Saturday, 5 January 2013

Centurion Running End of Year Analysis - 2012

I recently performed another bit of analysis, this time using the Centurion Running data from throughout the year as a kind of retrospective study of 2012, and also to look for any interesting trends. This wasn't anywhere near as in depth as the survey that we ran for the South Downs Way (there'll be more of that later this year), and the data that I used was just the checkpoint times available on the Centurion Running website. Despite this being a relatively simple look at pace profiles, we actually found some potentially interesting things, particularly relating to dropout rates. 

The nice thing about having access to the checkpoint times and not just the finishing times is that it gives us more scope to look at trends in pacing throughout the entire course. Also it is very useful to be able to pull out the dropout point of runners who were unable to complete the course, as I believe that this has highlighted some potentially quite interesting points relating to how runners are able to decide on which point to pull out at. I plan on doing something similar with the Western States data that I have played with in the past to see if we see similar trends to those shown below. More on that in a few weeks.

The report can be downloaded from the Centurion Running website here, but I have added the report below as well. Enjoy!

1 Introduction

2012 has been a fantastic year for Centurion Running. Following on from the inaugural North Downs Way 50 mile and 100 mile events in 2011, this year has seen a full timetable of ultrarunning events including the Thames Path 100 miler, South Downs Way 100 miler, North Downs Way 50 and 100 milers, Winter 100 and the now infamous Piece of String Fun Run. With the completion of the Winter 100, we have reached the end of the racing season and include here a retrospective for the main races in the Centurion Running 100 mile event calendar.

For the South Downs Way 100, we ran a survey of runners in a pilot project to look at factors relating to successful ultramarathon running completion. In 2013 we hope to roll this survey out to the other events to build up a database of information for 100 mile runners and allow us to look for interesting trends amongst the people who choose to run 100 mile events for fun. Please keep your eyes open for more information on this in the new year, and please take part as the more information that we are able to collect the better the findings will be.

For now, enjoy the holiday season and have a wonderful New Year. Happy running everybody!

2 Race Descriptions 

2.1 Thames Path 100
The Thames Path 100 miler (TP100) was the first event in the Centurion Running ultrarunning calendar, held as it was at the start of March. The race ran from Richmond in London to Oxford, following the Thames Path alongside the River Thames. With a total climb of only 2,500 ft, this was due to be one of the flattest 100 milers around and hopefully one of the fastest. The front runners didn’t disappoint, with Craig Holgate winning the men’s race in 15:11:15 and Mimi Anderson coming in first in the women’s race in 18:50:30. Despite the lack of any notable hills, the consensus was that there is no such thing as an easy 100 miler. The weather was perfect for running on the Saturday, with a dry week leading to excellent under-foot conditions, cool temperatures allowing runners to keep up the pace, and even some sun which came out later in the day to cast its rays over proceedings. But many runners were taken by surprise by the sudden change in conditions overnight, with temperatures dropping significantly, and sleet and snow causing chaos for runners still out on the course. Due to the severity of the conditions and the suddenness with which it had hit, the race was abandoned after 26 hours and 5 minutes. 14 runners were stopped out on the course before the finish line. All were instated as official finishers and will return in 2013.

2.2 South Downs Way 100
The South Downs Way 100 miler (SDW100) was the second event in the diary, held in the summer at the end of July. The race followed the South Downs Way National Trail from Winchester to Eastbourne, with a total ascent of around 12,700 ft. Conditions underfoot were excellent, with the rolling hills of the South Downs proving a beautiful back-drop to some fantastic performances, with Ryan Brown winning the men’s race in 17:04:26 and Claire Shelley winning the women’s race in 19:43:03. Conditions were once again fantastic, with cool temperatures and light drizzle in the morning soon turning to warm sun throughout the day.

2.3 North Downs Way 100
The North Downs Way 100 miler (NDW100) was next on the agenda, and was held only 6 weeks later in August. The race followed the sister trail of the SDW100, running from Farnham to Wye along the slightly more technical North Downs Way. With a total ascent of just under 10,000 ft, runners were nevertheless faced with the tough prospect of ascending the energy sapping steps of Box Hill midway through their journey. The men’s race was won by Brazilian elite runner Manuel Lago in 17:51:56, and the ladies’ race was won in a stunning performance by Alice Hector in 20:10:39. Conditions were warm due to the summer sun which had finally decided to make an appearance, leading to dry underfoot conditions.

2.4 Winter 100
The final race in the season was the Winter 100 miler (W100). The route for this race was designed as a 4-armed “cross” shape, centered on the Morrell Rooms in Streatley-upon-Thames. The idea here was to have a base where runners would never be too far away should anything happen given the potential problems that may occur due to the late November weather. The route took advantage of the intersection of the Thames Path and the Ridgeway National Trails, with runners heading out for four 12.5 mile out-and-back sections to Little Wittenham, Reading, Swyncombe, and Chain Hill. With a total ascent of only 3,840 ft, this was another relatively flat event, although runners would have plenty other things to deal with than hills. The week prior to the race, Britain was hit by heavy rain which had led to extensive flooding of the River Thames, and incredibly muddy and treachorous conditions underfoot. Race organisers were forced to make last minute changes to the route to avoid the dangerously swollen river (particularly at night). Also, runners had to contend with the extremely strong winds on the Ridgeway. Despite the inclement weather conditions, many runners persisted and completed the race under some very trying conditions. The men’s race was won by Richie Cunningham in 17:14:10 (an astounding time given the conditions), and the women’s race was won by Jean Beaumont in a fantastic time of 20:23:46.

2.5 Number of Runners
Figure 1 shows the number of finishers and non-finishers at each of the four races throughout the year. The number of competitors in each race was lower for the later events, particularly the Winter 100 which was limited to fewer than 100 entrants for safety purposes.

The proportion of finishers was greatest for the SDW100 and TP100 (around 70 %), possibly due to the fact that the conditions were good on the days of these races. The NDW100 showed a lower completion rate (59.48 suggesting that the conditions were perhaps tougher here this year (possibly due to the warm weather). The completion rate was incredibly low for the Winter 100 (42.67 likely a result of the incredibly poor conditions and muddy terrain on the day.

This suggests that the conditions on race day may be incredibly important in determining the number of finishers for a 100 mile race. It will be interesting to see the results in future years where races have different conditions to see if the finishing rate is affected by yearly changes in the weather.

Figure 1: Number of finishers and non-finishers for each race

3 Age

This figure shows the distribution of the ages of the competitors for each race. The distributions are split by male and female competitors, and are shown for all competitors that ran the races, and also for those that finished the races versus those that were unforunately unable to finish. The distribution of ages (in years) is shown in the format of a “violin plot”. This is similar to a boxplot, with a central dot indicating the mean of the data (which can be easily compared between groups), and with the frequency of other values indicated by the bulging of the violin. The areas where the violin is at its fattest represents the ages that occur most frequently within the group.

It is clear from these figures that there is no real difference in the age distributions of runners between the four races, or between finishers and non-finishers. This suggests that age is not a factor in determining whether or not a runner will complete an ultramarathon.

Figure 2: Age Distribution Amongst Competitors

4 Pacing

4.1 How Fast Were People Running?

In our recent SDW100 analysis, we looked at the pacing strategy of the runners across the entire field. From the results across the four events, we can get an idea of the pacing strategies used by the runners on each of these races. Comparing them allows us to see if different strategies were favoured in races with different conditions and profiles.

Figure 3 shows violin plots of the distribution of running speeds (in mph) of competitors between each of the checkpoints for each of the four races. The speed plotted at each checkpoint is the speed between that checkpoint and the previous one. No correction has been made for the terrain, so any hills which may exist along the course may also explain obvious dips in speed amongst all runners.

The violin plot shows the median (50th percentile) of all runners at each checkpoint as a white circle in the middle of a curved box. The curvature of the box indicates the number of people that ran at that speed, so is wider at the most commonly run speeds for that group.

Essentially, what this means, is that the box shows where the most common data points lie, allowing us to judge how the values are distributed in a simple visual way. The blue boxes represent the times for the male competitors, whilst the red boxes represent the times for the female competitors.

In all of the races, the pace started out high and gradually diminished over the course of the race, suggesting that the majority of runners favoured the approach of heading out at a good pace and gradually slowing down over time. Several sections of the races appear to stand out as being significantly slower for the majority of runners as compared to the rest of the race, suggesting that these may be the more difficult sections. On the SDW100 profile, it is clear that the sections between Queen Elizabeth Country Park (QECP) and Harting Downs, and between Cocking and Bignor Hill were particularly tough, with all competitors showing an obviously slower pace at these points. The other obvious point is the sudden drop in pace at Box Hill on the NDW100 course. Interestingly this drop in pace was never quite recovered for the rest of the race, suggesting that Box Hill may indeed be a fearsome beast to contend with!

Figure 3: Speed Distribution Amongst All Competitors At Each Checkpoint
4.2 Pacing of the Top Runners
Figure 4 shows the running speed as a function of distance for the winning runners in each event (top) and the average of all finishers across the four events (bottom). This allows us to see how the pacing of the race compared between the races. The left hand figures show the results for the male runners, and the right hand figures the female runners.

Looking at the race winners, it is immediately obvious that the pace for the TP100 was much faster than for the other events, which is likely a combination of the fact that it is a flatter course and the fact that the winner, Craig Holgate, is an immensely skilled marathon runner. In particular, the pace was much faster for the first marathon distance, but slowed quite considerably to a pace that stayed relatively consistent throughout. This was due to Craig pulling away and maintaining a steady pace on his own, not actually having wanted to take the lead so early in his first 100 mile race.

In the female race, Mimi Anderson’s pace was similarly fast to begin with but slowed after the first marathon distance. Inter- estingly, there was a very large drop in pace towards the end of the race, which may be a result of an incorrect entry in the data. Mimi’s race report did say that she had a bad patch at around this stage of the race, but it is unlikely that she slowed to quite such a slow speed as this.

Interestingly, despite the tougher conditions at the NDW100, the female winner, Alice Hector, ran a similarly paced (although ultimately slower) race to Mimi at TP100. In comparison, there is a clear difference in overall pace between Craig’s TP100 race and Manuel Lago’s NDW100 race. Manuel ran a relatively even paced race, although there are two points (50 miles and 100 miles) where his pace dropped quite considerably; the first at 50 miles, where Manuel had trouble with navigation on the course, and the second at the end of the race where his energy reserves were severely depleted.

Interestingly, despite the fact the conditions at the W100 were terrible whilst those at the SDW100 were perfect, the winning paces were quite similar between the two for both males and females. The pace for the SDW100 was typically slower than for the W100, probably due to the fact that it was significantly hillier (with Harting Downs and Bignor Hill being obviously tough sections). However, from mile 80 the pace at the W100 dropped significantly whilst the pace for the SDW100 increased. This may be due to the incredibly strong winds that hit runners on the last section along the Ridgeway on the W100.

If we look now at the average pace of the runners in each race, the most obvious conclusions to be drawn are as follows:
  1. Runners typically set off at a fast pace for the first marathon or so, then settled into a more steady pace for the rest of the event.
  2. Runners of the TP100 headed off the fastest, but after the initial marathon the average pace dropped to roughly the same (and potentially even slower) than for the other races. This is likely a result of runners heading off too fast on an “easy” 100 miler, and then soon paying for it.
  3. The most obvious difficult sections for all runners were Harting Downs and Bignor Hill at the SDW100, and Box Hill at the NDW100.
  4. The slower pace seen by the winners of the W100 in the last 20 miles or so was not seen on average across all runners. This is probably due to the fact that the weather improved significantly later in the day when most runners were running this section.
  5. Judging naively in the average pace of the runners, it would appear that the TP100 was the “easiest” of the four races, followed by the NDW100, the SDW100, and the W100.
  6. Having said that, the average pace for all four events was incredibly similar once runners had settled into a comfortable pace, suggesting that running 100 miles is similarly difficult whatever the circumstances.

Figure 4: Pacing Strategy of Top Runners and Average Runners

5 Non Finishers

DNFs (“Did Not Finish”) are an unfortunate reality for all runners. A lot of things have to go right on the day for a runner to complete a 100 mile race with zero issues, and unfortunately over the course of a race the chances of this happening are slim. Everybody suffers bad patches, and often these can be so bad as to necessitate pulling out from the race. By comparing the number of runners who drop out from each of the races, we can get a feeling for how tough they may have been.

Figure 5 shows barplots of the number of runners that pulled out before each successive checkpoint for each of the four races. This could mean that they pulled out at some point leading up to the checkpoint, or that they never left the previous checkpoint. The number of DNFs is scaled by the total number of runners to give the percentage of runners that DNF’d at each checkpoint. This allows the four races to be compared with one another. The number above each bar indicates the cummulative number of runners that pulled out by that specific checkpoint. The numbers are shown for male and female runners separately.

In most of the races (with the exception of SDW100), the checkpoint where most runners pulled out was the one at around the 50 mile mark. This suggests that 50 miles represents a psychological milestone for many runners, acting as a point where perhaps runners are happy to pull out knowing that they have worked incredibly hard to get to that point, but that there is still the same distance remaining until the end of the race. It is also a key point in the race from a difficulty point of view, as most runners are entering darkness soon after or at the point of reaching the 50 mile checkpoint. As to why the SDW100 does not follow this trend, it is difficult to say, but perhaps the volunteers at the aid station were particularly encouraging in getting runners to continue. Similarly, very few runners pulled out of any of the four races before the checkpoint that lies at roughly the first marathon distance.

Typically, after the halfway stage, runners pulled out at a fairly even rate up until the last few checkpoints. This is likely due to the fact that many runners felt that forcing themselves to run for 5-10 miles whilst feeling like they should pull out was more feasible than running for more than 10-30 miles say. Runners may be able to psychologically bypass the feedback from their body (which may feel like it cannot go any further) if the finish line is within a reasonable distance. This raises the interesting question of what a runner considers to be a “reasonable distance”.

Interestingly, despite the conditions being the hardest of all of the races, nobody dropped out of the W100 after the Bury Downs checkpoint at 83.3 miles. This is likely a result of the “4-pronged” layout of this race. When runners left Streatley for the last time, this likely felt like the final stretch despite being another 25 miles until the finish. The Bury Downs checkpoint was close enough to Streatley to allow runners to feel like they could drop out and get back to the base at the Morrell Rooms, but after reaching Chain Hill they probably felt that they were in a position where it would be just a case of retracing their steps to get to the finish. Also, the wind was blowing incredibly strongly against runners on the way out to Chain Hill, so it is possible that runners felt that the return journey would be much easier, being both downhill and with a tail-wind.

Figure 5: Cummulative Number of Non-Finishers At Each Checkpoint

6 Conclusions

The purpose of this analysis was to get a feel for the performance of runners across the four events in the Centurion Running calendar of 100 mile events. Given the large differences in the profile of the four courses and the conditions under which they were run, it would be naive to believe that any such analysis could accurately compare them. However, it is possible to draw some broad conclusions that add to our knowledge of the ultrarunning community:
  1. The number of non-finishers was quite similar for the four races, but as a proportion of the number of race starters it is clear that the W100 had the lowest finishing rate. This suggests that the W100 was the toughest event of 2012. However all four races were tough, and the number of non-finishers in the TP100 shows that there is no such thing as an easy 100 miler.
  2. There is no difference in the age distribution of runners amongst the four races, nor between finishers and non-finishers. Age does not appear to be a significant factor relating to the ability to complete a 100 mile ultramarathon.
  3. In all of the events, the strategy of most of the runners was to head out at a relatively fast pace and gradually slow over time.
  4. In particular, the first marathon distance was often run at a surprisingly fast pace, particularly for the TP100 and for the NDW100. Whether this is the optimum approach or results in an overall slower finish is an open debate, but it seems to have been the approach favoured by the ultimate winners.
  5. There were certain sections of the races that seemed to be particularly difficult, resulting in much slower times for these sections. These were:
    1. (a)  Queen Elizabeth (22.6 miles) to Harting Downs (27.2 miles) on the SDW100 - Not an obviously difficult section but with a large number of small steeper hills and two longer steep climbs making it difficult to settle into a regular pace.
    2. (b)  Cocking (35.1 miles) to Bignor Hill (41.7 miles) on the SDW100 - There are two big steep hils in this section (although interestingly not the steepest hills of the course).
    3. (c)  Box Hill (24.6 miles) to Reigate Hill (31.8 miles) on the NDW100 - Likely due to the fearsome steps of Box Hill.
    4. (d)  Bury Down Out (83.3 miles) to Bury Down Back (91.7 miles) on the W100 - This was due to an incredibly strong wind on the last section along the Ridgeway, and was particularly felt by those at the front of the pack.
  6. Almost no runners will pull out of a 100 mile racer sooner than the marathon distance
  7. The checkpoint at the halfway point is seen by many as a landmark point in the race, and many who need to pull out will make sure that they at least reach this point. Whether this is due to a psychological coping mechanism, assigning significance to certain “milestones” along the way, that darkness is falling presenting an additional stress or challenge, or simply because the halfway point is a convenient place to pull out as it is usually a larger well-manned aid station with access to drop bags, is difficult to say.
  8. Generally speaking, runners will not pull out when they are “close to the end”. In the cases of the SDW100 and NDW100, “close” appears to represent a distance of around 5 miles. For the W100, the distance appears to be longer at around 10 miles, likely due to the way in which the sections of the course were laid out, suggesting that out-and-back courses may affect the way runners consider the distance remaining in an event. For the TP100, many runners were unfortunately
    retired at the very end of the race due to the dangerous weather that came in.
The most interesting things from looking through these data is how far runners push themselves and how they can assign meaning to certain checkpoints over others. Similarly, the fact that the out-and-back nature of the course on the W100 seemed to result in a larger proportion of runners pushing on to the end despite the less than favourable conditions. This may add to the idea that a lot of the ability to complete longer ultramarathons is every bit as much a test of mental endurance as physical endurance.

We hope that this little look at the results over the previous year has been interesting, and perhaps even informative. If you are planning to run one of the events next year, perhaps this will give you a feeling for what to expect on the day, and hopefully this will help you to complete the race. 

No comments:

Post a Comment

Note: only a member of this blog may post a comment.