how do I log my swimming and Tai Chi activity?

AnnPT77 wrote: »

sollyn23l2 wrote: »

"good brand/models tend to be pretty close for a majority of people"

Not exactly true. Even the "best" brands tend to be pretty dramatically off. Really, the fitness trackers are about as good as spitting out a random guess, statistically speaking.

The statistics I've seen suggest otherwise, as do quite a few posts I've seen here. I'm talking about all day 24x7 numbers over longer time periods here, not just snippets. I'd have to redo the reading I did, which I won't, but my pathetic memory suggests that 50-70%ish or so of decent brands are reasonably close on that basis, better than "calorie calculator" estimates. There isn't much relevant research, though. Many of the studies showing huge error IIRC looked at snippets, like individual exercise sessions or short time periods. I'm talking about averages over periods of time.

Even with a tracker, doing the 4-6 week (whole menstrual cycles) experiment is vital.

My tracker, as I've said before, is off by 25-30% for me, as compared with around 9 years of pretty careful logging experience. Here's an important thing, though: The level of error, averaged over time, stays in that range.

Some people here set up spreadsheets and compared their tracker's estimate over time to their actual intake/weight changes, and found that though the estimate was not accurate for them, the percentage error over time was reasonably consistent. They could use that information to recognize that if they ate that X% more or less than their tracker estimate, things worked out pretty well.

I never did that, because I had maintenance calories and maintenance habits dialed in pretty well before I got the tracker. (Well, it's actually been two trackers so far, both off by about the same percentage.) If I weren't lazy, I'd probably run that experiment, because I think that if I monitored the difference between my tracker estimate and my calorie/weight data over a couple of months, I could derive a useful adjustment percentage that I could apply going forward. I just don't care, don't bother, because what I'm doing is working.

The tracker gives an estimate, of course, not a measurement. (It's a more nuanced estimate than a calorie calculator, but it's still an estimate.) It can be treated in exactly the same way as a calorie-calculator starting estimate, i.e., run the experiment for several weeks, and use the results to adjust calories. The difference is that because it's more nuanced, it can be used as described above, because it will "see" activity changes over time, unlike the calculator estimates.

I don't know what your statistics background is, and it may be much better than mine. There's a thing in statistics called the "law of large numbers". It doesn't actually apply in this case, but I think something conceptually similar potentially does, maybe call it the "squishy law of medium numbers". Some of the aspects of calorie expenditure will probably be over-estimated, some under-estimated. The long run average of those estimates may converge nearer the actual value than the error involved in each of those individual parts, or at least near a value to which a consistent adjustment factor can be applied to get a useful number.

YMMV - clearly does.