Olympic medal forecasting: it's easier to predict results than you might think

Wednesday, Aug 3, 2016, 08:12 PM | Source: The Conversation

Jeff Borland

The Australian Olympic team chef de mission, Kitty Chiller, has said she genuinely believes Australia could increase its medal haul from 35 in the London Games to 45 in Rio, including doubling the number of gold medals from eight to 16. But can countries really increase Olympic medal counts at will?

The new field of Olympics forecasting says the answer is a resounding no. Medal counts for individual countries from the Sydney 2000 to London 2012 Olympics have been predicted relatively accurately, using just a small number of demographic and economic characteristics, such as population size and GDP per capita.

Country rankings on those measures don’t vary much over time, meaning Olympic medal rankings are likely to stay much the same as well. Indeed, forecasts of Olympic performances for most major medal-winning countries based on their characteristics have been quite close to actual medal counts.

An early influential set of forecasts used just three variables to predict medal counts. For the 2000 Sydney Olympics, these predicted that Australia would win 52 to 57 medals, compared with an actual tally of 58 medals. The US was predicted to win 97 to 102 medals and won 92. New Zealand was supposed to win five medals against an actual count of four.

What, then, are the three variables that can make such a precise prediction of a country’s Olympic medal count?

Three important factors

The first variable is population size. If natural athletic talent is uniformly distributed over the world’s population, then it follows that larger countries will have more athletes with the greatest natural ability.

The second variable is GDP per capita. In countries with higher GDP per capita, the population is likely to have more leisure time to devote to sports and there will be better infrastructure and more funding for training elite athletes.

The third variable is whether a country is the host of the Olympics. Being the host has been associated with a strong positive effect on a country’s medal count (in the Olympics before and after hosting, as well as at the Games hosted).

Host countries spend more on preparing their athletes for the Olympics. The home crowd, familiarity with competition venues and favouritism by umpires also play a role.

New predictions

Four Olympics on from the 2000 Sydney Games, the approach to forecasting medal counts has been refined – mainly by finding some extra variables that explain a country’s Olympic performance.

Knowing whether a country has a communist government matters because those countries tend to spend more on elite sports. Countries that have higher female participation in elite sports also tend to win more medals.

A study published in early June applied a refined method of forecasting to the Rio Olympics. To test the validity of their method, the researchers began by forecasting medal counts at the London Olympics and comparing their results to actual counts.

They predicted the US should have won 105 medals, compared to its actual count of 103 medals, and the UK 60 medals against its actual tally of 65. For most other countries, the method did a similarly good job of forecasting. One country where the method didn’t do such a good job was Russia, which was forecast to have won 47 medals but actually won 65 medals.

So what of the Rio Olympics? The researchers have predicted the US will be number one on the medal table, with 98 medals. China is forecast to be second with 84 medals, and Russia third with 77 medals. The UK is said to be fourth with 62 medals.

Australia is forecast to come in at equal-seventh place with 33 medals. This suggests that the structural factors that determine Olympic performance will make it difficult to achieve Kitty Chiller’s goal.

As the host country, Brazil is predicted to improve the most, increasing from 17 to 33 medals to draw level with Australia on the medal table.

Other variables

There’s one big caveat to these predictions: they were made assuming a full Russian team at the Rio games. With fewer Russian athletes competing, other countries will have the opportunity to snag more medals, potentially throwing off the predicted results.

It’s important to note that there is nothing in Olympic forecasting that refutes the role of the talent, dedication and the effort of individual athletes in explaining their Olympic success.

What the science of medal forecasting shows is that the distribution of talent differs between countries; and that the scope to realise the potential of that talent depends on a particular country’s level of income and culture. These factors will play a strong role in determining who wins the Rio Olympic medal count.

The Conversation

Jeff Borland has received funding from the Australian Research Council.

University of Melbourne Researchers