Friday 26 February 2010

Olympic Metrics

I don't mind using medal counts as a metric for Olympic success, and I fully acknowledge that Canada badly failed in their "Own the Podium" campaign, but I think that the heavy focus on medal counts misses a lot of information. Clearly a fourth place finish is far superior to a fortieth place finish but we don't see that in medal counts. Another problem: sports like hockey or curling award only two medals, while most others, like short track speed skating, award large batches of them. If country A has an excellent hockey program, at best it's rewarded with two medals, if country B has an excellent short track speed skating program it can result in 8 or 10, but it doesn't seem to follow that country B is 4 or 5 times better at winter sports than is country A. I've thought of a few other simple metrics that would help communicate additional information.

a) Average percentile or ordinal/athlete: Take the average of the sum of the ordinal of the finish of all participants from a given country. Or if that would bias too heavily to sports with fewer participants, use percentiles. DNF or DQ count as one place lower than the last qualified finisher.

b) Average percentile or ordinal/event: Same as (a), except that we factor in breadth of participation so that breadth of qualification and participation is factored in. If a country has no participant in an event, we count that as a last place finish. If a country has two or more participants in an event, we take the average of their ordinals or percentiles.

c) Sport counts rather than medal counts: Given a metric for success in an event, e.g., best percentiles or most medals, we then identify a finish/sport. That number is then used to measure overall success. So, for example, a country finishes with two golds in hockey, they get 1st in hockey. They finish with 4 medals in short track but another country has 5 (with some weighting for ordinal, of course), they get second in short track, etc.

I don't think these need replace medal counts, but I think they'd be useful supplementary metrics. Next step, calculate these for the 2010 Games. (Don't hold your breath.)

No comments: