Why did my weight go up overnight?

Almost certainly food volume in the gut, sodium and water, or glycogen. A large meal, a salty one, or a rest day after a depleting session can each move the scale two to three pounds within a day on a body whose composition has not changed at all.

How much daily weight fluctuation is normal?

Up to about four pounds across two days is normal for a trained adult. The change you are actually trying to detect is a quarter to half a pound a week, which means the noise is five to ten times larger than the signal.

Should I weigh myself every day?

Yes, and never act on a single reading. More weigh-ins make a smoothed trend both more responsive and more reliable — the daily weigh-in is data collection, and the trend is the thing you read.

Why is my weight not moving in a surplus?

Either the surplus is not real, or it is too small to see against two weeks of noise. For a hybrid athlete the first is far more likely: endurance volume rose, intake did not, and the surplus was erased by training the maintenance figure never knew about.

Does creatine make you gain weight?

It adds two to four pounds of intracellular water over the first weeks — inside the muscle, not under the skin. It is not fat, and it will distort any energy-balance calculation done across the window in which it happened.

What the Scale Is Actually Telling You

What actually moves it

Cause	Magnitude	Timescale
Food volume in the gut	1–3 lb	Hours to a day
Sodium and hydration	1–3 lb	Hours to days
Glycogen and its bound water	2–4 lb	Days
Creatine loading	2–4 lb	1–3 weeks, then stable
Hormonal cycling	1–4 lb	Cyclical
Post-session fluid loss	1–4 lb	Hours
Fat or muscle change	0.25–0.5 lb/week	Weeks

Read the last row against every row above it.

Glycogen deserves particular attention because it is the one that fools athletes specifically. Muscle glycogen binds roughly three grams of water per gram of glycogen. A depleting long run empties some of it and the scale drops; two high-carbohydrate days refill it and the scale climbs by more than the run "burned". Neither event involved any fat.

The filter

trend[i] = 0.25 * raw[i] + 0.75 * trend[i-1]

An exponentially weighted moving average. Each new weigh-in moves the trend a quarter of the way toward itself, so a single high reading nudges it and a sustained shift moves it properly. A real step change is about 90% reflected after eight readings — fast enough to catch a genuine change within a week and slow enough that Sunday's dinner does not register.

Weigh daily. Read the trend. The daily weigh-in is data collection; the trend is the information. Those are different activities and conflating them is the error.

The refeed trap

The single most likely moment for someone to abandon a correct plan.

Coming out of a sustained deficit and eating at maintenance, expect three to six pounds in two weeks, and none of it fat. Refilling muscle glycogen binds two to four pounds of water for a trained male; the digestive tract refilling adds another one to two. The waist ticks up for the same reason.

Every visible signal says the plan is failing. Nothing about the plan is failing. This is exactly what refeeding looks like, and knowing it in advance is the only defence.

The creatine trap

Two to four pounds of intracellular water over the first weeks — inside the muscle, not under the skin.

To an energy-balance calculation done across that window it reads as a surplus you did not eat:

3 lb over 14 days = 3 × 3500 ÷ 14 = 750 kcal/day of apparent surplus

Your measured maintenance comes out 750 calories too low and every decision downstream inherits it. Start creatine at the beginning of a block, not the middle of a measurement window.

The scale cannot do this alone

Even smoothed, trend weight cannot distinguish "the surplus is working" from "the surplus is too big". Both look like the trend rising.

A tape measure can:

Trend weight	Waist	Reading
Up	Flat	Working. Hold everything
Up	Up	Above the lean ceiling. Cut 100–150 kcal
Flat	Flat	Maintenance
Flat	Up	Recomposition going the wrong way
Down	Flat	Losing something that is not fat

Ninety seconds a fortnight, and it answers the question the scale structurally cannot. Why body fat percentage does not go on this list.

What the scale is uniquely good at

Having spent this page on its limitations, the thing it does that nothing else can:

It is an integrator. It sums every calorie in and every calorie out — including the run your watch under-estimated, the fidgeting you did not notice, your particular running economy, your particular metabolism.

That is the property that makes measured maintenance possible:

weightChangeLb = trendWeight(today) - trendWeight(14 days ago)
energyBalance  = weightChangeLb * 3500 / 14
maintenance    = avgDailyIntake - energyBalance

No MET table. No activity multiplier. No assumption about anything. The scale knows what the watch guessed at, and a trained lifter's entire lean-gain budget of 179 calories a day is resolvable over a fortnight by the trend and not by any arithmetic you could do. The full method.

Reading it on the right timescale

Window	What it can tell you
One day	Nothing. Water and gut content
One week	Very little. Inside normal variation
Two weeks	The minimum for a maintenance calculation
Three weeks, same direction	A signal
Six weeks	A block-level answer

Three consecutive weeks pointing the same way is a signal. One is weather. The specific trap is a two-week plateau read as a stalled metabolism — fourteen days is exactly one maintenance window and well inside normal variation.

What to do next

Weigh in every morning for two weeks under the same conditions, then plot the raw points against the trend. Seeing a two-pound spread around a line moving 0.3 lb a week is a more convincing argument for ignoring daily readings than anything written here.