# Useful tools to measure training progress (or lack thereof) in endurance sports

Oct 6 · 10 min read

Using HRV4Training Pro to go beyond standard training load analysis.

In this post, I’d like to show how we can monitor progress (or lack thereof) in endurance sports using tools such as aerobic efficiency and cardiac decoupling analysis.

I will also show how training adaptations resulting from different training stimuli can be captured by these tools better than using standard training load analysis metrics such as chronic training load.

# Training load analysis and its limitations

Training load analysis is a simple and widespread way to summarize training duration and intensity in a single number (or graph). The analysis is so standard that pretty much any tool out there will provide some version of it. This analysis is mostly based on the work of Calvert, Banister, Savage, and Back (full paper here).

We can break down the analysis as follows:

1. Determine a basic unit for the training load analysis: for example Relative Effort in Strava, TSS in TrainingPeaks, or any metric that works for you, in HRV4Training (even just RPE or RPE x Duration are good candidates). The goal of this basic unit is to capture both duration and intensity of the workout, you can learn more here)
2. Once we have our basic unit, we compute a moving average for the past X and Y weeks, where X >> Y, so that X represents your Chronic Training Load (CTL) or the load you are used to take based on your historical data (typically 6–8 weeks), and Y your Acute Training Load (ATL): or the load you have been taking in the past few days or week (typically 1–2 weeks). More often than not, CTL is also called fitness, as the idea is that the more load you can take, the fitter you are, and ATL is called fatigue, as a steep increase in load will get you tired.
3. Finally, we can compute additional useful parameters from the ATL and CTL data, for example readiness to perform (freshness) or injury risk. Intuitively, the idea is that you should not load too much if you are not used to certain training loads (so a high ATL with respect to your CTL should be avoided).

Wonderful. What’s the problem with this?

While I find readiness to perform and injury risk informative when properly interpreted (read: do not obsess over anything that is based on a threshold, but try to look at trends and how things are changing over time), CTL can be a poor representation of cardiorespiratory fitness or of performance.

Additionally, I find it often dangerous for athletes to focus on metrics whose only goal is “more is better” which is hardly how the human body and performance work.

The goal of training is to provide the best stimulus at any given time, so that the desired adaptations can occur, leading to improved performance. What’s the exact stimulus that we need to apply depends on a variety of things, starting from our history, our current fitness level and of course our goal event.

Training load analysis does not take into account how your physiology changes in response to training. The only input to this model is how you train (duration and intensity) with no consideration for how your physiology is changing (if it is changing at all) as a result of training.

# Case study: my training plan

Before we start looking at different types of analysis to measure progress more effectively, let’s look at what was the training stimuli provided, as this is of course how we try to drive certain adaptations.

I have been preparing the New York City marathon on a low running volume diet as I was coming back from an injury. My training plan comprised of two main phases, a generic (or base) phase in which I mainly tried to rebuild fitness after the injury, and a specific phase in which I focused more on the demands of my target event (the marathon).

During the generic phase, I have done much low-intensity running and 1–2 sessions per week of short intervals (1–3 minutes x 6–10 reps, really hard). During the specific phase, on the other hand, I had to build towards long runs and work on my ability to sustain a lower intensity effort for longer.

Due to my injury firing up from time to time, I could not increase weekly mileage and had to sacrifice intervals sessions during the specific phase. While ideally, you’d be able to maintain what you have developed during the generic phase, the many days off I needed to take, lack of speedwork and overall low volume meant that I was getting slower and losing some aerobic fitness over time. Yet, I was also improving from other points of view, for example in my ability to run longer.

These aspects are not well captured by the training load analysis shown above, where things seem to improve, level off, and then improve again over time.

The limitations of the training load analysis are what motivated me in the first place to look for alternatives that could better capture progress in different endurance abilities.

Let’s look at some alternative ways to process workouts data to capture changes in performance.

# Aerobic efficiency and cardiac decoupling: what can they tell us?

Aerobic efficiency relates to your ability to sustain a given workload. Good endurance athletes tend to have high aerobic endurance, meaning that they can sustain a relatively high workload (for example pace or power), at a relatively low effort (typically measured in terms of heart rate). For the specifics of how this is computed, refer to this article.

There are many factors that can affect the relationship between pace (or power) and heart rate. A few examples are: running or cycling on trails or difficult terrains, (which reduces pace and makes your data not really representative of your fitness), very short workouts where heart rate does not reach a steady-state, environmental factors such as hot days or training at altitude, etc. — the list goes on. While many of these parameters are simply impossible to account for, what we can do is give you more control over what data is used to track changes in aerobic endurance. In particular, via the panel below you can filter workouts and environmental factors so that the resulting data is more representative of your aerobic endurance.

Here is how I configured my filters for running activities:

This way I make sure the analysis is not too affected by issues such as running on trails (by controlling elevation gain), short hard sessions (by filtering on time and heart rate) as well as particularly hot day or runs at altitude.

Let’s look at my aerobic efficiency for the same time period shown for the training load analysis:

We can now see much better how my performance dropped due to the injury, and how the first month of training my aerobic efficiency was very low (basically a high heart rate even for slow runs). As I keep training and doing lots of short intervals, my aerobic efficiency goes up (this is similar to what you could call an improvement in VO2max), long story short, I can run faster. My intervals times drop considerably at every session until mid-August:

In August I started the specific phase of my plan, and you can see how after a period of maintenance, aerobic efficiency starts to go down. This means I am getting slower due to the limited training I am able to perform and the fact that I have to target other adaptations (being able to run 42.2 km for example).

Note that the chronic training load plot shown at the beginning tells me that I am fitter, while I am actually running slower. This is why we need to use better tools that can capture different aspects of how our body is adapting to training.

## Let’s move to cardiac decoupling.

Cardiac decoupling relates to your cardiac drift. What’s your cardiac drift? Basically, your heart rate increasing as a result of your body getting fatigued, during the second part of a workout. This can happen for reasons such as dehydration or increased core body temperature, causing a reduction in blood volume as well as more blood to be directed to the periphery for cooling. This process causes lower stroke volume and an increase in heart rate that compensates for it so that cardiac output can be maintained.

To determine your cardiac decoupling, we compute the relation between output (pace or power) and input (heart rate) during the first and second half of a workout. Intuitively, if heart rate increases at the same pace during the second part of a workout, or if your pace reduces in an attempt to keep your heart rate below a certain value, it means that your aerobic endurance for the distance is not well developed.

As I started doing more long runs or long intervals with moderate intensity recoveries, my body got better and better at doing what I was training for: handling the distance.

Let’s look at the data:

Again we can see how post-injury my heart rate was not only very high at a given pace (as shown in the aerobic efficiency figure) but would also increase very rapidly during the second half of a short run (30 to 70 minutes, the same data is used for both plots according to the filters above).

As I get more weeks of long runs in, we can see how cardiac decoupling gets closer to one, meaning that the relationship between heart rate and pace in the first and second half of easy aerobic runs, is pretty much the same.

This analysis tells me that I am getting better at running for longer, but I am also getting slower and therefore I might need to adjust my race goal time for the marathon accordingly.

# Wrapping it up

In this post, I’ve covered some of the limitations of standard training load analysis and looked at other ways to analyze workouts data to determine training adaptations.

Looking at aerobic efficiency and cardiac decoupling can be more informative, as you can better understand which aspect is your limiter so that you know what to address based on your goal event.

Poor cardiac decoupling could mean that you need to work on your longer workouts while you can most likely improve your aerobic efficiency by doing speed work, as I have shown using my data.

As you can see, chronic training load is a bit of a simplistic way to monitor progress. It can still be useful (especially if we are healthy and fit and can train all the time there will be a strong relation between CTL and performance) but not looking at physiology to track how we are doing, makes little sense. Internal load (heart rate) is the only way to understand how hard we are working in terms of stress on the body, and therefore if we are improving, for a given workload (regardless of how you measure work: pace, speed, or power).

Finally, note also that you don’t need to obsess over pace, heart rate or power and overanalyze each individual workout. Physiology changes slowly. Fitness changes slowly. A single session’s aerobic efficiency or cardiac decoupling might be off because of tiredness, caffeine intake, or a myriad of other reasons. However, looking at medium and long term trends for contextualized physiological data, as shown above, is extremely effective and provides insights on which aspects are limiting your performance. Additionally, these tools do not require specific lab tests or racing, both being stressful events and disruptions in your training. This is exactly why we have built tools that rely on weeks of your regular workouts data and let you filter our abnormal days in HRV4Training Pro.

# Where to go from here?

In this post, we have seen how we can analyze workouts data to capture training progress. As we hopefully have established, more is not always better and we need to account for our physiological responses to training.

In this context, another important bit is to understand when we should be applying a given training stimulus so that we can better assimilate it and improve performance in the long term.

Marco holds a PhD cum laude in applied machine learning, a M.Sc. cum laude in computer science engineering and is currently enrolled in a M.Sc. in human movement sciences and high-performance coaching.

He has published more than 50 papers and patents at the intersection between physiology, health, technology and human performance.

He is the co-founder of HRV4Training and loves running.

Written by

## Marco Altini

#### PhD in Data Science, Founder of HRV4Training.com, Traveler, Passionate Runner, Immigrant.

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