Marathon Pacing Part 1: New Research Investigates The Challenges of Good Pacing

marathon pacing
Marathon pacing is notoriously difficult. There are so many factors that affect our ability to pace over 26.2 miles, especially if we are trying to run the distance as fast as we can. Even the top elites regularly get it wrong. It’s also a rather poorly understood area. So how can we learn more about it, and how can we (as runners) benefit from more research into this area?

We have two connected articles for you, to explore and learn from the new research findings. These are:
Part 1: New Research Investigates the Challenges of Good Pacing
Part 2: A Look at The Physiology of Pacing


PART 1: New Research Investigates the Challenges of Good Pacing

A brand new research project from Anglia Ruskin University, and supported by data from The Flying Runner community, is providing an intriguing window into how well runners predict and pace marathons. A large group of nearly 800 runners at London 2015 were asked to predict their marathon finish time in the three days prior to the event, and researchers then looked at how they actually paced their race.

The results are fascinating. They give an insight into the large-scale way in which the majority of runners appear to pace poorly, and the inaccuracies involved in predicting likely performance. On the flip side, there were early indications of how some groups of runners appear to be better predictors than others, and an opportunity to explore why, and whether this means they are pacing more effectively.

Asking these questions starts to reveal information that we can all potentially learn from, such as how factors like age and gender affect predictions, the positive effects of experience, some of the differences between faster and slower runners, and the combined impact of psychology and physiology.

None of this gives us easy shortcuts to getting pacing “right”… In fact it often throws out more questions than answers. But as with all areas of learning, this is part of the fun!

This article provides a surface-level overview of some of the research findings. The research itself will be published in the next few months. Most importantly, neither this article nor the research is about giving coaching advice: It’s not an attempt to define how marathon runners “should” pace, but a look at what’s actually happening and how we might learn from that.

Why is good marathon pacing so hard?

Here’s the trouble: it’s not easy to practise marathon pacing. It’s not like running a 5k, where you can turn up at parkrun every week and try out different approaches. Even most elite athletes can only tolerate around three full-effort marathon races each season. And after the massive training operation involved in getting to the marathon start line, most of us don’t feel inclined to take a risk with how we run the race. So we don’t often experiment with what might work best for us individually to pace a 26.2-mile race well.

Add in to this factors such as the specific conditions on race day, or your variable training condition year-on-year, and suddenly the goal of planning a very precise pacing strategy becomes more and more difficult to pin down.

Ability to pace has also got a lot to do with the fact that we are humans and not robots. Emotional and psychological factors are inextricably linked to performance. This is not an easy area to study: it’s less quantifiable than the “hard science” elements of physiology, but making more inroads into researching this area and connecting it with performance adds a whole dimension of interest that can’t be ignored in the overall picture of marathon running.

What’s this new research all about?

So it’s clear that there is plenty of room for more learning, discussion and research. There are some big questions that can be asked to help us understand the most efficient way to pace a marathon, such as:

  • What do we know about how the body responds to endurance effort, and how can we use that information to get better at pacing?
  • How do we find people who are very good at pacing their races (regardless of how fast or slowly they are running)?
  • What are these good pacers doing? How are they running their races?
  • Is it possible or viable to use that information to improve your own approach to pacing?

It’s exactly these questions that Dr Dan Gordon and his team in the Sport & Exercise Sciences Research Group at Anglia Ruskin University have been addressing.

The “marathon time prediction” survey this spring was kicked off by us at The Flying Runner, initially as a bit of a fun competition for the London Marathon Expo. As we planned this, we realised we’d have an interesting data bank on our hands, so asked Dan if he’d be interested in using the data for academic research purposes. He was – it fitted perfectly with his current work on pacing. We gathered data from over 1,000 runners taking part in marathons in spring 2015, asking simple questions: predicted finish time, age, gender, experience of running marathons.

Dan’s team then analysed in detail 778 of the London runners who were in the survey, looking at their actual 5k splits, and creating groups by the factors of age, gender and experience.
The vital link was asking people to predict their finish times. The majority of predictions were made within three days of running the same race in London. Using this information, Dan’s team was able to understand how closely people actually, ran compared to how they expected to run, and therefore to look at how well they were pacing.

Dan’s team also took the data of the entire body of 38,000 runners who ran the London Marathon in 2015, and looked at their 5k splits. This was to ensure that the research group were not different in any way from the rest of the runners, not self-selecting or behaving differently.

A few of the interesting findings

There are some fun findings as well as ones which can potentially be applied to improving pacing.

Stick your neck out and venture a view on these questions:

  • Do most runners a) go at an even pace throughout, b) speed up through the race, c) slow down a bit towards the end, or d) slow down a lot throughout the race?
  • Do you think women are better than men at predicting their finish times?
  • Do you think younger runners are likely to be better predictors than older runners?
  • Are more experienced marathoners better predictors than less experienced?
  • And are faster runners likely to be better predictors than slower runners?

Think you know? Here’s what the research showed…

The whole group

In the results for the whole group of 778 runners, the pacing profile showed a steady slowing of speed throughout the whole race. In the graph below, the red line shows normalised speed, with vertical error bars showing the range of results around the average. (Normalised speed means that the average speed of the runner is taken as 100%, so that it’s possible to compare across different speeds.)

marathon pacing

Normalised speed across the whole group of 778 runners.

So there’s a very clear drop of speed for the whole group across the race. This is replicated (to a greater or lesser degree) across almost every single group when they are split out by gender, age and experience or the combinations of those factors.

Are men really better than women?

Well (annoyingly for us girls!) yes, when it comes to predicting marathon times. But only slightly.

pacing your marathon
As a group, men were in fact more accurate predictors than women. The average prediction from males was 419 seconds away from their actual finish time, or just short of 6 minutes. Females, on the other hand, were an average of 531 seconds adrift, or just under 9 minutes.

In terms of speed, interestingly, women overall were starting at a faster normalised speed than men but also dropping off less right at the end of the race (see graph below – green line shows women, blue line shows men).

marathon pacing

Normalised speed by gender. Green line shows women, blue line shows men.

Does older mean wiser?

Older runners, on average, are more accurate predictors than younger runners. Runners in the 70-79 age group were the most accurate (only 31 seconds away on average), and runners in the under-20 age group were the least accurate (1107 seconds on average).

Here’s a graph showing the spread by age groups (as grouped by the London Marathon). Personally when I look at this, I can’t see that much of a meaningful difference between the ages, yet it’s still striking to see how consistently everyone is dropping off the pace throughout the race.

marathon pacing

Normalised speed by age groups.

Dan’s team are investigating the idea of “global experience”, in other words whether life experience in general helps with decisions about pacing. Knowing yourself better, understanding common pitfalls, having improved self-control, and having more experience of coping with stress and difficult conditions could all be relevant factors. How will that be balanced against inevitable decline in physical robustness with age? It will be interesting to see if the researchers find more to differentiate the age groups when they publish the research papers.

Does marathon experience matter?

In short, yes, absolutely. The number of marathons run previously could be one of the most meaningful differentiating factors. It’s not always clear to see the difference in prediction accuracy among those who have run between 1 and 10, but the differences between complete novices and experienced runners with more than 10 under their belt are certainly clear.

marathon pacing
More experienced runners on average are better predictors than less experienced runners. If we place all those who had run fewer than 5 marathons in one group, and those who had run 5 and above in another group, the less experienced runners ran actual times of 481.8 seconds (just over 8 minutes) outside their predictions, whereas the more experienced were 238.6 seconds (just under 4 minutes) outside their predictions. This equates to a 49.4% difference between the two groups.

But here, in my view, is where it starts to get even more interesting. The scattergraphs here show the difference in predictions between runners with no previous experience, and those who had run more than 10 marathons previously. The black line shows where a perfectly accurate prediction would sit, and each red dot shows a runner, with predicted times on the vertical axis and actual finish times on the horizontal axis. So a dot above the line shows a faster actual time than predicted time.

No previous marathons – this shows a looser spread of predictions, forming a kind of ice-cream cone shape with faster runners showing better prediction ability than slower runners.

marathon pacing

Prediction accuracy of runners with no previous marathon experience.

More than 10 marathons – this shows a tighter clustering around the line, less dispersed, across the slower paced groups as well as faster.

marathon pacing

Prediction accuracy of runners who had run more than 10 previous marathons.

Looking at this in terms of how far off novices were with their predictions compared with those with 10 or more previous marathons, it’s clear to see the difference in the chart below:

marathon pacing

Comparative prediction error of runners with no previous marathons, and more than 10 previous marathons.

The faster the better?

marathon pacing
Faster runners as a group are significantly better at predicting than slower runners.

The fastest 5% of runners in the survey predicted their finish times (which averaged 2 hrs 43 mins) to within 71 seconds, so were about 0.7% adrift. That’s compared with the slowest 5% (averaging 6 hrs 20 mins) who were 47 minutes off, or 12.4% difference from prediction.

What does that look like?

This graph shows the fastest 25%, with the black line showing predicted times:

marathon pacing

Prediction accuracy of the fastest 25% of runners.

Compare that with the slowest 25%:

marathon pacing

Prediction accuracy of the slowest 25% of runners.

The top chart is very obviously much more tightly clustered around the black line, whereas the bottom chart is very dispersed. The top 25% were much more accurate at predicting.

Even more interestingly, here’s what that looks like in terms of pace across the race:

marathon pacing

Normalised speed of the fastest 5% (black) and 25% (red) and the slowest 5% (green) and 25% (blue).

The black and red lines show the fastest 5% and 25% runners, and the green and blue lines show the slowest 5% and 25%. It’s very clear to see that the faster runners ran a more evenly paced race throughout, while the slower runners ran a significant positive split. And you can see that the very bottom 5% started off way too fast and paid the price.

Explore the data yourself

Intrigued by the findings? We’ve created a tool to enable you to explore the data yourself. You can look at the pacing patterns of different groups of runners, and choose how you want to view the groups by finish time, age, gender, experience, and prediction accuracy.

You can, for example, compare runners with a predicted and/or actual finish time close to your own target pace, and to select runners “like you” in age and experience, and look at how they paced their running.

Click through to the “Explore Marathon Research Data” tool here.

Key messages

All of the findings will be explored in much more depth in the research papers when they are published. For now, the key message is that exposure to race conditions (experience) is key to improved pacing, prediction and performance. Pacing is a learnt response but cannot just be acquired through training, racing is essential.

We’ll take a look at the physiology and psychology of this in Part 2.

What next for the research?

Dan is working with colleagues to run similar studies at marathons in Paris, Berlin and Florida. In 2016, Dan’s team also plans to extend the research at London with richer information, involving a group of 200 runners who will have VO2 max testing and physiological profiling, will predict their finish times, and will wear a heart-rate monitor in the marathon.

We’re hoping that The Flying Runner community will get involved!

If you have a place in the London Marathon in 2016 and would be interested in taking part in the research (including free VO2 max testing) register your interest here

Read on for
PART 2: A Look at The Physiology of Pacing

Subscribe to the Flying Runner bi-weekly newsletter for more inspiring articles and promotions on our beautiful gifts!

We promise we'll never spam you or share your details with anyone else.

You've successfully subscribed! Please validate your email address by clicking the link in the email we just sent you. Thank you :)

Leave a Reply

Your email address will not be published. Required fields are marked *