Ironman New Zealand 2013: A Deeper Look at the Run
There is something deeply unsatisfying about observing what you believe to be a pattern without a good explanation for it. The example in this case is the seemingly poor – relative – performance of men under 40 at Ironman New Zealand. At least when I examined the times for the top 20 placings over the weekend, men in those categories were consistently slower than the expected averages. I’d assume a slow race and be satisfied with that were it not for other age divisions showing better than average performances and a course record being set. With race day heat as a possible candidate for blunting performances I wondered if I could see any evidence for this by looking at splits in more detail.
The initial idea was to examine the detailed run splits for deterioration in pace; I would expect issues with heat (along with pacing and nutritional errors) to have their biggest impact in the latter half of the marathon. On a practical level the fact there was only detailed split data for the run – and even that was missing one timing point – limited further analysis anyway.
The charts above show the average run pace (in minutes/kilometre) by age group at each timing mat. There was no data for the 35K mat so pacing at that point is extrapolated. Some of the variation in pace can be explained by the imprecise placing of mats, they probably weren’t all at exactly 7km intervals, but the overall trend, slowing over the marathon, is expected. In fact there’s nothing remarkable in this breakdown: there is no clear indication of significant deterioration among certain age groups. I suspect this was always likely to be lost in the averaging.
An alternative approach, perhaps more realistic than looking at individual age groups for signs of poor performance, would be to consider run pacing by finishing time as in the chart above. I’d wondered if what I saw in the age group top 20s was fast age groupers pushing themselves too hard on a hot day when they were more susceptible to the impact of pacing and nutritional errors, while those over 40 were more prudent in their pacing strategies. If so, then the faster finishers might show more indication of deteriorating performance on the run. Apparently not, if anything the chart shows larger performance degradation in the slower age groups with the most significant change happening mid race (probably an amplification of bad timing mat placement).
As I was already playing with the data and having struck out with all lines of enquiry I thought I’d look at the top level splits and see how the age divisions performed overall. These averages – which I clearly should have started with – show a different picture. In the men’s field the average finish time and splits trend as I would expect with 30-44 year olds being the fastest categories; the women’s field is a little less clear, but again closer to what I’d consider a standard Ironman.
The pattern I’d observed in the top 20 of each age group doesn’t extend beyond that chart. Averaging over the entire field removes those differences and instead I find a typical Ironman race. To an extent that makes those slower performances seem more anomalous, but my explorations have yet to devise a good means of observing this more accurately. I remain unsatisfied.
But taking steps towards examining race splits in more detail should enable more interesting comparisons between events in the future. Side-by-side we may be able to observe differences in pacing under different conditions. I’ll come back to this when I have more – and more useful – results.