KBC Dublin Marathon Analysis

40th Anniversary Edition, on Finishers, Pacing, & Hitting the Wall

barrysmyth
Oct 28 · 8 min read

Yesterday was the big day. The start of the 40th edition of the Dublin Marathon, now one of Europe’s top marathons, and, only for yesterday, it was the largest mass-participation sporting event in world. Just in the nick of time, the weather gods got their act together to provide us with a crisp and bright autumnal day, near-perfect marathon conditions.

The Dublin Marathon began its journey in 1980 when 1,421 runners crossed the finish-line. 40 years later, there were more volunteers that there were inaugural runners, and this year’s 22,500 registrations sold out just 40 days almost a year ago. Between runners, spectators, volunteers, it was estimated that more than 300,000 people came out to participate, spectate, and support Dublin’s milestone marathon.

What better to do on a lazy, recovery, Bank Holiday Monday that pour over the results from the big day…

Participants, Finishers, and DNF’ers

The official KBC Dublin Marathon registration number was 22,500, but according to the race data at the time of writing, 17,973 runners (11,264 men vs 6,673 women) toed the start-line yesterday, meaning there were just over 4,500 (20%) no-shows.

Figure 1 shows the number of starters, based on the different age groups used by the Dublin Marathon: under 19s (U19), Seniors (19–34), 35–39, 40–49, etc. The average age for men and women was just over 42, and the most popular age group, for men and women, was the 40–44 year-old group, with a full two-thirds of runners older than 40.

Figure 1. The number of male and female runners by maximum age group age.

Just over 200 starters did not finish (DNF) ––137 men and 75 women –– or a little more than 1% of runners. The Dublin Marathon dataset includes timing/pacing information at 10km intervals throughout the course and Figure 2 shows when in the race these runners dropped out, based on their last recorded segment time. For example, if a runner only had a time recorded for the first segment of the race (the start to the 10k segment) then they must have dropped out somewhere during the second segment (10–20k). We can see that women were more likely to drop out during the early part of the race, while more men dropped out later in the race; 56% of female drop-outs occured by the 20k mark, compared with just 43% of male drop-outs.

Figure 2. The percentage of DNFs for male and female runners by race segment.

Finish-Times

The average finish-times of runners, based on gender and age group, is presented in Figure 3; in this case we have excluded runners >75 years-old because there are too few of them to generate reliable statistics.

On average the fastest runners, for men and women, were in the 35–39 age group. Men in this age group had a mean finish-time of just under 4 hours (237 mins) while for women it was just under 4.5 hours (267 mins). Older runners were slower to an increasing extent and the rate of slowing with age was similar for men and women, relative to their average finish-times.

Figure 3. Mean finish-times for male and female runners by maximum age group age. The dashed lines correspond to the average finish-times for all male and female runners.

Pacing Profiles

To get a sense of how runners completed the marathon we can look at their pacing over the various sections of the race. Let’s start with the pacing profiles for the race winners (male/female and overall/national) in Figure 4. The male winners (El Goumri and Scullion) and the overall female winner (Gedefa) ran negative splits –– their second-half pace was slightly faster than their first-half pace –– while the national female winner (Cooke) ran a more or less evenly paced race, overall. However, negative splits are not so common among other runners: less 10% of recreational runners ran negative splits, with no difference between males and females.

Figure 4. The pacing profiles of the overall and national race winners.

How then did the rest of the runners pace their race? To compare large numbers of runners, with varying finish-times and paces, we need to use the relative pace of a runner during the various race segments, rather than their actual pace. To calculate the relative pace of a runner we we divide their pace in a race segment by their average race pace. For example, if a runner had an average pace for their marathon of 5 mins/km, and they ran the first 10k at 4:30 mins/km, then their relative pace for this segment would be 0.9 (4.5/5), indicating that they ran 10% faster during the opening 10k relative to their average race-pace.

Figure 5, shows the relative pacing profiles for runners based on (a) gender and (b) finish-time range. On average, runners ran the first half of their race about 5% faster than their average race-pace (relative pace = 0.95) before starting to slowdown; 42% of women and 35% of men ran their first 10k as their fastest segment, rarely the best plan in a marathon.

Men paced their race less evenly than women. During the first 30km their relative pace was faster than that of women, but men slowed more than women did from the 30km mark onwards. For example, during the slowest segment of the race (30–40k), men ran about 7.5% slower (relative pace = 1.075) than their average race-pace, while women ran just 5% slower (relative pace = 1.05).

Figure 5(b) shows similar pacing profiles but based on three finish-time ranges: less than 3 hours, 3–5 hours, and more than 5 hours. It should be clear that faster runners tended to pace their race much more evenly than slower runners. The relative pace of those finishing in under 3 hours was consistently close to their mean race-pace for the majority of the race, only slowing during the final 2.2km. For slower runners there was more varied: slower runners started faster but finish slower, whereas faster runners started as they meant to finish. For example, runners finishing in more than 5 hours tended to start their race about 10% faster than their mean race-pace, and slowed to about 10% slower than their mean race-pace towards the end of the marathon.

Hitting the Wall

Slowing significantly during the second half of the marathon is associated with that most iconic of marathon hazards, hitting the wall. When runners hit the wall their pace deteriorates significantly and rapidly over a sustained period of time, often until the end of the race. The effect is usually associated with the 30–40km portion of the race, and it is commonly thought to be due to a combination of poor pacing, inadequate in-race nutrition, or disrupted training. Of course we don’t have access to information about nutrition or training, but we can estimate whether a runner hits the wall by looking for significant slowdowns during the 30–40km segment.

To do this we calculated the relative wall pace of a runner by dividing their 30–40km pace by their average first-half pace. For example, if a runner had a 30–40km pace of 6 mins/km and they ran their first half at an average of 4:30 mins/km then their relative wall pace was 1.33 (6/4.5), indicating a 33% slowdown. As to the degree of slowdown that is indicative of hitting the wall? Mostly its a case of “you’ll know it when you see it,” which isn’t very helpful as a data science principle, but for this analysis we will use 1.33 as the critical threshold. In other words, if a runner has a relative wall pace of >1.33 then we deem them to have hit the wall; minor variations in this threshold don’t change the trends that we find in any material way.

How did our Dublin runners do? Figure 6(a) shows the percentage of male and female runners who hit the wall based on their final finish-times. Overall, just over 10% of men hit the wall this year, compared with less than 5% of women. There is a sharp increase in the percentage of runners hitting the wall as finish-times increase. 32% of male runners, who finished around the 5-hour mark, hit the wall, compared with only 8% of females; in other words, 4-times as many 5-hour male finishers hit the wall.

If that wasn’t bad enough for us men, not only did more of us hit the wall, when we hit it, we hit it harder, regardless of finish-time; see Figure 5(b). Men, on average, slowed by just over 45% (relative wall pace = 1.45) when they hit the wall, compared to 43% for women.

The Sprint Finish

Finally, as we reach the end of the marathon, a brief look at how runners dealt with the final 2.2km dash to the finish-line. In Figure 5(a), we can see that, on average, runners did speed-up during the final segment of the race, at least relative to the 30–40k segment, women moreso than men. Overall 62% of women and 47% of men managed tospeedup during this final segment (Figure 7), but very few –– 13% of women and only 9% of men –– ran their fastest pace of the race during this dash to the finish. Although fewer men managed to achieve this final speedup, when they did they improved on their 30–40k pace by more than 6%, whereas those women who achieved a final speedup only improved their pace by just over 5%.

Figure 7. (a) The percentage of runners who speedup during the final 2.2km of the marathon, relative to the 30–40k segment. (b) The degree of speedup for those runners who speedup during this final segment.

Conclusions

It is time to wrap up. We’ve looked at some of the key statistics from the 40th Dublin Marathon to make the following observations:

  • Just over 20% of registrants failed to start the race and of the 17,937 starters, just over 200 (about 1%) failed to finish, with most of these (approximately 80%) dropping out before the 30km mark.
  • Most runners were 40 and older but the fastest average times were achieved by those in the 35–39 age group, with average finish-times slowing consistently with age after that.
  • While the race winners skewed towards negative splits, running the second-half of the race faster than the first-half, negative splits were far less common in general, with less than 10% of runners completing a negative split.
  • Just over 8% of runners hit the wall yesterday (11% of men and 4% of women) and, in general, men a far more likely to hit the wall than women, and they slow more when they do. What might be termed ‘peak wall’ occured for men and women who finished in about 5 and a half hours.
  • At the end of the marathon many runners (47% of men and 62% of women) managed to dig deep to find a final burst of energy to speed-up by 5–6% during the final 2.2km of the race.

Running with Data

A collection of articles at the intersection between data science and endurance running.

barrysmyth

Written by

Professor of Computer Science at University College Dublin.

Running with Data

A collection of articles at the intersection between data science and endurance running.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade