Fast Road Race Months

I am looking here to see if there is a time of the year when a runner is likely to run their fastest. To do this I need to find a series of runs that operate through the year, with a large number of runners in varying conditions. parkrun does this with their weekly free timed training runs. parkruns are held all over the country each week and in all weather conditions. The number of parkruns means that many variations are cancelled out if you look at them all over a few years. I am going to look at whether there is a fast week or month of the year using the parkrun series of runs for analysis.

To do this I have a few assumptions:
The population of runners changes every week but the profile of the mass of runners does not change
Given enough courses in enough areas, variations due to weather are reduced
Club runners and male runners run faster than others so this needs to be considered

Each course is given a weekly big percent based on the fastest week for each class of runner for that course. These big percents are averaged for all the courses that ran that week. This gives an indication of how fast each week is.
This weekly average for all parkruns is then given a week number (1st January is week 1, for example) and over the years that park runs have run, all corresponding weeks are averaged together to give a value for each of the 52 weeks in the year.

This plots to give a profile of runs and average finishing times over the years that parkrun has been operating. This gives the following plot.

And this is how runners speed vary throught the year:

January0.748 February0.763 March0.752
April0.762 May0.763 June0.766
July0.78 August0.768 September0.777
October0.742 November0.739 December0.744
Chart of monthly speeds for 5km parkruns

I think that the chart clearly shows that for a 5km timed training run, there is a peak in averaged finishing times in the middle of the year. This is probably around the time when most runners are running at their best.

These numbers are based on 5km training runs (parkrun), and I think the same speed profile would count for other 5km racer in the UK. I also think that this shows for 10km races as well based on my experience, however I cannot show this because there are no 10km races with published results held every week of the year for several years. There are a couple of 10km races that run twice a year along the same course, and when I get chance I will look to see how these compare. The same goes for half and full marathons but I don't know of any that run on the same course at different times of year to compare how finishers speeds compare through the year.