Thursday, 16 June 2016

West Highland Way Race 2015 Splits Analysis

To help gain an insight what type of splits to use in the West Highland Way Race it can be useful to look at previous years races.  For this article I have taken 2015 race splits published on westhighwayrace.org, and processed them in three different ways:
  1. First half vs second half pace to look how much runners slowed
  2. Comparison between individual runners splits% vs average split%
  3. Comparison between individual runners splits% vs even paced split%
Lots of my own prep to complete for this years race (in two days) so I won't spend long explaining how I did the analysis, instead I'll just dive directly into the results and then wrap up with conclusions.

 1. First half vs second half pace to look how much runners slowed

ratio of Average Pace second half / Average pace
This graph shows the ratio of the pace for the second half (Tyndrum to Fort William) divided by the average pace for the first half (Milngavie to Tyndrum). A value greater than 1 shows a positive split, while a value of exactly 1 is an even split.  There were no negative splits in 2015, we had two in 2014 so it can happen!

The most obvious takeaway from this plot is that there is a broad trend of faster runners running more even splits and slower runners running slower splits.

The second observation is that there can be a lot of difference in first half and second half even for runners that finish around the same time, especially for those in the middle of the race.

The third observation is that the top 8 runners all had slowed less than 20%, with three slowing less than 5%.

Paul Giblin slowed by 9.9%, which is substantially less than he slowed in 2014 where he slowed by over 16%.  This suggests that racing Robbie Britton hard in the first half of the 2014 had penalty on how well he could maintain pace in the second half.  Compared to the rest of the field though even in 2014 he was still finishing far stronger.  In 2015 ran faster without the pressure of any close competition - he was in effect race himself and was behind even as late Kinlochleven but was able to overhaul his 2014 self over the Larigmor with some very impressive running.

Second placed runner Neil MacNicol came very closer an even split only slowing down by 1.7%.  It's worth reading Neil's race report to see how it unfolded.  Neil ran an exceptional race, especially considering it was his first outing over the distance.

By contrast if we look at the tail enders then then are slowing down 40 to 75%.  This level of slow down will be unlikely to be down to planned pacing, I suspect it's more to do with runners encountering problems as the race unfolded rather than just going out too fast.  I'm sure going out too fast will have been a big factor though, you are much more likely to struggle later if you go out too fast.

2. Comparison between individual runners splits% vs average split%

Speed difference between Actual runner splits 
For the second part of the analysis I computed the split % for each leg and compared these to the average split %, this shows how well using average splits for the whole field is able to predict what splits the actually saw in the race.

The clearest takeway from this is that in the middle of field, between 22 and 30 hrs the average spits do a reasonable job for most of the runners, with most being only 10% off, with a decent number below 5% suggest good match.  The way these difference manifest themselves will be different for each runner, a runner who runner a more positive split than another runner might be equally far away from the average so the accuracy relative to the average splits can end up the same.

As we move to the front of the field the average splits provide a much poorer fit for most runners, you have to get out of top 8 before you seen any close to average splits.  A different splits model would be appropriate for front of field.

Also if we look at the tail ended, runners over 30hrs, again the average splits start to be a more proxy for the actual splits runner see.  This will be related to the the significant slow down that these runners experience.

3. Comparison between individual runners splits% vs even paced split%

Individual runners splits vs even effort splits
For the 2015 West Highland Way Race I created a set of even effort splits that where based on 2014, adjusting for parts of the race where I had problems such as tending to blisters or succumbing to injury late in the race.  I ran the 2014 pacing by heart rate and got the effort level throughout race right, finishing strong despite injury.  These adjusted splits are my best attempt at estimating what even effort splits would look like. 

Even effort splits aren't even paced for indivudual sections - some sections like the first to Drymen or the final descent to Fort William are relatively fast even with the same effort level thanks to the easy terrain or downhill, while by contracts Inversnaid to Beinglas or going over the Devil's Staircase are significantly slower pace wise even if you are maintaining the effort level.

If we look the plot we see a very similar trend as we saw for the first graph that looked at average pacing in the first half vs second half.  Clearly even splits works much better for the front of the field than the majority of the field, especially the tail enders.

For the front of the field the top eight are actually closer to the even effort splits than the average splits.

By contrast the rest of the field the even effort splits increasingly show a poor correlation to the actual splits runners saw on the day.

However, there are still runners where even effort level splits were representative even up to 25hrs.  This suggest that good pace judgement isn't just the preserve of the elite, other members in the field can do it too.

Lessons learned from the analysis

It's very clear that the elite runners are able to maintain their pace far better than the majority of the rest of the field.  This is likely a factor of the training they do and their genetic disposition to ultra racing, as well as experience with how to eat, drink and manage themselves through the whole race.  Part of this ability will also be experience in pacing, they know what they are capable of and pace accordingly.

There is also a factor in that not everyone has a perfect day come race day, those at the front of the field will mostly have been having a good day.  Even great runners who have a bad day will end up slowing badly and moving further away from even effort splits.  This applies even more to us mortals further down the field.

Last year I was one such runner, slowing by 14.55%.  I was very close to even splits as planned up to Bridge Of Orchy, but stomach problems that began at Tyndrum eventually got so bad that I had to walk for the majority of the rest of the race.  My training and pacing weren't to blame, I was fit and pacing for a sub 20hr time, but my day went wrong because of other factors.  In my case the other factor was that I picked up a cold days before the race, and then just hours before the start I strained by back so was in pain at every step.  The cold and back injury meant that I ended up taking pain killers/flu tablets during the race to keep on top of things.  I believe it as the co-codamol pain killers that were the mostly likely cause of the stomach problems.

My experience from last year illustrates that there are many things that can cause you to slow down, so under your control, some out of control, but the analysis can't tease apart these individual stories, all it can pull out is general trends.

When planning your own race we can learn from even these broad trends.  If you want to have a perfect day of racing like the front of the field then it makes sense to pace like them, and this means that even effort splits are likely to be good aspirational place to start from.

Even effort splits have runners start the race well within their capacity, starting off at an intensity that they know they can come close to maintaining for the whole race.  Some runners are really good at judging this by feel, others are terrible at it and even when they try to hold themselves back still go out too fast.

Practice approach to pacing with even effort level

For my own racing I made a number of mistakes in pacing in ultra's before settling upon using a HR monitor as a guide for managing my effort level.  The HR monitor is a good proxy for effort level so if you pace yourself within an appropriate HR band then won't be too far even effort pacing without ever having to look at splits.

For myself for this year's West Highland Way race a HR range of 130 to 140 is roughly appropriate, this is 75% to 81% of my Lactate Threshold HR, or 70 to 76% of my max HR.

I have also computed even effort level splits for this year's West Highland Way Race and will publish these in a follow up article.  These splits will take account of a small route change in this year's race so won't match exactly to those I used in 2015 and for this articles analysis.

1 comment:

  1. This comment has been removed by a blog administrator.

    ReplyDelete