Trend weight is a smoothed line through your scale readings; scale weight is the single number the scale showed this morning. The scale is measuring your body plus whatever water and food happen to be in it today, so it moves by a kilo or more from one morning to the next even when nothing about you has changed. The trend line averages those swings away and shows the direction underneath. This article gives the exact formula Zyra uses, the numbers it produces, and how to read the chart.

illustrative synthetic data76 kg77 kg78 kg79 kg80 kg81 kg82 kgday 1day 8day 15day 22day 29day 36scale readingtrend weight (EMA, 7-day half-life)start 81.1 → trend 78.7 kg
Fig. Daily scale readings (dots) against the trend weight (line) — an exponential moving average with a 7-day half-life, α = 1 − exp(−ln 2 / 7) ≈ 0.0943, the same smoothing Zyra Workouts applies. Each new reading moves the trend only about 9% of the way toward it, so a single heavy or light morning barely registers while a real drift shows within a week or two. Illustrative synthetic data, not a person's log.

What the scale measures

A bathroom scale measures total mass at that instant. Total mass includes things that change hour to hour: the water your body is holding, the glycogen (stored carbohydrate) in your muscles and liver and the water bound to it, the food and drink still in your gut, and what you have or have not yet excreted. None of those is stored energy in the sense that matters for a nutrition goal, and all of them shift with yesterday's salt, carbohydrate, training and hydration.

So a single reading is precise about the wrong thing. It tells you exactly how much you weighed at 07:10, and very little about whether your body changed this week. Two readings a week apart can differ by a kilogram in either direction with no change in fat or muscle at all.

In plain terms: the scale is a good instrument attached to a noisy quantity.

What a trend line does with it

A trend line treats each reading as one noisy sample of an underlying value and estimates that value from many samples. The simplest version is a plain average of the last N days. The version almost every tracking app uses — including MacroDiet and Zyra Workouts — is an exponential moving average, or EMA: a running average in which every reading counts, but recent readings count more and old readings fade out geometrically.

The reason for preferring the EMA over a fixed window is that it has no edge. A 7-day plain average jumps when a high reading enters the window and jumps again when it leaves a week later. An EMA lets the reading in gradually and lets it go gradually, so the line stays smooth.

MacroDiet's own in-app note puts the idea in one sentence: scale weight data tends to be quite noisy, and the weight trend is the signal in that noise.

The exact smoothing we use

The constant is set by a half-life of 7 days, and the code is short enough to quote:

HALF_LIFE_DAYS = 7 α = 1 − exp(−ln 2 / 7) ≈ 0.0943 trend[today] = α × scale[today] + (1 − α) × trend[yesterday]

The first reading seeds the line (trend[0] = scale[0]); after that each new reading pulls the trend about 9.4% of the way toward itself. A half-life of 7 days means exactly what it says: after 7 days, half of a reading's influence remains; after 14, a quarter; after 21, an eighth.

Two worked cases show what that does. Suppose the trend sits at 80.0 kg and this morning's reading is 82.0 kg — a typical post-weekend spike:

trend = 0.0943 × 82.0 + 0.9057 × 80.0 = 7.73 + 72.46 = 80.19 kg

The 2 kg spike moved the trend by 0.19 kg. If tomorrow's reading is back at 80.0 the trend drifts back toward 80.0 and the spike is essentially gone within a week.

Now suppose the 82.0 kg reading is real and every following day also reads 82.0. The trend closes the gap by half every 7 days:

Days since the changeShare of the change absorbedTrend value (80 → 82 kg)
09.4% (one reading)80.19 kg
750%81.0 kg
1475%81.5 kg
2187.5%81.75 kg
2893.75%81.88 kg

This is arithmetic, not a trial result. The same constant runs in the Zyra Workouts Insights tile (labelled "Smoothed Trend") and in the figure above, which is drawn from synthetic data.

Reading a trend chart

Three habits make the chart useful.

Read the slope, not the gap. The question the trend answers is "which way, and how fast, over the last two to three weeks?" Compare the trend value today with the trend value 14 or 21 days ago. A difference of −0.8 kg over three weeks is a real signal; a scale reading 0.8 kg above yesterday's is not.

Expect the lag, and expect it to point the right way. Because the trend is mostly made of earlier days, it sits above today's scale reading while you are losing and below it while you are gaining. That is not an error; it is the smoothing doing its job. If you switch from a deficit to maintenance, expect the scale to rise a little toward the trend as glycogen and gut contents return to normal.

Give real changes three weeks. The table above says a genuine change is 87.5% visible after 21 days. That is also why calorie-adjustment logic keyed to the trend waits rather than reacting to one week — see how apps estimate energy expenditure and why has my weight loss stalled?.

The short version

Scale weight is one noisy reading; trend weight is an exponentially weighted average of all of them with a 7-day half-life (α ≈ 0.0943). One odd morning moves the trend by about a tenth of its size; a change that persists is half absorbed in a week and nearly fully absorbed in three. Read the slope over two to three weeks and let the daily number be what it is — data for the line, not a verdict on the day. Training and nutrition decisions belong on the trend. Educational overview only — not medical advice.