You gained 2 kg overnight because the scale weighs everything, and almost none of what changes overnight is fat. Two kilograms of fat would take about 15,400 kcal above your expenditure at the 7,700 kcal/kg planning value; even a very large day is a fraction of that. What moved the number is water — stored with carbohydrate, held with salt, shifted by training — and the plain mass of food and drink that has not left yet. This page puts numbers on each part, shows what the trend weight does with a spike, and gives the test for when a gain is real. It is arithmetic, not a trial result.

Water, glycogen, sodium and what you ate

Start with the impossible explanation and rule it out. Take the 75 kg example (TDEE 2,269 kcal) after a 3,500 kcal day:

  • surplus = 3,500 − 2,269 = 1,231 kcal
  • fat-equivalent = 1,231 ÷ 7,700 = 0.16 kg

Even a 5,000 kcal day is a 2,731 kcal surplus, or 0.35 kg. The scale said +2.0 kg. So at most a sixth of the jump could be tissue, and in practice less, because part of a day's surplus is used before it is stored. The remaining 1.8 kg or so has to be something that weighs a lot and carries few or no calories.

SourceDirectionWhat it isHow it clears
Glycogen and its waterUp after a carbohydrate-heavy dayCarbohydrate is stored in muscle and liver together with water; a big refill after a low-carb stretch weighs more than the food didOver the following days as the stores are used
Sodium and water balanceUp after a salty mealThe body holds extra water to keep its salt concentration steadyA day or two, as the salt is excreted
Food and drink in transitUp in proportion to volume, not caloriesA litre of anything weighs a kilogram; a large meal plus drinks is easily 1–2 kg of mass before it is digestedOver the next day or so
TrainingUp after a hard or unusual sessionWorked muscle holds extra water for a whileDays
Menstrual cycleUp for part of the cycleCyclical water retentionWith the cycle
Fat and lean tissueSlowly, either wayThe only part that reflects energy balanceIt does not clear — it accumulates at the rate the weekly arithmetic sets

Two features of the table matter. First, every row but the last is *mass without matching calories*, which is why a calorie count cannot predict tomorrow's scale reading. Second, every row but the last is reversible on a timescale of days. That is the whole basis for reading the trend instead of the scale: fat moves by tens of grams a day and never reverses on its own; water moves by kilograms and does.

We are deliberately not quoting grams-of-water-per-gram-of-glycogen ratios or how many hours a salty meal takes to clear; we do not have a verified source to hand for those figures, and the argument does not need them.

In plain terms: a kilogram of water and a kilogram of fat weigh the same, and the scale cannot tell them apart. The calendar can.

What the trend weight does with a spike

MacroDiet's trend weight is an exponential moving average — a running average that gives the newest weigh-in a small fixed share and the previous trend the rest. The half-life is 7 days, which sets that share at α = 1 − e^(−ln 2 ÷ 7) ≈ 0.094. Each morning:

  • trend today = 0.094 × scale today + 0.906 × trend yesterday

So a +2.0 kg spike on one morning moves the trend by 0.094 × 2.0 = 0.19 kg. What happens next depends on whether the spike persists.

Scale readingDay 1Day 2Day 3Day 7Day 14
One-morning spike (+2.0, then back to baseline) — trend excess+0.19 kg+0.17 kg+0.15 kg+0.10 kg+0.05 kg
Three-morning spike (+2.0 for three days, then baseline) — trend excess+0.19 kg+0.36 kg+0.52 kg+0.35 kg+0.17 kg
Spike that never leaves (+2.0 from now on) — trend excess+0.19 kg+0.36 kg+0.52 kg+1.00 kg+1.50 kg

Every cell is the recurrence above, applied day by day. Read the rows against each other. A single morning of +2 kg is a 0.19 kg nudge that halves every week — by day seven the trend is only a tenth of a kilogram above where it was. Three days at +2 is half a kilogram on the trend, fading afterwards. And a jump that stays is, by definition, tissue or a new steady state of water: the trend walks towards it, reaching half the distance in seven days and three quarters in fourteen.

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.

That third row is the test built into the tool. MacroDiet's own Weight Trend screen puts it plainly: "If day-to-day fluctuations stress you out, follow the trend." The same smoothing is why one high day barely changes an expenditure estimate that reads the trend over three weeks — see why weight loss stalls.

When it actually is fat

A gain is real when it behaves like the last row of the first table and the last row of the second: it does not clear, and the trend keeps rising towards it. Practical version:

  1. Day one or two: ignore the scale, weigh anyway, log the food honestly. The trend will move by a fraction of the jump.
  2. One week on: if the scale is back near its old range, it was water and mass; the trend excess is under 0.1 kg and shrinking. If the scale is still up but the trend has risen less than half the jump, it is too early to call.
  3. Two weeks on: if the scale is still up and the trend has followed it most of the way, some of it is real. Now the arithmetic explains it: two weeks of eating 500 kcal a day over expenditure is 7,000 kcal, or about 0.9 kg — which is what the trend will show.

Notice that the second case in step 3 needs a *sustained* surplus, not a day. One day cannot do it; the one bad day page has the week's arithmetic. And if the trend has been rising slowly for a month during what was meant to be a deficit, the cause is upstream — logging or the target — and why weight loss stalls is the page to read.

The short version

Two kilograms overnight is water, glycogen, salt and food in transit; the fat-equivalent of even a 3,500 kcal day is about 0.16 kg, because fat costs 7,700 kcal a kilogram. The 7-day-half-life trend turns a +2 kg morning into a 0.19 kg nudge that halves every week. If the scale is still up after one to two weeks and the trend has followed it, some of it is real, and two weeks of surplus is what explains it — not the one night.

Educational overview only — not medical advice. Zyra is a training and nutrition tracking tool; the rates, caps and floors on this page describe how the calculator behaves, not what any individual should do.