Race Result Predictor

One of the problems with racing in Colorado is that you show up at a local hill climb, with hardly any prize money to speak of, and you notice Tom Danielson warming up in the parking lot. It’s easy at this point to predict that you won’t be winning…

A good friend of mine just sent me a link to a road race “predictor” that scrapes registered riders from bikereg.com‘s website for an upcoming race, and compares those data with each rider’s USA Cycling points, in an attempt to predict the winner. Sounds fun, eh?

I love hill climbs, and the Mt. Washington Hill Climb has a list of registered riders on bikereg.com, which includes me, so this seemed like a perfect test. I started off by going to the race predictor website (http://www.road-results.com) and I entered the link to the Mt. Washington confirmed rider list (http://www.bikereg.com/events/ConfList.asp?EventID=9819) to see how I could expect to do when I race up that hill in August of 2010. What do you know, I’m predicted to get 268th place, and Chris Hurst is predicted to win. Chris Hurst and I both did the Mt. Washington hill climb in 2009, and I was about 38 minutes faster up the 7-mile climb. Maybe the website knows more than I know, and maybe Chris has gotten 39 minutes faster, or I’ve gotten 39 minutes slower…

Okay, to be fair, here are a couple of reasons why it might not be working in this case:

  1. Most races in Colorado are not USA Cycling events, so those of us who race here don’t have many USAC points. In Colorado most races are American Cycling Association sanctioned events instead. I do have a few points, but not many.
  2. I don’t think that the road race predictor algorithm takes into consideration what type of race is being predicted. If a rider has lots of USAC points, but those points were earned from winning crits, using these data to calculate how a rider will do in an upcoming hill climb probably won’t be that accurate.

While I think it’s not exactly fine tuned at this point, I love the idea of this website, and by adding a few additional metrics into the predicting calculations this could be much more accurate. If they looked not only at your points, but the type of race in which you earned those points, that would be a huge improvement. And if they combined this with the results for that same race from previous years, that would also be helpful.

What I’d really like is to have someone who races in USA Cycling events, and who does have points, to write up a little report for us. Any takers? We’d love to hear what you think of theĀ  road-results.com website and how well/poorly it works. Most importantly, is it fun cycling technology?

And if any IT geek Colorado racers out there have spare time on their hands, they could do something similar for us by scraping the American Cycling Association website for rider results and upgrade points.