Performance curves for single activities

, Posted in Features

With the latest update, Runalyze can now display power, pace and heart rate performance graphs for individual activities. These graphs show the best or maximum performance for all possible time periods – from one second to the entire activity duration.

This type of diagram should be familiar to cyclists in particular, who know this diagram as the “Power Duration Curve” – in that case, however, only for power. We go one step further and offer this diagram also for pace and heart rate.

One thing to keep in mind: These diagrams “only” show the performance achieved within a single activity – and not what the athlete could theoretically achieve. In addition, of course, only contiguous sections are taken into account. Therefore, zigzag curves (as in the 2nd example for an interval training) are also absolutely valid.

Example: relatively constant effort

Power curves for an activity with a relatively constant effort.

In this case, power and pace each show a clear peak for up to just under a minute before the average performance decreases only slightly over the rest of the course. This is a fairly typical picture for an activity with a relatively constant effort.

The data can be interpreted (in terms of power) as follows:

  • A maximum power of about 320 W could be achieved.
  • A power of over 300 W could be held for a maximum of approx. 30s.
  • A power of over 280 W could be held for a maximum of about 3 minutes.
  • The maximum power for any period between 5 minutes (275 W) and one hour (250 W) differs only slightly.

Example: Interval training

A somewhat different picture usually emerges for interval training, where a jagged curve usually emerges. The fact that the best performance for 5 minutes (278 W) is greater than for 3.5 minutes (265 W) is not a contradiction.

Performance curves for an interval training of 10x400m (in about 1:30, 2′ rest).

In this example, 400m intervals (uphill) were run with the fast sections in about 1:30 and the slow sections in about 2:00. As a result, the curves drop significantly until about 3:30, since a segment of this duration can contain a maximum of one fast section. Only then is there a renewed increase until about 5:00. This best performance over 5 minutes now includes two fast sections and one slow section (2×1:30 + 1×2:00 = 5:00). Analogously, it goes on and on for longer sections, so that the waveform shown here is created.

This makes the curves very different from a histogram, which would show about 15 minutes of highest load (> 300 W) for those 10 repetitions – and not the information whether this load was done in one piece (tempo endurance running) or in sections (interval training).

Outlook

Of course, we would like to offer more features based on these individual performance curves – especially the view of the performance curves for e.g. the last 90 days or any period (e.g. comparison 2021 vs. 2020). These features are planned and should follow this year if possible. But of course they need a lot of server capacity and will probably only be available for paying users (in the upcoming premium level). In the current preparations for this, for example, we are talking about server processes that are already running for several days.

3 thoughts on “Performance curves for single activities

  1. Hi Runalyze team,

    these graphs sound very interesting.

    I have one question to better understand the graphs:
    If I looked at such a graph, I would expect that the quantity of interest stays strictly above the y-value for the time intervall shown on the x-axis. So, the graph would show the maximum min-value of any 5-minute intervall during the training. This, inevitably, leads to a monotonously decreasing graph.

    As you lay out in your example, you are using the average value over the time intervall and, thus, the graph can become non-monotonous.

    Maybe, this hint could lead to a second set of graphs you can implement with only minor changes to the algorithm behind it.

    Cheers,
    Konstantin

    1. You are right that the ‘maximum min-value’ method would lead to monotonously decreasing graphs. I’m not 100% sure whether other competitors use ‘maximum min-value’ or ‘maximum avg-value’ as we do. Not sure if that’s a difference between cyclists, that may be interested in ‘how long can I produce at least X Watts’, and runners, that may be more interested in (maximum) averages?

      When thinking about the pace data, the raw signal is too noisy to calculate correct ‘maximum min-values’. And to be honest, I, personally, am not interested in ‘how long can I run 3:30/km or faster’, I’m interested in ‘how lang can run 3:30/km in average’.

      As the calculation of those graphs is not done in a few miliseconds, I guess we can’t provide those graphs very easily for ‘maximum min-values’, but we’ll keep that in mind!

Leave a Reply to Konstantin Cancel reply

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

Start your analysis now

Create a free account and start analyzing your activities.

You can also sign up with your Facebook or Google account.

Questions?

Any questions left? We are there for you. Ask your questions in our Forum or Facebook Community.

Need help?

Do you have troubles with your account?
Contact our Support.