The Data of the Chicago Marathon

Lessons learned from 400,000 Chicago marathoners (2005–2016)

TLDR;

  • An analysis of more than 400,000 Chicago Marathon results from 2005–2016.
  • Athlete gender has a greater impact on various performance metrics (speed, pace variation, hitting the wall) than age.
  • While men are faster than women, women are more disciplined. Women enjoy more even pacing and hit the wall far less than men.
  • Faster runners are more likely to be repeat marathoners. Men are more likely to repeat than women and older runners repeat more than younger runners. And the experience gained from repeat marathons has a positive impact on performance in terms of speed, pacing, and the tendancy for athletes to hit the wall.
  • This is part of a series of posts that I am writing as part of an ongoing analysis of marathon data. So if you are interested in this sort of thing then you will find a growing number of articles here.

Introduction

As I am writing this post the results are just in for yesterday’s 2016 Chicago Marathon, the 39th edition of one of the world’s most popular marathons. This year saw 41,357 runners toeing the start-line in near-perfect race conditions and the event did not disappoint with its fair share of photo finishes, returning champions, and new winners. As usual most of the commentary and analysis has focused on the elite field — and well done to those super-humans that most of us can only marvel at from afar — but what of the other 40,000 or so runners thet compete every year? What can we learn from this much larger group recreational runners as they wind their way through the Windy City?

As it happens I’ve been looking at the data of the Chicago Marathon (among others) for the last few months as part of a larger personal project to see what the data can teach us about recreational marathoners. Finisher results for Chicago from the period 2005–2016 are readily available online, providing access to a treasure trove of data, from numbers of participants, gender, age ranges, finish-times, even 5k split-times. All in all I have collected, cleaned, and compared more than 416,000 finisher records and in what follows I will try to summarise some of the more interesting findings. Note that for ease of analysis the following does not include an analysis of the wheelchair races.

Who Runs the Marathon Anyway?

What type of person is willing to put themselves through 26.2 miles of hurt, the months of high-mileage training that come before, not to mention the pain of injury or the agony of self-doubt that defines the marathon? Is the marathon just for the committed few? The life-long athletes? The over-achievers? Not so, it seems. When we look at the data we see all sorts of participants, young and old, male and female, fast and slow. The modern marathon is truly a mass-participation event, not to mention one of the few occasions where recreational athletes can toe the line with the best of the best.

Let’s look at the breakdown of participants for Chicago. In much of what follows we will divide participants by gender (male versus female) and age (under 40s versus over 40s). In the bubble graph below we show the average number of participants per year for each of these divisions for the years 2005–2015; that is, excluding 2016. We can see that there have been an average of 34,420 runners over the period 2005–2015. Approximately 56% of participants are male and 60% are under 40.

By comparison, the data for this year (2016), with over 40,000 finishers, is presented below and shows a greater balance between the genders and ages: 54% are male and only 56% are under 40.

Finish-Times or Finish-Lines?

The saying goes that its about finish-lines not finish-times, but still many of us are interested in getting to that finish-line as fast as we can. Below is a graph of the average finish-time for each year, measured across all participants by gender and age-group. We also separately show the male and female winning times for each year.

Perhaps the most striking feature of this graph is the difference between the average finish-time and the average winning time. Not surprising to most of us but striking nonetheless. It highlights a typical difference more than 2-hours between the winner and the average marathoner.

As expected the winning times for men and women have not changed by much year on year. The male course record was set by Dennis Kimetto in 2013 with a time of 2:03:45 and the female record by Paula Radcliffe in 2002 (not shown here) with a time of 2:17:18. That said, the last two years have seen male winning times that are slower than any since 2007.

The average finish-times calculated across the entire field, are less stable, largely owing to a significant blip in 2007 as a result of a major heatwave that saw October temperatures hit the high 80’s, leading race organisers to halt the race late in the day.

It is interesting to note how gender has a greater influence on finish-times than age. For example, the difference between the finish-times of men and women varies from about 30 minutes while the difference in finish-times between the under 40's the over 40's runners varies by less than 10 minutes, in favour of the younger runners.

The graph below takes a closer look at finish-times by examining the average percentages of runners crossing the line for different finish-times. We can see a sharp rise in the number of people crossing the line up until the 4-hour mark when just over 20% of men and women finish. Before the 4-hour mark there are more men than women finishing. After the 4-hour mark the proportion of finishers starts to fall off again, with a greater proportion of women finishing than men.

Once again we can see how gender has a greater influence on finishers than age. In fact there is very little different between the relative proportion of youner and older finishers crossing the line.

It’s the Pace that Kills!

Finish-time is just one measure of performance for marathon runners. How a runner’s speed or pace changes over the course of a race — their pace variation — is also of interest. Pace variation is often seen as an indicator of ability and, generally speaking, less variation in pace is considered to be a sign of a more disciplined and able runner.

We can compare the pace of participants during the first and second half of the marathon to calculate the percentage of pace variation. The above graph shows how this pace variation changes for men and women, younger and older runners, as finish-times increase.

Clearly the degree of pace variation changes with finish-time, regardless of gender or age. Faster, more able, runners exhibit less difference between first and second-half pacing than slower runners. The fastest runners exhibit pace variation of less than 5%, for example, but for 360-minute finishers, the first and second-half pace of male runners varies by more than 25% before levelling out. A similar effect is seen for female runners, albeit with a lower pace variation for a given finish-time before leveling out at about 20% by the 360-minute mark. This lower pace variation for females suggests that women are more disciplined than men.

Once again, age has a more modest impact on pace variation. There is very little variation due to age for women until after the 300-minute mark. For men, age plays a more significant role at faster finish-times. What is interesting is that for both men and women, older runners appear to be more disciplined (lower pace variation) than younger runners. Is this the wisom of age?

Split Personalities

As we approach race-day most of us will start to think about our pacing strategy. Should we go all-out early on to compensate for an expected slow-down in the second half? Or should we take it easy during the early stages and try to speed-up later? Or is it best to keep our pace as steady as possible, as the fastest runners seem to do? The first of these pacing strategies, going out fast and slowing in the second half of the race, is called a positive split. The second, starting slow and speeding up in the second half, is called a negative split. And the third, running the same pace throughout is called an even split. So, what types of pacing strategies are at play in the Chicago Marathon?

In the above we graph the percentage of participants who perform positive, negative, and even splits based on their finish-times. Across all finish-times, the majority of participants (approximately 85%) perform positive splits; their second-half is slower than their first half. Less than 5% run negative splits, completing a faster second-half than first-half. And just over 10% manage to run an even split; note that for an even split the first and second-half times must vary by less than 2%. However, the graph shows that these different split-types are not evenly distributed across finish-times.

Generally speaking, among faster runners there are relatively more negative and even splits (and therefore fewer positive splits), when compared to slower participants. This is consistent with the prevailing wisdom that even and negative splits correlate positively with ability and better race performance. As we move through the finish-times, the percentage of positive splits grows steadily to over 90% by the 300-minute mark. It’s worth noting that women are consistently better a producing an even split than men for a given finish-time beyond 150 minutes, whereas men produce consistently more positive splits.

Calculated over all participants, the average finish-times for even, negative, and positive splitters is 242, 256, and 280 minutes, respectively, suggesting that an even pacing strategy is correlated with faster finish-times whereas a positive pacing strategy is correlated with slower times. A word of caution however: correlation is not the same as causation and so it is unwise to conclude that forcing an even split, for example, will guarantee a better finish-time.

Hitting the Wall

The dreaded wall is the stuff of marathon nightmares. One minute you are motoring along, gliding through the miles and feeling strong, and the next your legs feel like concrete, your breathing becomes laboured, and the will to continue is no more. You’ve hit the wall! The prevailing wisdom is that bonking is caused by the depletion of the glycogen stores in our liver and muscles — the primary sources of fuel for most marathon runners — and avoiding it requires dedicated training, a sensible pacing strategy and careful race nutrition.

When runners hit the wall or bonk (to use the more colourful language of marathon lore) they tend to do so after the 20-mile mark, but the literature remains somewhat undecided on the extent of the pace changes experienced. For some, pace can slow by 20–25% while others report slow-downs of 50% or even higher. While our race data doesn’t tell us directly whether someone actually hits the wall, we can make an educated guess by looking for participants who experience a significant slow-down in the second-half of their race. But what is a significant slow-down?

For our purposes we focus on participants who slow down by at least 33% in the second-half of the race. For instance, we will consider a runner who runs the first-half at 6 minutes per mile and then slows to just over 8 mins per mile in the second-half (a 33% slow-down) to have bonked or hit the wall. But if this runner slows to 7 minutes per mile (a more modest 16% slow-down) then we will not consider them to have hit the wall; they have slowed because they have fatigued but they have not bonked.

According to the data, using this 33% threshold, just under 10% of all Chicago marathoners bonk, but as with split-types, the likelihood of hitting the wall changes with finish-times and therefore ability. The graph below shows the percentage of participants who bonk for different finish-times; as before we also show how this changes with gender and age.

It is clear that the tendency to hit the wall is strongly influenced by finish-time (and therefore ability). The fastest runners rarely hit the wall; the incidence of bonking among runners finishing within 3 hours is less than 1%. Beyond the 3-hour mark, however, things don’t look so good. 5% of 4-hour finishers hit the wall and for participants who finish between 5 and 6-hours 15–20% hit the wall.

But that’s not all. Once again gender appears to play a critical role, much more so than age. Male runners are 3–4 times more likely to hit the wall than women, at least for some finish-times. For example, about 25% of 300-minute male finishers hit the wall while women with the same finish-time do so only 7% of the time. Age only has an impact afer the 300-minute mark. Then older runners are increasingly less likely to hit the wall than younger; the wisdom of age once again.

As with pace variation, the incidence of bonking tends to flatten out for slower finish-times beyond the 360-minute mark, and even begins to fall. These slower finishers are presumably under less extreme levels of exertion (notwithstanding their 6+hours out on the course). They also have more time to re-fuel effectively during their race to protect against hitting the wall.

Never Again? Not a Chance!

There is something about marathon runners that makes us gluttons for punishment? The months of training and the 26.2 miles on the day are rough enough, but on top of that some of us will repeat the experience year after year.

In our Chicago Marathon data we find that just over 19% of participants complete more than one marathon (at least, during the 2005–2016 period) and on average these repeaters complete an average of 3 marathons during this period. Men more likely to repeat than women — 22% versus 16% — and these men ran 3.17 marathons on average, compared to 2.76 marathons for the women.

Unsurprisingly, ability (in terms of finish-time), has a significant bearing on whether a participant is likely to repeat his or her marathon experience. Faster runners are much more likely to be repeaters than slower runners as shown below. For example, just under 50% of runners finishing around the 180-minute mark are repeaters, but only 30% of those with 270-minute finish-times repeat.

It is interesting how, this time, age does have a greater influence than gender: while men are somewhat more likely than women to repeat, older runners are a lot more likley to repeat than younger runners. But actually this is to be expected since older runners will have had more opportunity to run multiple races than younger runners.

Practice Makes Perfect

Do these repeat marathoners enjoy performance benefits? The graph below shows how both finish-times and pacing improve with the number of repeats. Regular marathoners tend to run faster races with more even pacing.

In addition, below we see that the percentage of people hitting the wall also improves with experience. At a 33% bonk threshold just over 10% of first-timers will hit the wall whereas those who have run 5 marathons will hit the wall only 5% of the time.

Lessons Learned

In this study we have analysed the results of more than 400,000 runners, from the last 12 years of the Chicago Marathon, from 2005–2016, examining various aspects of performance, from finish-times and pacing to hitting the wall and the benefits of experience. We have found that although men are faster than women, women appear to have greater race discipline; they present with more even pacing and they hit the wall less often. We have also seen the benefit of experience and how repeat marathons can improve finish-times, pace variation, and the rates at which runners hit the wall.

This is one of a series of posts that I am writing as part of an ongoing analysis of marathon data. So if you are interested in this sort of thing then you will find a growing number of articles here.

Appendix — Male/Female Performance Adjustments

An argument can be made that the above analysis is not a like-for-like comparison between men and women. There are physiological reasons why men are capable of faster finish-times than women and so comparing a male runner with, for example, a 180-minute finish-time to a female runner with the same finish-time is not a fair comparison. A 180-minute female finisher is likely putting in a much greater effort than male runner with the same finish-time.

This calls for an adjustment to be made in the finish-times of women to facilitate a fairer comparison. But what adjustment should be used and will it effect the results? One way to look at this is to calculate the ratio of average male finish-times to average female finish times and use this as an adjustment factor. For our Chicago Marathon data male finish-times are, on average, 91% of female finish-times. So men are 91% faster than women and to get a like-for-like comparison between men and women we can adjust female finish-times by this degree. Thus, a 200-minute female finisher is compared to a 182-minute male finisher.

Below we show the pace variation graph and the percentage of bonkers graph using these adjusted female finish-times. The net effect is that the female graphs have effectively shifted left an amount that corresponds to this 91% adjustment and while this reduces the difference between the genders it does not eliminate it. Far from it. Women still present with much better pace discipline than men and they hit the wall far less often, especially after the 180-minute finish-time.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.