Training Bands and Training Volume

Posted

Speed has always been my personal struggle. I never think I’m fast enough and I always fall for the same mistake that first-timers make: I do too much, too quickly. I still don’t believe I can ever run a sub-6:00min/mi (<3:43min/km) and hold it for longer than 150 meters, but that hasn't stopped me from trying to become better at getting there.

To improve, we learn to push our limits until we pass certain thresholds, just enough to stimulate adaption, but not so much that we cause injury. Consequently, we rely on various forms of measurements to properly manage training load such as training volume, Training Impulse (TRIMPs), Ratings of Perceived Exertion (RPEs), Training Stress Score (TSS), and Training Effect (TE), among others. All of them, in one way or another, attempt to answer the same question: are you training hard enough so that you are improving?

As with all models, they each have their shortcomings, one of which is how averages are frequently used. This is an important point because it is the main difference between every “training score” calculation compared to Smashrun’s Training Bands.

Unlike TRIMPs and all of its variations, Training Bands do not attempt to rate your training effort. Instead, it is a detailed account of your effort distribution. To understand Training Bands, it is best to look at how we approach training sessions.

Tracking Effort Distribution with Training Bands

For the sake of simplicity, imagine that you are planning out your running schedule for the week and you decide to allocate 70% of your training volume to your aerobic zone, 20% to anaerobic, and 10% to your max. The easiest way to track this is to look at individual runs as a session and categorize them in some way such as easy, tempo, and speedwork.

A more granular approach would be to track your effort distribution for one run, which is what most training platforms do. Again, to keep it simple, we will only look at three zones: easy, moderate, and difficult.

Effort distribution for one run.

This is often useful for looking at one run but, when you are training for an event, looking at an isolated instance is less than ideal. One 10 mile run could be broken into multiple segments within the same session to include a warm up, several pick-ups, a short tempo, and a cool down.

How would you quantify the contribution of each of these segments if you’re only looking at each run as a whole? This is where Training Bands do the hard work.

Training Volume by Effort

Smashrun looks at each trackpoint of every run and allocates it to a bucket, which we call a training band. It does this for pace, heart rate, and hill grade. Using our previous example, if we took all of the easy zones from a bunch of different runs and added them up across all runs, we will end up with one training band.

One training band.

This becomes tremendously complex with actual running data, because one run could have thousands of trackpoints. Your entire training history could be hundreds of thousands of trackpoints. Each of those trackpoints go into a specific band. When you combine all the bands together, you get Smashrun’s Training Bands.

Another way of visualizing it is to look at individual area graphs, which you can actually do when you click on specific bands. Each color represents one band. When stacked together, they represent your total training volume.

Area graphs

Training Bands illustrate volume at a different level, because you are not just looking at total duration of all runs. Instead, you are looking at the total duration of runs at different training efforts. Whether you are viewing your Training Bands for pace, heart rate, or hill grade, each band represents how hard you performed, for how long, and for what percentage of your total training.

Combined with training volume based on heart rate zones, it is possible to use Training Bands to get a sense of your training load. This, of course, is not a direct calculation of your load. Instead, it is a quick way to visualize your effort distribution and training intensity over different training cycles.

Scrolling training bands.

As I mentioned in last week’s post about using Pace Trends, overload (the tipping point at which you start to improve) happens with planned recovery periods.

Use Training Bands to see when you spent more time running hard so you can better manage your future training volume.

% tempo

% easy/recovery

Identify the time periods when you did the most speedwork. Did it help your overall training? Or did it set you back because you were too burned out afterwards?

Intervals

Have you ever wondered what someone’s training might look like if they stick to the 10% rule to gradually increase volume? It’s like something out of a textbook.

10% rule

It gets even more interesting when we look at the same person’s Training Bands for heart rate, viewed as a percentage instead of absolute duration. You’ll see that they ran less in the beginning but spent a lot of time near their max, whereas they’re running much more now but spend much less time in the red zone. By running more, aerobic capacity just naturally increased, which is as it should be!

HR distribution.

Training Bands is really where Smashrun bridges the gap between small data (your individual runs) and big data (your total training duration by effort distribution). It is a much more meaningful representation of your training volume because it shows you how every second of every run contributes to your training as a whole.

  1. Laurent

    While this is an impressive display of data visualization, I am not sure how much insight I can derive from it.

    If all of my runs were always on the same course, in the same exact conditions (hydration status, terrain type, elevation, external temperature, freshness level, etc.), then training bands would help me decide how much I am improving under any particular stimulus.
    But, my runs are not like that.

    Since I moved from California to Colorado, I went up a mile in elevation. I also went from running on all flat terrain to very hilly roads and trails. I also had to adapt to much cooler temperature and much higher winds.
    So, look at my track points over the past few months and you would conclude I am putting more and more effort into my runs, for slower and slower results.
    Or you could just as well conclude that I am not training as hard, or putting as much care into my running.

    That’s one point.

    Secondly, if the time I spend below 70% of my max HR appears to increase, is it because I purposefully focus on aerobic training or is it because I can handle the same loads with less cardiac output? (i.e., getting fitter)

    In that sense, it would be better to always show input (effort) and output (normalized pace) in parallel, so that things can be correlated to some degree.

    • Jacklyn

      Of course, context is necessary. It’s one of the reasons why we track Training Bands for pace zones, heart rate zones, and hill difficulty. It’s entirely possible to draw correlations between the three and, just because, it’s not accounting for certain external factors (like a recent move) doesn’t mean that it can’t be insightful.

      Looking at factors like hydration status and terrain type are much harder to quantify and use as an adjustment factor. For something like freshness level, it’s entirely subjective and, often, once you start combining qualitative measurements with quantitative measurements, you’re giving yourself a chance to occasionally game the system.

      It’s also important to keep in mind that we only have so much bandwidth for information. Every decision to emphasize one thing is at the expense of something else. Our goal is not to overwhelm runners with too much data, but make it easier to manage the information they collect.

      That’s why Training Bands is so simple. It just tells you exactly what you’ve been doing. It’s a quick and easy way to visualize the distribution of your training volume based on pace buckets, heart rate zones, and hill grade. It just reports on your actual raw data, unadjusted and without averaging anything.

      We would like to eventually introduce some form of normalized graded adjusted pace but that will be for the Pace Trends and creating that data set will help us plot the input and output side by side.

      • olivier gounot

        Numbers not working for trailers? Seems I read that somewhere lately… ;)

        “We would like to eventually introduce some form of normalized graded adjusted pace but that will be for the Pace Trends and creating that data set will help us plot the input and output side by side.” => Sounds good! Can’t wait! What else do you have in your sleeves? :)

        • Jacklyn

          Ideas are plentiful! So much that our brain is running out of room to accommodate all of them. Execution is a different story…. (sigh) the cool stuff we could do with just one other developer. I’m working on putting up a forum. We’ll create a home for everyone’s ideas there. Hopefully, without shooting ourselves in the foot.

      • Laurent

        I am not saying that what you did sucks, far from that. All I am saying is that there is a point when data *aggregation” loses value. And that point comes when the data you aggregate is too disparate because it was gathered in very different circumstances.

        Granted, we both agree that you cannot quantify or control the conditions of any particular run. Hence, you cannot account for what would put the data into context.
        Yet, it is possible to fix/isolate/select some of the variables (course, time of day, pace, HR zone, cadence or combinations thereof) and compare such benchmark runs over time.

        As for displaying input and output simultaneously, you are so close…

        https://dl.dropboxusercontent.com/u/29886031/tmp/modal_views.png

        If there was a way to stack these 3 views into a single view, I would not have to click back and forth to see how the distribution of my efforts influenced the distribution of my paces.

        Peace out. Sorry if my prior comments appeared to be harsh.

  2. Lars

    These pace bands are determined automatically, right? So would there be a way to easily relate them for example to the pace categories as defined by Jack Daniels, e.g. Easy, Interval, Marathon etc.?

    • Adam Styles

      Running in never easy after a few Jack Daniels.

    • Jacklyn

      Yup – pace bands are fixed. Although, the current paradigm for selecting specific bands limits you to the selection of one training band, one training band and everything above it, or one training band and everything below it.

      If you wanted to select a custom combination of training bands that falls in the middle, you wouldn’t be able to do it.

      What we’ll have to do is give users the ability to select any combination of training bands to see the volume allocated to custom pace categories, such as those designed by Jack Daniels. It would be equally awesome to be able to do the same thing within a specific run using the Pro map, so you could also see your Avg. HR, cadence, distance, and duration for a pace category.

  3. Wilfried

    I like the training bands but as I have more than 3 years of runs I always have to select a part of my timeline. Each time I select a new band I have to make the selection again. It would be great if the selected timeframe in the overview section could be the basis to analyze.
    Wilfried.

    • Jacklyn

      Do you mean switching between pace, HR, and hill grade or just one training band within a given filter? Currently, I can highlight a section within my timeline, brush over different time periods within a given training band, switch to a new training band, and continue with the same time period.

      Although, when I switch to HR or hill grade from pace, I definitely lose the selected time frame. That’s a bug that Chris is fixing (woohoo!)

      • Wilfried

        I ment switching between pace, HR and hill grade. Good to hear that it will be fixed.

  4. Laurent

    Another thing I forgot to mention before is about HR zones. I have not found the place where I can specify my HR zones. I assume they are automatically calculated using the “traditional” percentage ranges. (70-79% = aerobic, 80-89% = anaerobic, etc.)
    It would be nice if this could be made a profile setting. I base the distribution of my zones on my estimated lactate threshold, so the cookie-cutter zones may not correspond very well to that.

    • Jacklyn

      Yup. It’s another feature that we’ve accounted for in our development schedule. (I’m pretty excited about custom HR zones as well!) We’ve had a couple of users request it, actually. Some want only three custom buckets, others want multiple buckets. I guess we gotta figure out how we’ll have to handle it. I need to look at how various training platforms display it and get a feel for best practices.

      • Laurent

        Good to hear. The traditional ranges can be as inaccurate as the various formulas for maximum HR. (the best max HR formula can easily be off by 10 beats)
        For HR zones, the lactate threshold HR can be anywhere between 80% and 95% of your maximum HR, depending on your age and level of fitness. So, any ranges based on that could be off by a lot as well.

        Finally, there is a pet peeve of mine with HR zones…HR zones are an example of what statisticians call data binning or data discretization.
        This being said, I consider HR to be a continuous variable, in the sense that it’s hard to imagine your HR jumping from 140 to 170 from 1 second to the next.
        Binning or discretization of continuous variables is a really bad idea. Your discard information in the process and it is worse than bad rounding.
        Email me for more details on what I mean.