I charted four years of weigh-ins as one slope

2026-10-06 10:03

I don't trust before/after photos. I exported 1,461 daily weigh-ins from 2019–2024 and potted a 28-day rolling regression slope instead.

The result is less a dramatic arc and more a boring staircase: 11 flat months, two regains that both started during work travel, one 14-week decline, then a long plateau I kept calling a stall. The biggest slope shift didn't come from cutting carbs.

It came from adding a 20-minute walk after dinner and keeping my protein around 120g.

Raw weight still annoys me daily, but the slope tells me whether the month is actually moving. I keep it pinned above my desk. Anyone else tracking trendline slope, or am I just making my own mood ring with extra steps?

2026-10-09 04:25

If you’re fitting one slope to four years, you’re fitting a line to a staircase. Try a change-point model or segmented regression; it’ll give you dates and slopes per segment.

I tag travel days and compare slopes with/without them. Question: what’s the autocorrelation in your residuals?

Mine was ~0.6 on daily weights, so naive p-values were fiction.

A 28-day rolling slope with n=28 is basically one independent observation, not 28.

2026-10-11 01:58

Tbh the travel regains are the most interesting bit for me. I had a similar staircase and when I tagged trips, the positive slope showed up within 2–3 days and faded after a week of normal sodium/sleep—so it looked more like water/glycogen than fat.

Did you try the regression on a 3-day median instead of raw daily weights? In the trials they smooth, but this is messier.

Also curious: did the long plateau have a stable slope or just hover around zero? I keep wondering how much of my own “stall” was measurement noise vs actual maintenance.

2026-10-11 15:55

Running a cafe, this hits hard. My weight graph looks the same: flat, flat, then a little cliff after every holiday catering week.

What helped me was a boring rule — I pre-portioned a small cup of whatever I was taste-testing and logged it before service, instead of grazing from the spoon. Also weighed the same three lunch options once, so I stopped guessing when I was hungry and surrounded by pastry.

Sharp question: on your two work-travel regains, was it more the restaurant portions or the disrupted sleep/step count? I’m trying to figure out which lever actually moved my own staircase.