§ · Forecast
Calculatedon your phone.
Five states. One math model. Fourteen-day lookback. Seven-day rolling average. The whole forecast runs in your iPhone's binary — turn airplane mode on and the projection still updates on the next weight log.
§ 01 · The 5 forecast states
The forecast is a 5-state model. Each transition is triggered by your actual data, not by a manual switch.
Locked
Default state when you have a set goal weight and you haven't logged enough data yet to confidently project a goal date.
Unlocked
Enough recent data accumulated. The forecast surfaces a projected goal date and current trend rate.
Goal-reached
Your 7-day rolling average has met or crossed your goal weight. The deck flips into a maintain-and-confirm pattern.
Maintaining
No active loss or gain target — the forecast tracks deviation from your maintenance band so you can spot drift early.
Reversed
Your 7-day rolling average is moving the wrong direction relative to your stated goal. The forecast flags it explicitly so you don't have to read the chart.
§ 02 · The math, in plain terms
BiteDeck uses Katch–McArdle for BMR — body-fat-aware, more accurate than Mifflin–St Jeor at the tails of body composition. Multiplied by your activity factor, that gives maintenance TDEE. Your logged intake against that number is your daily deficit (or surplus).
You can run the same calculations yourself: free TDEE calculator · Katch–McArdle BMR calculator.
§ 03 · The 14-day lookback and 7-day rolling average
Daily weight readings are noisy — water, glycogen, sleep, salt, hormonal cycles all move the number a couple of pounds in either direction without the underlying trend changing.
The forecast uses a 14-day window for the trend regression and a 7-day rolling average for the displayed weight. Two weeks is long enough to absorb the noise. Seven days is short enough to respond when the trend actually shifts.
Why not 30 days? Real change shows up in 14. A 30-day window would lag the moment your forecast started moving — exactly the wrong tradeoff when you're trying to confirm a plan is working or course-correct one that isn't.
§ 04 · Reversed-trend detection
The reversed state is the lever. If your 7-day rolling average is moving the wrong way relative to your goal — a gain trend during a cut, a loss trend during a bulk — the forecast flips into reversed and surfaces it on the Home screen. You don't have to read the chart to notice; the app tells you.
This is the part most calorie apps don't do. They show you a chart, you eyeball it, and the trend goes unflagged for another two weeks while you wonder why your plan isn't working.
§ 05 · Why on-device matters
Most weight-tracking apps do this same math on a server. You send up your weights, your intake, your activity; the server fits a model and returns a projection.
BiteDeck does it on your phone. The Katch–McArdle constant, the 14-day regression, the 7-day rolling average, the state machine — all in the binary. Airplane mode doesn't break it. There's no upload to opt out of.
For the broader argument: what local-first means.
§ 06 · The free TDEE / Katch–McArdle calculators
Both calculators run the same math BiteDeck runs daily, but one-shot: enter your numbers, get a result. Use them to sanity-check the forecast or to test a different body-fat estimate.