDP) scaled to the whole world.

This makes little difference to our predicted ranking, except for moving Brazil up from 17th to 16th place and Kazakhstan down from 18th to 19th, and acts to slightly dampen our medal counts. These new counts are shown below against the real 2012 results:

External links

The BBC's More or Less team conducted a similar statistical prediction of what countries might be expected to win in the way of medals with the help of the Kellogg School of Management at Northwestern University; here is their analysis of the results.

The Guardian teamed up with the Royal Statistical Society and four academics at Imperial College London to produce very interesting alternative medal tables for the 2012 Games, taking into account factors including GDP, population and team size. You can see their final alternative medal table winners here.