During the pandemic, and in-between knee surgeries, I started focusing on aerobic base training runs. The guidance I followed is from Steve House’s “Training for the New Alpinism.” One thing I noticed was that focusing only on aerobic base training significantly improved my time on his “box step test.” The test (described on pg. 179) consists of putting a 16 kg kettlebell into a backpack, wearing mountaineering boots, and doing 1,000 step ups on a one-foot box for time.

What surprised me is that when I was primarily doing CrossFit and “fast” runs, my box step times were in the 42-44 minute range. But after a period of time spent only climbing and doing aerobic base training, I hit a PR of 35:38.

This shouldn’t have shocked me since I had read “Extreme Alpinism” by Mark Twight in 2003 and “Lactate Threshold Training” by Peter Janssen in 2004, but since I’m a knucklehead I ignored the science and just figured that if a workout was harder then it had to be better. Interestingly, my jiu jitsu suffered from the same mentality. When I was a white and early blue belt, I always thought I should roll full speed since I would be competing at full speed. Why dial things back if you’re going to be going all out when it counts, right?

Wrong. My jiu jitsu improved the most from drilling and doing positional training at initially very low, but gradually increasing, intensities. Credit to Ian Lieberman for recommending this approach.

I’ve been using a Garmin heart rate monitor for years, but I’d never analyzed the data before. I had a sense that my base pace mile times had significantly increased by around a minute and a half to two minutes per mile over the last two years and wanted to look into that. My sense of this came from just scrolling through Connect and looking at average pace per mile for a few base runs. The two hypotheses that I thought were most likely were either that I had a serious underlying health condition or that doing base runs on an elliptical at a 120 average heart rate explained the difference.

What I learned from analyzing the data was that:

1) My base run pace was slower, on average, by 13 seconds per mile. This is nowhere near the 1-2 minute mark.

2) The 9:15 or 9:30 base runs I had remembered and used as my mental benchmark were very short in duration. These were 1-3 mile runs, typically around 2 miles, and that’s not at all comparable to a 9 or 11 mile base pace. Credit to Jim Barnum for asking a question that lead me to look at this.

3) When I was faster, I logged around 25 more hours running around 110 more miles over the previous few months compared to the slower period.

4) My memories, feelings, and eyeballing a few data points are not nearly as reliable as actually analyzing the data- to a degree that I’m frankly pretty uncomfortable with. I’m wondering where else in my life my perception is so different from reality…

You can see all of the Python code and several data visualizations here. Processing the Garmin data was a little trickier than I thought it would be. The broad approach to the analyses was to:

  1. Limit the dataset to include only running, hiking, treadmill running, mountaineering, and the box step test.
  2. Convert the total times for each run from hour:minute:second string format into a total number of seconds.
  3. Calculate the cumulative seconds of training time for each month then convert this back into total number of hours spent training for each month.
  4. Create monthly averages for base run heart rates, base run speeds, “fast” run speeds, and “fast” run heart rates. Base training is defined as an average heart rate under 146.
  5. Convert the monthly averages back into minutes-and-seconds-per-mile.
  6. Limit the data to September 2020 or later.
  7. Create a bar chart of training distances, a scatter plot of average paces, and then combine these into one data visualization.

The above shows the average monthly base run paces plotted against the monthly cumulative base run mileage. It also shows the trends of speed and training volume over time.