Training (mostly) slow to race (kind of) fast

Physiological adaptations to 80/20 running: the data

Marco Altini
10 min readDec 20, 2016

Disclaimer: This post is a short case study on my personal data. I wasn’t blessed with much athletic talent, but I am really passionate about running and data science, which is what motivated me in the first place to build HRV4Training, a platform to measure and interpret physiological data. You will not read about any world record, but decent improvements backed up by solid data, and hopefully it will be useful to some.

My 10 km PR came a few years after starting with polarized training. Play the long game.

My blog has moved to Substack, you can find it here.

The research

Training slow to run fast(er) might seem counter-intuitive, however, quite some research in the past 15 years showed how elite runners (and not only runners) spend much of their time training at low intensities (see Stephen Seiler’s work). Following early research, interventions have been carried out to randomize runners in groups including a greater amount of low-intensity training, and groups including more moderate-intensity training. Results showed consistent improvements in running performance for groups training at lower intensities for most of the time, typically 80% of the overall training load. Many runners, coaches, and authors have been preaching low-intensity training for years, under slightly different guidelines, but all following the same core principles (see for example most of Phil Maffetone’s work — the 180 formula, MAF test, etc. — and more recently 80/20 running by Matt Fitzgerald, who also wrote this nice piece on the topic on RunnersWorld). We also ran an analysis of HRV4Training users generated data highlighting how faster runners tend to train at lower intensities, check out this blog post for the full story.

The classic mistake

The problem with the average recreational runner is that we tend to get stuck into too much moderate intensity training, having a hard time recovering, increasing training volume, performing well during the hard sessions and also being more prone to injuries.

I was exactly that kind of runner. After some initial progress mainly due to basic adaptations to running as a sport, I’ve spent the past five years stuck with zero progress in running performance (sometimes inevitably as life got in the way). Last summer I’ve started reading more on running and I eventually decided to give this different type of training a try. Surely I was running slow in training before as well, or at least I thought I did, but things changed a lot when I started with 80/20 running.

Below I’ll go through my data and provide a few tips on tracking adherence and progress as well as show physiological adaptations in both resting measurements (heart rate and HRV) and heart rate during workouts.

Trying it out

To get started with low intensity (or 80/20) running you really need a heart rate monitor. Our perception of what is low intensity in terms of pace is most likely too intense physiologically speaking (see first plot below). Once you have a heart rate monitor, defining what is low intensity for you can be done in many different ways, a simple one is Maffetone’s 180 formula, or you can do 70% of your heart rate reserve added to your resting heart rate (so your threshold would be 0.7*(HR max — HR at rest) + HR at rest, in my case 0.7*(194–50)+50 = 150 beats per minute). Different methods will give you similar results, the point being, that’s the low-intensity threshold for your runs.

Here is how it went for my first few runs since I switched, about 6 months ago (August 3rd, 2016):

What you can see above is pace over time. Light blue dots indicate workouts after I started with 80/20 running, the dot size is my average heart rate for the run, so you can see it gets smaller as I start 80/20 running, as I was trying to stay below the threshold. The circle indicates my first 80/20 training, a.k.a. the slowest training in my life. There are still bigger dots at a faster (smaller numerically) pace, as quality work (tempo, intervals) is still present.

The first run was so slow, that it ended up being my slowest training in 7 years since I started running.

It definitely took some patience and commitment at the beginning.

Tracking adherence

I made a couple of plots to track my adherence to this training program. As you can see below, there is much more low-intensity work in the second part of the year, which confirms that before I was basically always going too fast/intense (very few light blue dots on the left):

Same data as above (pace over time, dot size this time is the inverse, so bigger = lower heart rate) but for the entire year. The dot color this time is determined based on my average heart rate for a workout being either below or above the pre-defined threshold of 150 beats per minute.

Similarly, we can look at the heart rate distribution for my data before and after August, showing again a much lower intensity (lower heart rate) for my workouts after starting 80/20 running:

Running below threshold often turns out as running at threshold, as we can see from the second distribution peaking at 150 bpm, but that’s good enough for me :)

Even more revealing is the maximal heart rate plot. In this one I plotted my heart rate max for any given training, before and after starting 80/20 running:

Apart from potential outliers (HR max >200 bpm), in this plot, we can see that 1) I was getting close to my maximal HR basically in each training before I started training at low intensities — i.e. always too intense 2) after starting low intensity training the distribution gets bimodal, we have two main peaks, which quite clearly separate low-intensity training and quality work (tempo, intervals).

The plots above show you how you can keep track of your workouts and make sure you follow the 80/20 or mostly low-intensity training. The main question I had however was another one, as I knew I was sticking to the plan, was I making any progress?

Making any progress?

Low-intensity training, especially at the beginning, can get frustrating (and boring) as you force yourself to run very slowly. Be patient. Physiological adaptations such as increasing your aerobic fitness takes time in the order of months, not days, especially if you are already used to training/running.

Racing short distances regularly to track progress is an option, but also can be a big disruption in training and stress on the body. The end game is improving racing performance, however, there are other simple ways to track progress as we go.

In my opinion, a good approach is the MAF test, which consists in running at your threshold heart rate (same you picked for your low-intensity workouts) for 1 mile, check the time it takes, repeat every month or so. By keeping two parameters constant (your heart rate and the distance you run) you will be able to track progress as it will take less time to cover the same distance at the same intensity.

Alternatively, you can simply check your typical low-intensity run. As creatures of habit all runners tend to often run the same routes, so pick the one you run the most often, and check if your pace gets better over time, at the same threshold heart rate. Here are my pace and heart rate for my most typical training, a 10 km, 110 meters elevation gain “recovery loop”:

A couple of points from the plot above: pace decreased (higher values in the plot) dramatically when I started the 80/20 running. However, as aerobic fitness improves and your body adapts, the pace started increasing again (lower values at the right end side of the plot), highlighting improved fitness. There is a gap in the data as I was in Europe in September/October and I couldn’t run this route.

It’s important to use always the same route as you want most parameters to remain unchanged (elevation gain for example). Clearly, other factors will also have an impact over long periods of time, for example, temperature, yet this is the most practical way I found to track my progress. As heart rate on a daily basis can be affected also by many other parameters (training, stress, etc.), I would recommend using a route that you run weekly, so that you can have more data points and can spot trends instead of just noise in the data.

Alternatively, we have built analytics in HRV4Training Pro which makes it easier to track how your heart rate changes over time at a given intensity (what we call aerobic efficiency), as described in the post below.

Physiological adaptations

I’ve shown already above some physiological adaptations as a way to track progress, as the pace at the same intensity (heart rate threshold) will slowly get faster. These adaptations are mainly due to improvements in aerobic capacity, however, I’ve seen much faster changes in resting physiology.

Below are my heart rate and heart rate variability (HRV) over the past year, measured each morning using HRV4Training:

Heart rate (top) and heart rate variability (rMSSD, bottom) during this year. A reduction in heart rate is typically associated to improved cardiorespiratory fitness. Similarly, an increase in HRV can be associated to higher fitness as well as reduced physiological stress on the body.

The quick reduction in resting heart rate and increase in HRV put my mind at ease as I thought something positive (physiologically speaking) was happening rather fast.

I believe part of these changes might be due to the less overall intensity / load I was putting on my body, which might have been constantly overworked before.

Two other main adaptations that I believe should show up in most people were quite obvious. Training at low intensity allowed for much faster recovery, you go less hard not only on the cardiorespiratory system but also on the muscles, less DOMS (delayed onset muscle soreness), etc. which in turn allowed me to increase significantly training load, something I’ve always struggled with (50% more running in 6 months, from 30–35 km / week to 70+ km / week, eventually settling at around 90 km / week in the following years). Below you can see chronic and acute training load over time (also available in HRV4Training):

Chronic and acute training loads are computed on 42 and 7 days moving windows, according to the Banister model.

These data are particularly interesting especially when we look at them together with resting physiological data. Both heart rate and HRV show signs of reduced stress (high HRV and low HR) while training load was significantly increasing. This might be an effect of the more polarized training as the body is given more time to recover, due to all the low-intensity training, with respect to the previous “always moderate” regime.

The second adaptation, that influenced both resting physiological data and heart rate during exercise, is improved aerobic capacity / cardiorespiratory fitness. It took some time to see these changes coming, but they are clear now as in the past month I PR’d all distances from 1 km to the half marathon (eventually after about 1 year of polarized training I went from 1h42' to 1h24' in the half marathon, which is probably close to my limit at this point).

How do we track changes in cardiorespiratory fitness? I’ve done quite some work on this topic during my PhD, the key is to contextualize sub-maximal heart rate. Intuitively, a person that can keep heart rate lower (with respect to her/his maximal) during a certain effort (e.g. running at a certain pace), can potentially perform better than a person that is already maxing out during the same effort (e.g. at the same running pace). In HRV4Training we use the speed over heart rate ratio as a variable representative of changes in cardiorespiratory fitness. We basically contextualize HR at different speeds so that you can track how your body adapts over time, without the need of performing a specific exercise or protocol. Here is how the data looks for my year:

Speed over heart rate for each workout. This parameter is representative of cardiorespiratory fitness and performance.

We can see above that I first plateau, then switch training to 80/20 and slowly build up to being able to sustain a higher speed a the same heart rate, which eventually resulted in my improvements. This plot provides very similar information to the one shown above where we look at changes in pace for the same route when keeping heart rate at threshold (constant), with the advantage that here we can put together all workouts, as heart rate is contextualized by pace, similarly to what is covered in the blog post about aerobic efficiency.

Closing remarks

Personally, I found low-intensity training extremely beneficial in improving performance. I could see physiological changes in resting heart rate, HRV, sub-maximal HR (and therefore estimated VO2max) and improve performance more in these past 6 months than in the previous 7 years of running. These were not the only changes I implemented, for example in the meantime I’ve also adjusted diet and lost some weight. As some of these changes were simultaneous, determining causality is challenging. However, I believe the biggest changes (increased training volume and cardiorespiratory fitness) are mainly due to this more polarized training, and to finally getting out of the “always moderate” training typical of recreational runners.

We have also brought this analysis to HRV4Training, so that you can track both adherence and progress similarly to what is shown above, within the app and web platform.

James Witts’ piece on training polarization featuring my case study. Runners World UK, June 2019.

How can you track training polarization?

Training polarization can now be tracked in HRV4Training, see some screenshots below.

Above you can see how my data looked before starting a more polarized training program, when I was stuck in moderate-intensity training (always training at way to high heart rates, only 23% of my workouts were easy at that stage), and how it looks now that I mainly train at low intensities (78% of my workouts in the past month are now easy), the last screenshot shows improvements in estimated VO2max following 6 months of polarized training (as VO2max is estimated using as one of the main parameters sub-maximal heart rate normalized by pace, changes in aerobic capacity resulting in the ability to run at low intensities faster, will reflect in improvements in VO2max)

Marco holds a PhD cum laude in applied machine learning, a M.Sc. cum laude in computer science engineering, and a M.Sc. cum laude 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.

Twitter: @altini_marco



Marco Altini

Founder, Data Science @ouraring Lecturer @VUamsterdam. PhD in Machine Learning, 2x MSc: Sport Science, Computer Science Engineering. Runner