This chart really tells the whole story...
The chart shows a few things:
- My steps total on weekdays has generally increased over time.
- Each week, I generally see a few days over and under the trend line.
- On 6/4 I had my lowest step total since 4/2.
- I then proceeded to have five consecutive weekdays below the trend line.
This was all a bit bizarre to me, as I reviewed my data on 6/11 at mid-day. But then I started thinking about what had been going on lately:
- Slowly increasing my intensity and overall amount of exercise.
- Run at least four miles, every day, since some time in February* (see note below).
- I hadn't really taken a break in about four months.
So, I took 6/11 off. No running. And guess what? The next day I felt a lot better. I hypothesized that my activity level over time had reached a point of cumulative fatigue, where my body was just wore out and needed a day off. But could I have prevented that?
USING THE DATA
A few years ago, I read a book that talked about the impact of cumulative fatigue. Basically there is an advantage of running on tired legs over time, but too much running on tired legs ends with the body breaking down. Fairly intuitive stuff, but there is some disagreement in the literature ranging from people who see little risk in cumulative fatigue to those who see a lot of risk and believe that running less can be inherently better.
The expert opinions are fine and all, but what can I do to keep myself from having fatigue induced down weeks? I'm taking a multi-pronged approach:
- Forcing myself to take one day every two weeks off from running.
- Developing a metric that tracks my aggregate fatigue load in recent periods (past two weeks), then compares it to prior historical trends to spot possible issues.** (see note 2 below)
- Training a model based on the above derived metric, in order to predict days where I may be approaching a fatigue issue, and should take a day off.
I'll post more in the future on this, but generally, I'm taking the short term step of forced days off, but hope to have a model to predict "needed days off" in the near future.
*I may give some more background on my running "habit" at a later time. But the general of it is: I run 40-50 miles every week, but don't run races. I used to run races, but generally don't have time anymore. So I just run for exercise. It's good fun.
**The metric I'm playing with is aggregate steps over the period, "penalized" for variance. Variance is generally good for cumulative fatigue, because a single low-mileage day can *theoretically* allow healing from many high-mileage days.