A trailing seven-day mean on a steady slope shows your weight as it was three days ago (our arithmetic, not a published finding). TrendWeight's default trend lags about nine days by its own documentation. So your weekly weight average is a delayed number, and skipped days change it further.
Three things people call a weekly average
The same phrase covers three different calculations. Pick one on purpose.
The calendar-week mean. Add the readings you have from Monday through Sunday and divide by how many you have. This is what a spreadsheet's AVERAGE does with blank cells. It is a mean of available observations (our term), and it is a seven-day mean only when all seven days are present.
The trailing seven-day mean. For each day, average that day and the six before it. It moves daily and needs no fixed week.
The exponential trend. Each day's trend is the previous trend plus a fixed share of the gap between today's weight and that trend. The Hacker's Diet uses 10 percent: trend today = trend yesterday + 0.1 × (weight today − trend yesterday), and it assumes one measurement per day. NIST writes the general recurrence for control charts as EWMA(t) = λ × Y(t) + (1 − λ) × EWMA(t − 1); set λ to 0.1 and you have the Hacker's Diet update. TrendWeight, per its documentation read on October 20, 2026, sets alpha to 0.1 and fills missing days by linear interpolation. Our TrendWeight guide covers its method.
Why skipped weekends move the number
Weight has a weekly rhythm. In 80 adults who weighed daily (4,657 measurements), weight was highest on Sunday and Monday and fell through the week. In 1,421 people in a weight-loss maintenance program, Monday was highest and Friday lowest, and the Monday-to-Friday swing averaged around 0.35 percent of body weight, about 0.3 kg. Those are group averages. Yours can differ.
Now the arithmetic, which is ours. Suppose two high days (Sunday and Monday) average g above the other five. A full week's mean sits 2/7 of g above the mean of the five low days. With an illustrative gap of 0.6 lb, skipping both high days pulls the weekly figure down by 2/7 × 0.6, about 0.17 lb. Skip Thursday and Friday instead and it goes up.
The direction depends on which days you skip and on your own pattern. A week missing weekends can sit lower or higher. No study read here gives a correction for missing weekdays, so this post offers none. It offers a way to see the problem.
How much lag each one carries
A moving average reports the past. On a steady slope, a trailing seven-day mean averages values from today back to six days ago, and the middle of that window is three days back. So it reads the line as it was three days earlier. That is derived arithmetic, not a measurement.
Say you lose a steady 0.2 lb a day (illustrative). The seven-day mean runs about 0.6 lb above the line.
The exponential trend lags more. TrendWeight's documentation says its default lags about 9 days' worth of weight change. At alpha 0.1, that is six days slower than the seven-day mean. The trend is smoother, not faster. TrendWeight also offers Holt presets that track a slope as well as a level; we have not tested how they lag.
Lag applies to every average. When your weight changes direction, any average shows it days later.
Which to use when days are missing
If you weigh every day, use the trailing seven-day mean, or the calendar-week mean when all seven days are in. If you skip days, use the exponential trend with interpolated gaps. It has no denominator to change when a day drops out.
Our view: a column labeled "weekly average" that silently averages whatever exists is the wrong design. It changes its own denominator and hides that it did. If you keep one, print the reading count beside it and treat a count under seven as partial.
Tradeoffs: interpolation draws a straight line from Friday to Monday, so it cannot recover the Sunday rise you did not record. And the nine-day lag is a price. If you want a ready-made version, TrendWeight's exponential method is documented in full. If your readings sit in the Apple Health app, our guide to tracking weight trends with Apple Health is the place to start.
None of these averages measures fat. They average total scale weight, and that is all. If weighing or the number starts to drive your eating or mood, it is fine to stop and talk to a doctor.
A spreadsheet recipe with eight illustrative days
- Put dates in column A and weights in column B. Leave missing days blank. Never type 0.
- Calendar-week mean: =AVERAGE(B2:B8). Next to it put =COUNT(B2:B8). A count under 7 means a partial week.
- Trailing seven-day mean: on each row, =AVERAGE over that row and the six above it, again with a count.
- Exponential trend: fill each gap with a straight line, filled = previous known + (next known − previous known) × days since previous ÷ days between. Start the trend at your first reading. Then each row is =previous + 0.1 * (filled − previous).
Here are eight made-up days, in pounds. Thursday, Saturday and Sunday are missing.
| Day | Reading | Filled | Trend (0.1) |
|---|---|---|---|
| Mon | 180.4 | 180.4 | 180.40 |
| Tue | 179.8 | 179.8 | 180.34 |
| Wed | 179.6 | 179.6 | 180.27 |
| Thu | missing | 179.3 | 180.17 |
| Fri | 179.0 | 179.0 | 180.05 |
| Sat | missing | 179.33 | 179.98 |
| Sun | missing | 179.67 | 179.95 |
| Mon | 180.0 | 180.0 | 179.95 |
- Calendar-week mean, first Monday to Sunday: 179.70 from four readings.
- Trailing seven-day mean on the second Monday: 179.60, also from four readings.
- Exponential trend on the second Monday: 179.95.
Three methods give three answers from the same eight days. None is wrong. Each answers a different question, so pick one and stay with it.
Trends across months, not weeks
We make Trendline. Trendline is an iPhone app that reads Apple Health. You can follow weight and body-fat trends across one month, three months, six months, one year, or all time.
Questions people ask
How do I calculate a weekly weight average with missing days?
Adding up the readings you have and dividing by their count gives an available-observation mean, not a seven-day mean. If days are missing, fill the gaps by straight-line interpolation and run an exponential trend, or keep a reading count beside the average.
Why does my weekly weight average change when I skip weekends?
In a study of 80 adults who weighed daily, weight ran highest on Sunday and Monday. If those days drop out of the week, the mean is built from a different mix of weekdays, so it can shift without any change in you.
How many days behind is a seven-day average of weight?
On a steady slope, a trailing seven-day mean equals the line as it was three days earlier. That is our own arithmetic, not a published finding. Noisy or gappy data will not follow it exactly.
Does a weekly weight average tell me how much fat I lost?
No. An average of scale weight is still an average of total mass. It cannot say how much of a change was fat, water, glycogen or lean tissue, and no average can add that information back.