Daily weight and trend weight are two separate numbers
Imagine five mornings on the scale: 201.4, 199.8, 202.1, 200.6, 200.2. The second and third mornings sit 2.3 lb apart, a swing that water and food can produce in a single day (the guide to overnight jumps covers why). Each reading combines a slow quantity (the weight of your tissue) with a fast one (that morning’s water and food). The trend tracks the slow quantity.
Researchers handle weight records the same way. A 2020 study of 1,421 adults removed the long-run trend from each person’s daily weights before measuring the weekly rhythm left over, and found weekend gains and weekday losses on the order of 0.35%. The trend the researchers set aside first is the part that tracks where weight goes over months.
How smoothing pulls a trend out of noisy readings
Smoothing treats each weigh-in as evidence and weighs it against what came before. One reading that lands far above the line moves the estimate a little. When the next several readings land up there too, the evidence agrees and the trend follows.
How far a single reading can move the trend depends on two things: how noisy readings usually are, and how fast real weight plausibly changes. A method that assumes very little noise chases every jump. One that assumes too much noise lags behind a real change for weeks.
Some methods also look backward. Once tomorrow’s reading arrives, it says something about today, so a smoother can revise the recent past with later data. That’s why the last stretch of a trend can shift after you log a new weigh-in, while older history stays put.
Averaging the last seven days in a spreadsheet is the familiar alternative. It lags behind real changes by about half its window, and it has no good answer for a week with only two readings in it.
Why a trend needs a range
A trend is an estimate, and estimates come with uncertainty. When readings are frequent and agree with each other, the trend’s range is narrow. When they’re sparse or scattered, it widens, because many different lines would fit the same few points.
The sample chart below shows this with invented data. Sixteen days in the middle have no weigh-ins, and the shaded range around the violet trend opens up across that stretch, then tightens again once readings resume.
- Weigh-in
- Trend
- Range
- Plan
- Goal
- Forecast
- No weigh-ins
A trend drawn as a bare line hides that difference. A line through a month of daily readings and a line across three weeks of silence look identical, even though one rests on far more evidence.
Reading the rate per week
The rate is the slope of the trend, usually shown in lb/wk. A rate of −0.3 lb/wk ±0.4 says the data fit a weekly change anywhere from −0.7 to +0.1 lb. Because that range includes zero, the current record can’t distinguish slow loss from holding steady. A few more weeks of weigh-ins usually settle it.
The rate also gives context for the pace of change. The CDC notes that people who lose weight at a gradual, steady pace of about 1 to 2 lb a week are more likely to keep it off. That figure is a population finding, and the right pace for your own plan is something to settle with your clinician.
Four common misreadings of a trend line
- Treating a one-day dip as a turn. A single low reading after a weekend moves the trend slightly. A change in direction shows up as several readings, over a week or more, on the same side of the line.
- Expecting the trend to match the scale. On most days the reading and the trend differ, sometimes by more than a pound. The difference is mostly fluid and digestion.
- Treating the newest stretch as settled. The latest part of a smoothed trend has the least data after it and will move the most as readings arrive.
- Ignoring the weekly rhythm. Research on weekly cycles puts the heaviest readings early in the week. Comparing a Monday reading with a Friday one mixes that rhythm into the change.
Where the trend lives in Balance Point
On the Trends chart, Balance Point draws the trend as a solid violet line with its range shaded around it and each weigh-in as a dot. The trend comes from a filter designed for weigh-ins that arrive on an uneven schedule, and the app then smooths history using later readings. Today shows the current trend value, your last recorded weigh-in, and the weekly rate with its range, for example “−0.3 lb/wk ±0.4”. A separate rate chart plots the weekly rate over time with its own uncertainty. The explainer on the trend and forecast goes further into the filter, and the forecast carries the trend forward to a window of likely arrival dates.