What the Scale Really Shows, and How the Metabolic Forecast Reads It

The usual approach

Ask most calorie tools what a deficit will do and you get one rule for everyone: about 3,500 calories equals one pound. Eat 500 fewer calories a day and you should lose a pound a week, every week, in a straight line.

Researchers have tested that rule against real people, and it falls short. In a 2013 analysis, scientists compared its predictions with the results of seven weight loss studies where food was supervised or measured, and found that the rule badly overestimates the weight people actually lose ¹Thomas et al., 2013Commentary with dataThe rule tested against 7 supervised weight loss studies.

One reason is that the body does not stand still. As weight changes, the energy you burn changes with it, so the response to eating less is slow and curved. A model of adult metabolism published in The Lancet found that body weight responds to a change in intake with a half time of about a year ²Hall et al., 2011Modeling studyA model of adult human metabolism and weight change over time. That means it takes about a year to get halfway to your new steady weight, not a few weeks.

The rule also ignores timing. It does not know when you ate, how you slept, or how your body in particular responds. And it never checks its own answer.

What the scale really shows

Over a few days, much of what a scale shows is water, not fat or muscle. Two things move it most.

The first is stored carbohydrate. Your liver and muscles store carbs as glycogen, and glycogen is stored with three to four parts water ³Kreitzman et al., 1992Clinical study11 adults, 4 days on a very low calorie diet. Eat a carb-heavy meal and you refill that store, along with its water. Cut carbs and you empty it, and the water leaves too.

That is why the first week of keto often looks dramatic on the scale. In a classic metabolic ward study, six adults each spent 10 days on an 800-calorie ketogenic diet and 10 days on an 800-calorie mixed diet. On the ketogenic diet, 61 percent of the weight they lost was water and 35 percent was fat. On the mixed diet, water was 37 percent of the loss ⁴Yang et al., 1976Metabolic ward study6 adults, three 10-day diet periods. The extra weight that came off on the ketogenic diet was mostly water. A later study found a related pattern: weight lost in the first weeks of a diet carries less stored energy per pound than weight lost later, which is one more reason a fixed rule misreads the start ⁵Heymsfield et al., 2012Clinical study23 adults (CALERIE) plus 75 adults (Kiel), body composition by DXA.

The second is salt. Sodium changes how much water your body holds outside its cells. In a randomized crossover study of 51 healthy adults, a high-salt diet expanded that fluid compared with a low-salt diet ⁶Oppelaar et al., 2023Randomized crossover trial51 healthy adults, a low-salt and a high-salt diet. So salt can move the scale without any change in body fat.

Put those together and you get the weekly rhythm many people notice. In a European study of 1,421 adults who weighed themselves at home, weight rose over the weekend and fell during the week, a swing of about 0.35 percent ⁷Turicchi et al., 2020Observational analysis within a trial1,421 adults weighing themselves at home. For a 200-pound person, that is about 0.7 pounds up and back down in an ordinary week.

What the Metabolic Forecast does

The Metabolic Forecast is our answer to the usual approach. Instead of one formula for everyone, it is an engine that runs your own metabolism forward and shows your weight, your energy and, when you eat keto, your ketones for the next 72 hours.

• It starts with your history. Up to 60 days of meals and their timing, your ketone and glucose readings, and your weigh-ins. If you connect Apple Health or Health Connect, it adds sleep, steps and recovery. Those are read for the forecast and never stored.
• It runs forward one hour at a time. Every hour it weighs your last day of eating, your latest meal, the time of day, your fast, your sleep and your training. In the forecast, carbs pull ketones down fast, and the rebuild takes most of a day.
• It reads the scale differently. It separates real tissue from the water stored with carbs and the water held by salt. A carb-heavy Saturday does not read as fat gain, and the water leaving on Monday does not read as fat loss.

It learns you, then it grades itself

The engine begins with published human physiology, then shifts toward you as your data comes in: your real daily burn and your own ketone response. The longer you log, the more of the model is yours.

Then it does something a formula never does. Every forecast is saved with an honest range. When your next reading or weigh-in arrives, the engine checks what it predicted. Misses widen the range and adjust your model. An honest range holds about eight readings in ten, and the engine holds itself to that standard.

That is the difference between a guess and a forecast. A weather forecast earns trust because tomorrow checks it. This one is checked by your next reading.

It knows how you eat

Not everyone who tracks is in ketosis. On a Mediterranean or balanced plan, the forecast shows no ketone curve at all. Ketones appear only when your own data shows ketosis.

The order of the work matters too. Every number is calculated first. Only after the numbers are final does AI write the explanation, in English or Spanish, and the AI cannot change a number.

Why we built it this way

Four ideas sit under the whole design.

• Personal. Built from your history, not an average.
• Honest. It shows a range, not false precision.
• Accountable. It is checked against what actually happens.
• It grows with you. More logging means more of the model is fit to you.

In practice, that changes how you plan. Thinking about a higher-carb day? See the dip and the rebuild before you commit. Short on sleep? See what it does to tomorrow. And when the scale jumps after a salty dinner, the forecast does not count that water as fat.