The Road to Egmond

Feb 22, 2024

Get the app now and experience it for yourself!

Are We Over-tapering? Or maybe under-tapering? We analyzed data from 80 runners preparing for the 2024 Egmond Half Marathon. How many followed the standard advice to reduce training volume by ~50% in the last period before the event? What were the extremes in tapering behavior? And did the taper have more than a marginal impact on the race results? Read on!


A taper has since long become a standard part of the preparation for any endurance (running) event. The general principle is that by reducing our training workload in the week(s) before our goal race, our body can recover so that we’ll start our race with fresh legs without losing too much of the fitness we built up. Although there is some evidence that tapering has benefits, the science is still far from conclusive about the ideal length, workload reduction, or optimal intensity during the tapering period. An interesting research on Strava users showed some significant associations and could be a starting point for discovering causal effects of tapering on race results.


So, what did Vortza users participating in the Egmond Half Marathon do regarding their training load and tapering? We computed the training load by summing the running distance of the last seven days, for every day three months before the event up to (but not including) the race day, for every runner. The development of the training load for this population of 80 runners can be seen in the chart below. 

The “Average” line shows the average weekly training distance of the population. The other lines show the top and bottom 20%. What we can see from this overview is that:

  • Around 10% of the participants barely did any (running) training.

  • Only the ambitious runners seem to follow a training plan with an apparent increase in weekly distance, showing a ramp-up in volume from around two months before the race.

  • On *average* (we’ll get back to that later), there is a noticeable decrease in weekly training in the week before the event, among even the most casual runners.

  • For ambitious runners, there is a significant drop in training volume in the period 6-4 weeks before the race, after which it rises to new highs. This may be explained by using easy weeks in structured training plans but is not confirmed by hard evidence.

Next, let’s look at the taper in some more detail. We computed the taper as the relative difference between the training distance in the last week before the race and the long-term weekly training distance. This was done for every runner and expressed as a percentage. A 50% taper means that a runner’s training volume in the last week was half their weekly average. A -50% “taper” indicates that someone ran 50% *more* than usual in their last week. The distribution of the taper volume reduction is shown below.

On average, the runners seem to follow the existing recommendations and reduce their training by ~40% during race week. However, there is a significant variance, where some runners don’t train at all, while others even *increase* their load in the last week. Why would that be? Were the “slackers” maybe injured after too much training before? Were the “overdoers” feeling guilty about not doing enough so far? For possible answers, let’s look at a detailed comparison between overall training volume and taper, where every dot is an individual runner:  

Here, we can see that everyone who put in 0 km of training in the last week didn’t train too much before, which reflects their casual approach to running. However, some who (even vastly) increased their final-week training volume had already run considerable weekly distances before. Again, the (quantitative) data alone does not explain why some runners increase their training in the last days before their race. Interviews could provide more insight into this behavior.


Now, for the remaining burning question: what’s the effect of the taper on our race results? We must disappoint you as the analysis based on our small-ish sample is inconclusive: in a model for predicting the race time based on overall training volume, sex, age, weight, and taper, the latter has a negligible effect. A larger sample with more data per sub-population could lead to better insights.


If you want to support future analysis like this, please consider donating your (Strava) data by installing our app. If there’s any aspect of running behavior you’d like to know more about, let us know!

Share blog post