Olympic data help decide victors in race for 2020 funds

Speedier decision making will boost the possibility of medal success
On track: analytics allows study of athletes such as Sir Chris Hoy © Charlie Bibby

Listen to this article

00:00
00:00

The world-beating performance of medallists at the Rio Olympic Games this summer will probably be matched by the speed at which Team GB’s performance levels will be displayed on computer screens at the UK’s sport-funding body.

The data will help UK Sport determine the funding British athletes will receive for Tokyo in 2020 — and possibly even for Olympic Games further in the future. The body hopes speedier decisions about funding will boost the possibility of athletes’ success because training for the next games begins just a few days after they take the podium.

Time waits for no one in world-class competitions, says Simon Timson, director of performance at UK Sport. “The quicker and more accurately we can make our decisions and get the right support to the right athlete and the right sports, the greater our competitive advantage,” he says.

UK Sport agrees individual medal targets with 38 Olympic and Paralympic sports plus an aggregate target with the government as the basis for investment.

“Three years ago, we’d often sit in reviews of sport and investment meetings and ask what potential a group of athletes really had,” says Mr Timson. “We’d have to take the sport’s word for it and we’d have some doubts. Track records would be patchy and often the likelihood of successful performances looked marginal, so we would have to err on the side of caution.”

In 2014, UK Sport’s intelligence team began working in partnership with Quebec-based CGI to create a portal to analyse and report on a wide range of data that can inform its decision making for the 2020 Tokyo Olympics.

The techniques that are being used include logistic regression — a method of predicting outcomes based on several variables — to analyse previous and current performance data of UK athletes. This is combined with data, such as the performances of other world-class competitors in big sporting events and the normal swing in the number of medals from event to event on a sport-by-sport basis for each nation.

The software lets UK Sport look at failure and success rates for athletes who are, or were, on its “podium programme” of potential medal winners. Mr Timson says: “It gives us a lens to look into how ambitious and realistic the proposed target range for Tokyo might be. So we can understand whether we should be investing in fewer or more athletes.”

Previously, such calculations would have been too complex, says Mr Timson. “The analysts had to spend most of their time collecting, ‘cleaning’ and ordering the data. So they had less information and were able to perform fewer analyses.

“We now have a much clearer understanding of each athlete and each sport’s medal potential, which is crucial when you’re making four to eight-year investment decisions.”

The software allows UK Sport to focus on potential implications of their decisions. “And it enables us to put the facts and evidence on the table for individual sports,” Mr Timson says.

“A sport might claim a particular athlete could win a medal for a specific event in 2020, or even in 2024, and we can see from the analysis that to have a 40 per cent chance of achieving this the athlete would need to be 10th in the world,” says Mr Timson.

“If the athlete is outside that range, it would take something pretty special to convince us that this is a worthwhile investment.”

Data are also fed into a forecasting algorithm for the Rio games that simulates Team GB’s performance 250,000 times to understand the range of potential outcomes and predict the most likely number of medals it will win — which currently is 53.

“It has changed the dynamic and the nature of conversations we have with sports,” says Mr Timson. “We can be prudent, targeted and precise now in where we place our investments. It takes out much of the uncertainty and risk in our decision making, enabling us to make the most precise and best use of the £500m National Lottery and government funds we receive [for 2013-17].”

The first phase of the project, now complete, was the automation of data collection but high-volume analyses still takes two to three weeks.

Phase two will be live analysis of the Rio results as they come in for each event, plotting athletes’ trajectories towards the podium in 2020.

Carl Statham, director of digital transformation at CGI, says other countries are investing in similar software, including the US, the Netherlands, Brazil, Australia and New Zealand and the winning nations will be those which use it most effectively. “Sport has been slower than sectors such as finance to deploy analysis, but the UK is ahead — other countries have not put so much thought into it,” he says.

Copyright The Financial Times Limited 2017. All rights reserved. You may share using our article tools. Please don't copy articles from FT.com and redistribute by email or post to the web.