The Paris Marathon Predictor

Enter your time on a distance between 17 and 25km. The calculator will predict your marathon time and pace for you.

Distance:
Time:
::
 

What's this?

There are a good number of marathon time calculators on the internet. Unfortunately, most of them are targeted toward elite runners, or at least runners who log crazy hours in training. What's been lacking is a calculator that will work for the average runner.

Imagine you could interview a great number of runners of all speed levels, who had run a half-marathon a few weeks before a marathon, and then ask them:

  1. What was their half-marathon time?
  2. What was their subsequent marathon time?

From their (numerous) answers, you could apply some statistics and get an average marathon time as a function of one's best half-marathon effort. That's how this calculator was designed. Use it to predict your marathon time from a time on a distance close to that of a half-marathon, run a few weeks before your marathon, at best effort.

I must thank Joseph Hanan whose work in comparing half-marathon and marathon results (see article) inspired this work.

How does it work?

Data was retrieved from the Paris half-marathon and marathon for the years 2008, 2009 and 2010. I found the names of the runners who had run both the (March) half-marathon and the (April) marathon and linked the two. Applying a polynomial fit to the data, I came up with the following equation:

marathon time = -0.419041667*(half-marathon time)^2 + 3.5535 * (half-marathon time) - 0.984

...where all times are expressed in hours.

To get the equivalent half-marathon time from the time and distance entered in the above form, I use a Riegel predictor:

half-marathon time = entered time * (21.0975 / entered distance)^1.06

How accurate is this?

This is a model established from the statistics of almost 10 000 runners having run both the half-marathon and marathon in Paris in the 2008-2010 period. It will not reflect the capabilities of all runners but will instead reflect an average marathon time for an amateur runner based on his half-marathon time.

Note: To "clean" the data somewhat, data was rejected if the ratio of marathon to half-marathon time was above 3 or below 1.7.

René Ghosh, December 2010