Under Development

Posted on 6 Aug 2009 at 12:52

 

Gallery

Over the hill

Surely someone has addressed these issues and written software to ease the pain of the calculations? An extensive Google search said not. Nothing on Tucows, either. The only reference to handicapping techniques I found online applied to horses.

What I've got so far is a SQL database with tables for runners, courses and races. Course details include length and also elevation changes, and I devised a Riegel-like formula for compensating for hillyness. I get elevation information from www.mapmyrun.com, which is more accurate than the data you get from GPS devices such as the Garmin Forerunner. These are great for measuring time and distance, but are way out on elevation.

The program stores all the results I can lay my hands on for all runners. To calculate the targets for a new race, you do the following:

1. Add a new race record and allocate a known course to it

2. Load every runner into the entry list (one button click does this)

3. Press Calculate Handicap

4. When the results are known, enter the actual times

5. Press a button that removes the runners that didn't run

The calculator takes the latest three races for each runner and converts the time into a flat 10K equivalent by removing any hillyness and 'Riegeling' the distance to 10K, then it calculates the mean. Then the process gets reversed by adding in a hillyness factor for the target race and Riegeling the 10K average to the actual race distance.

There are issues, including runners that run a race that's not representative (if they weren't really trying, for example); runners with no race history; whole races that are not representative, such as when weather conditions were exceptional; 'reverse handicap' races, where the slowest starts first, then the next one sets off after the difference in their target times and so on; and allocating points that count in a handicap series.

Target practice

Tests over six months show an improvement in target accuracy, with the bonus that everybody knows that their targets are based on historical performance rather than arbitrary factors. Usually more than half the runners have a target/actual difference of less than one per cent. One female runner has a cumulative difference over six races of just 22 seconds. As a bonus, the targets are self-adjusting, so if you train harder and beat the target time, that result feeds back into the target for your next race. If your target gets harder it's your own fault for running better. It's hard to cheat by running a deliberately slow race in the hope of getting an easy target next time - and not many people would do that, I hope.

The program still needs a little polishing before I can let other people use it, but I'm hoping that it will soon be in a fit state to let out into the big wide world - complete with new features that do age adjustments as well. And if you detect shades of self-interest in that, you'd be completely wrong!

Author: David Robinson