Moving Forward

Data in Triathlon by Thorsten Radde

By InnerVoice

October 3^rd, 2018

The True Kona Favourite

I’m a longtime triathlete. I think I did my first triathlon in 1988. In total, that year in Germany, I think there were 5 triathlons, so every race you did had a bunch of superstars in the field. I would come in second last or something like that, but triathlon was always something that was very interesting to me. My participation has gone up and down over the years depending on my job and other things in my life, but I’m happy to have something in my life that I enjoy doing, and also following when I can’t race.

The question that I started with was whether you could predict who the true Kona favourite was. The Americans always looked at whoever won Ironman North America events and they would make them their favourites, whereas in Europe we had a much broader view to other races that were going on, like the big races in Germany. I began thinking: there must be some way of analyzing who really has the best results, the best performances, and determine the Kona favourite from that. I’m a computer scientist, so working with data and writing the programs to analyze data is literally my day job, but this was more fun than working with the regulatory requirements in the financial industry.

I knew how to do the analysis and come up with an ‘opinion,’ but the problem was that there wasn’t a database of results, and there wasn’t any ongoing analysis. I was able to dig up 4 or 5 years worth of results for Kona, Frankfurt and a couple of other races, and it all started from there. As soon as I put all the information up on the website, I realized that there’s a ton of people who were interested in this type of thing. I got some pretty quick feedback from people all over the world - some of whom I looked up to and never thought I’d ever get to speak to - and it turned out that I’d created something that’s useful to a lot of people. I decided to keep working on it, and it grew from there.

Photo courtesy of BMC-VIFIT SPORT PRO TRIATHLON TEAM POWERED BY UPLACE

Kona Rating Report

The initial objective of the results analysis was to try to distinguish between someone who went fast or slow: was it because of the course, or because the athlete had a bad day? Then someone suggested that using that data, I can make predictions. It snowballed from there. Now I do 3 posts per race. First is when the start list is beginning to fill up, then when the start list is confirmed I’ll add seedings and predictions for the race, and I’ll also do post-analysis with how fast or slow the course was, and how the major incidents impacted the results. I’ve ended up adding 70.3 races, and a money list, and extended out to the different race organizers as well. Now, the challenge is to present the information in a more digestible format, which means adding more than just tables that are interesting to data geeks.

Of course, I'm a bit of a nerd in delving down into this data, but I think can really improve the commentary and presentation of the sport. It can make it more interesting than just discussing the one or two people that are being shown on camera. For instance, in a regular marathon, about halfway through you’re pretty certain that X, Y and Z are good athletes and since they’re in the lead pack one of them will win. Whereas in triathlon, with the three different legs, people have different strengths and weaknesses, so someone can be 30 minutes behind and still be a chance to make the race interesting. That’s something that’s hard to capture when you just look at where they’re at on the course. If commentators had this information at their fingertips, they could be preparing the viewers for what might happen, or what they should be looking for half an hour down the road. Things like that. I’d love to work on creating a dashboard for commentators (I would need a lot more data to actually do this) to make their job easier and create a really interesting experience for the viewers.

I would like to see TriRating grow and be able to make it more interesting, but that is limited by the amount of hours that I’m about to put into it. It’s not making enough money for me to leave my job, which is when I’d be able to really expand it. I’d like to extend the analysis to age groupers, but that means I’d be analyzing 10 to 15 races rather than just 2, so you can see how it quickly balloons out in terms of the time needed. But for now, I’m really happy with the position that I'm in and that I’ve created something that people find valuable.

This year's Kona Rating Report is a data-driven look at the Ironman World Championships, 132 pages of in-depth analysis and information about the race, all professional participants and their chances on October 13th.