Friday, December 10, 2010

Triathlon Rating

Usually this blog is about my own training and races … this entry is a little bit different but still very much focused on trathlon. I’d be happy to receive feedback on my thoughts, please leave a comment!

Macca’s IMTalk Interview

I’m a few weeks back in listening to my podcasts, so it wasn’t until Friday that I listened to IMTalk’s podcast interview with IM Hawaii winner Chris McCormack. A great, long interview – if you haven’t listened to it yet, you can find it here.

One of the things Macca mentioned started me thinking. I don’t have the exact words, but his point was that US races were easy to win but all the press was focusing on. Races in Europe (even the high profile ones) were much harder and also had a much bigger depth than the US races. One of his examples was Marino Vanhoenacker’s 7:52 in Austria – all this got in the US was a “Austria is a fast course” but not much more, so he wasn’t really on most people’s list for delivering a good performance in Hawaii. My idea at this point: There should be a way to rank the relative performances of athletes in different races.

Rating in Chess


When I was younger, I was quite a serious chess player. Chess has a great scoring system to compare the strengths of the different players, see this wikipedia article on the Elo Rating System (named after Prof. Arpard Elo, who developed the mathematical foundation). The score in chess in an artificial number, but it still was quite meaningful in comparing different players.

Known Scoring Systems in Triathlon

In triathlon, the only scoring systems I know are points based. In each race, there is a certain numbers of points to win, usually based on placing. Races are usually in a tiered system, i.e. better tiers dish out more points.

The “official” systems I’m aware of is the Pro qualifying points race for IM Hawaii by WTC and the points system for determining the ITU olympic distance world champion. I’ve also found the site triathlonranking.com that does a scoring of long distance racers. I’m not aware of any official status of this site, it seems to be a portfolio site by Wim De Doncker, a relatively well known Belgian triathlete who’s now also working as a freelance web designer. It’s also not completely up to date.

Problems with Points based systems

A points based system is pretty easy to set up, but I see a number of drawbacks:

  • Frequent racers have an advantage over athletes that start less often.
    A good example is Craig Alexander – his only IM race is Hawaii which does not give him a lot of chances to earn points.
  • It is very hard for a point based system to "fairly” reflect the strength of the field.
    The points you get depend a lot on who also shows up on race day.
  • A points system does not offer a way to compare different courses.
    An example for such a system would be golf’s course rating. Almost all golf courses have a “par” score of 72. A course rating of 71 shows that this course is two shots easier than a course with a rating of 73 (for more, see another wikipedia article).
  • A point system is not predictive.
    Just because you have x points, it does not mean you should be able to race in y time.

Ideas for a Triathlon Rating System

Compared to chess, triathlon has a natural way of scoring: your finishing time! Therefore, a better ranking system would have to incorporate these main ideas:

  • an athlete’s rating should be his finishing time – this would naturally lead to different ratings based on race distance (Olympic, Half IM, IM-distance) – also makes a rating automatically predictive
  • courses should be rated as well – for example by comparing finishing times for the same athlete on different courses you could say that course x is by 15 minutes harder than course y
  • comparing expected and actual results by the athlete’s in a specific race would also allow you to determine if it was a hard year (e.g. in Hawaii a year with heat and strong winds) or an easier one and therefore allow you to compare different year’s results on the same course

Obviously, there are quite some tricky issues to resolve before this can be workable, but I think that the result can be quite interesting. I know that there is some previous work by Neil Hammond, but his main focus was on picking the race that would give you the best chance for qualification (ending in 2007 it’s also not quite up to date).

I think this is something interesting to look into. Does anyone know of anything similar to this? Any other comments or ideas? Also, if anyone knows of a database with decent quality race results, please give me a shout! Thanks.

No comments:

Post a Comment