OnWeight — Source of Truth: Logic, Formulas & Flows

Status legend ✅ Documented — taken directly from the Figma frames / spec table. 🟡 Assumed — a reasonable rule I've proposed to make the prototype compute; needs your confirmation. 🔴 Open — not yet specified anywhere; decision required.

Last updated from: Key user flows, Goal Setting Flow, WeeklyLossRate, CutStrategy, MenstrualCycleWeightChanges, Trending Weight & Phases, the per-sport GS_14A_* frames, and the sport rules table.

1. Purpose

OnWeight helps a combat athlete cut to their fight-day weight safely and on schedule. The app's job is to interpret daily weigh-ins — telling the athlete in plain language whether they are on track, then exposing the underlying numbers for those who want them.

The core intelligence is three layers:

Trend — smooth noisy daily weigh-ins into a true trajectory.
Projection — extrapolate that trajectory to weigh-in day.
Verdict — classify pace (safe vs too fast vs behind vs off-track) and coach accordingly.

2. Inputs & data model

2.1 Athlete profile (set in onboarding)

Field	Source	Notes
`biologicalSex`	✅ `SignUp_BiologicalSex`	Gates menstrual-cycle flow + some strategy warnings. Also selects the division ladder (`divisions.men` / `divisions.women`) via `recommendDivision(sport, wt, sex)` — map `female → 'women'`, else `'men'`.
`currentWeight` (kg)	✅ `SignUp_CurrentWeightInput`	Starting / walk-around weight.
`goalWeight` (kg)	✅ `SignUp_GoalWeightInput`	The athlete's target fight-day body weight. User-facing term is now "your limit" — the weight you must hit on the scale (the onboarding screen reads "Your limit", not "Goal weight"). The field / code name stays `goalWeight`; only the athlete-facing copy changed. See the Lexicon note in §18.
`sport`	✅ `SignUp_SelectSport`	Boxing / Muay Thai / BJJ (gi/no-gi) / Taekwondo / Judo / MMA.
`hasFight` + `fightDate`	✅ `SignUp_Date_Fight` / `SignUp_NoFight`	No-fight users get an open-ended plan.
`weighInPreference` (time of day)	✅ `GS_02_WeighInTimePreference`	When the daily fasted weigh-in reminder fires.
`morningNotif`	✅ `GS_03_MorningNotifOptIn`	Opt-in to the reminder.
`cutRate` (% bw/week)	✅ `WeeklyLossRate`	See §5.
`cutStrategies[]`	✅ `CutStrategy`	Fight-week tactics. See §8.
Cycle settings	✅ `GS_09A`–`GS_13A`	See §7.

2.2 Competition rules (per sport — auto-filled, user-confirmable)

Source: ✅ sport rules table + GS_14A_CheckSportDetails_*, GS_15A–GS_21A.

Sport	Equipment req.	Equipment weight	Re-weigh-in	Hydration clause	Hydration cap	Weigh-in timing
Boxing	no	n/a	no	no	n/a	Same day (amateur), day before (pro)
Muay Thai	no	n/a	no	no	n/a	Same day (amateur), day before (pro)
BJJ (no gi)	yes	0.5 kg (rashguard + pants)	no	no	n/a	Within 2 hours
BJJ (gi)	yes	1.2 kg	no	no	n/a	Within 2 hours
Taekwondo	yes	1.2 kg	yes (5%)	no	n/a	Day before (with re-weigh-in 5%)
Judo	yes	1.2 kg	yes (5%)	no	n/a	Day before (with re-weigh-in 5%)
MMA	no	n/a	no	no	n/a	Day before

Note (✅): any non-equipment sport weighs in in underwear only. Hydration-clause fields exist in the model (GS_20A/GS_21A) for promotions that test hydration, even though none of the default sports above enable it.

2.3 Daily log

entries[]: { date, weight } — one fasted morning weigh-in per day (✅ AW_01–AW_04).
Editable date via the date picker; one canonical entry per day.

3. Target-weight calculation

The number the athlete must hit on the scale is not always their goal body weight — equipment and weigh-in timing change it.

`` scaleTarget = divisionLimit // the official class limit bodyTarget = scaleTarget − equipmentWeight // what the BODY must weigh (equipmentWeight = 0 for underwear-only sports) ``

✅ Equipment-required sports (BJJ, Taekwondo, Judo) add the gi/gear allowance back: you may weigh more on the scale because you're dressed, so the body must come in lower by equipmentWeight.
✅ Underwear-only sports (Boxing, Muay Thai, MMA): bodyTarget ≈ scaleTarget.
🟡 In the prototype this is illustrated as goalWeight = 74.4 (body) and maxWithGi = 76.0 (scale, +1.6 allowance). Replace with real division limits when wired to a class table.

3.1 Re-weigh-in (Taekwondo, Judo)

✅ A second weigh-in caps rehydration: the athlete may not exceed +5% of the division limit at re-weigh. This limits how aggressively they can dehydrate, because they must be able to recover within the cap.

3.2 Weigh-in timing → rehydration window

✅ Documented per sport; 🟡 implication for the cut:

Same-day (amateur boxing/Muay Thai) → ~no rehydration window → cut must be almost entirely body-fat/true weight; water cuts are unsafe.
Day before (MMA, TKD, Judo, pro) → ~24–32 h to rehydrate → moderate water cut viable.
Within 2 hours (BJJ) → minimal rehydration → conservative water cut only.

🟡 Proposed rule: weigh-in timing sets the maximum safe fight-week water cut and therefore how much of the gap can be left to fight week vs. must be lost as true weight during camp.

3.3 Out-of-healthy-range guard (±20 kg) ✅ (built — `GoalOutOfRange`)

When the entered limit is more than ~20 kg from current weight (either direction), the goal step raises a calm "Out of healthy range" gate instead of silently building a plan: it names the risk plainly ("This target looks like a mistake"), points the athlete to a doctor + sports dietitian, and states that OnWeight won't build an aggressive plan toward an unsafe target. A "Keep it anyway" branch lets the athlete proceed against advice (they own the number), but the recommendation and tone stay conservative. Within ±20 kg there is no popup. Lives in fc-strategy-info.jsx; see §18.

4. Trend & projection (the engine)

4.1 Trend smoothing ✅ (implemented)

Daily weigh-ins are noisy (water, food, glycogen). The displayed trend is a 5-day exponentially-weighted moving average:

`` trend[i] = Σ ( weight[i-k] · 0.75^k ) / Σ ( 0.75^k ) for k = 0..4 ``

Raw daily values render as faint dots; the trend is the bold line.
trendingToday = the most recent trend value (today's "true" weight).

4.2 Velocity ✅

`` weeklyVelocity = 7-day change in the trend // kg/week, negative = losing dailyVelocity = weeklyVelocity / 7 ``

4.3 Projection ✅ (implemented, linear)

`` daysToWeighIn = campDays − todayDay projected = trendingToday + dailyVelocity × daysToWeighIn // TRUE-weight projection gap = projected − bodyTarget // + = will be over the limit toGo = trendingToday − bodyTarget // kg still to lose from today ``

🟡 Currently a straight-line extrapolation of current pace. A more honest model would project along the planned rate (the chosen cutRate) and flag divergence — see §6.

4.4 Cycle-aware projection ✅ (implemented) — the weigh-in buffer

The number that matters is what the scale reads on weigh-in day, not the athlete's underlying true weight. If weigh-in lands in a water-retention phase (luteal), the scale reads higher than true weight — so the plan must pre-cut extra true weight as a buffer.

`` weighInOffset = cycleOffsetKg(weighInDay) // §7 — kg the scale reads above true weight that day projectedScale = projected + weighInOffset // honest scale reading at weigh-in cycleBuffer = max(0, weighInOffset) // extra TRUE weight to pre-cut trueTarget = bodyTarget − cycleBuffer // the true-weight number the plan aims at scaleGap = projectedScale − bodyTarget // the gap that actually decides made/missed ``

The projection cone (Trend tab) draws a widening fan from today to weigh-in; its upper edge carries weighInOffset, so the honest "could I be over?" risk is visible. Verdict: cone clears / straddles / misses the limit. This replaces the over-confident single dashed line (addresses F4).
The cycle-adjusted plan (Cycle screen) draws the planned glide path undulating with the cycle: the expected scale reading = true plan + cycleOffsetKg(day), dipping in follicular and rising in luteal, but always landing on bodyTarget at weigh-in. The true-weight target sits cycleBuffer below it.
Worked example (prototype): weigh-in lands in luteal with weighInOffset ≈ +0.5 kg, so trueTarget = 74.4 − 0.5 = 73.9 kg. The athlete must trend to 73.9 true so the scale shows 74.4. (Default amplitude is now 0.5 kg — see §7; personalised upward from logged data, e.g. the +0.8 kg in the Learned from your data screen.)

5. Cut-rate strategy (the pace thresholds) ✅

Source: WeeklyLossRate. Rates are % of body weight per week:

Option	Rate	Guidance
Maximum	1.0% / wk	Beyond this, muscle loss is likely, impacting strength & power. (Hard ceiling.)
Aggressive	0.75% / wk	Effective but may impact performance during camp.
Recommended	0.5% / wk	Optimal — preserves muscle, supports strong performance.
Conservative	0.25% / wk	Slow & steady; minimal performance impact, needs a longer timeline.

`` plannedWeeklyLossKg = currentWeight × (cutRate / 100) ``

✅ 1.0%/wk is the safety ceiling — sustained loss faster than this = "cutting too fast".
🟡 Feasibility check at onboarding: if (currentWeight − bodyTarget) can't be reached by the fight date at ≤1%/wk, warn the athlete (timeline too short / consider a higher class).

6. Pace classification — the gauge

Source: Trending Weight & Phases shows a 4-zone gauge: Low · On Track · High · Risk, evaluated on the 7-day average. This maps to the four feedback screens (AW_08A–AW_8D).

Gauge zone	Meaning	Feedback screen	Tone
On Track	Loss rate ≈ planned; projection lands at/under target	`AW_08A_OnTrack`	green
Low	Losing too slowly; projection misses target but recoverable	`AW_08C_SlowProgress`	amber
Risk (slow)	So far behind it isn't safe to make weight at ≤1%/wk	`AW_8D_NotOnTrack_CTA`	red
High	Losing faster than planned; over-cutting	`AW_08B_TooFast`	amber

🟡 Proposed thresholds (NEED CONFIRMATION). Let r = current 7-day-avg loss as % bw/wk and target = chosen cutRate:

On Track: target − 0.15% ≤ r ≤ target + 0.20% and projected gap ≤ +0.2 kg.
Low (behind): r < target − 0.15% and the gap is still closable by date at ≤1%/wk.
Risk (off-track): the remaining gap cannot be closed by the fight date at ≤1%/wk → CTA to change plan / move up a class.
High (too fast): r > target + 0.20% (and/or r approaching/over the 1%/wk ceiling).

🔴 Confirm the exact boundary numbers and whether the gauge is driven purely by rate (r) or by projected gap, or a blend of both.

7. Menstrual-cycle adjustments 🟢 (evidence-based defaults — now sourced) — drives the math, learns per athlete

Source: GS_13A_MenstrualCycleWeightChanges + a literature review (2024–25). Shown only when biologicalSex = female and the athlete opts in (GS_09A). All values are customisable defaults; contraceptive users may set all to 0; athletes with absent/irregular cycles are handled separately (§7.4).

✅ What the evidence says (replaces the old guesses).

The change is small, fluid-only, and absent in many women. A large systematic review (Kanellakis et al., 2023, Am. J. Human Biology) found the majority of studies report no significant body-weight/composition difference between phases.
Where detected, the magnitude is ~0.45 kg average (full-cycle tracking), explained almost entirely by water, not fat; individual ranges run up to ~2 kg.
Mechanism & timing are well-established: progesterone peaks mid-luteal → fluid retention; late-follicular estrogen raises extracellular fluid. Retention peaks late-luteal into the first 1–2 days of menstruation, then clears. In athletes, body mass + total body water rise from follicular → luteal.
Implication: the old +1.0 / +2.0 kg luteal defaults overstated it, and a −1.0 kg follicular offset has no basis — follicular is baseline. Default amplitude is now 0.5 kg, and the product leads with annotation over arithmetic because the effect is so individual.

7.1 The water-weight curve ✅ (implemented)

One smooth curve is the single source of truth for every cycle chart and the projection (§4.4), so they can never disagree. Water offset (kg the scale reads above true weight) follows a cosine over the cycle: it dips to −retention in follicular (~day 10) and peaks at +retention in late luteal (~day 24).

`` cycleDayAt(campDay) = (dayToday + (campDay − todayDay)) mod cycleLength cycleOffsetKg(campDay) = retention × −cos( 2π · (cycleDayAt − 10) / cycleLength ) cyclePhaseAt(campDay): <5 menstrual · <13 follicular · <16 ovulation · else luteal ``

Model offset (cosine, around the cycle mean):

Phase	When	Model offset at phase centre	Note
Menstruation	days 0–5	~0 kg (falling off the luteal peak)	puffiness often lingers day 1–2
Follicular	days 5–13	−retention (≈ −0.5 kg)	water low — lightest readings
Ovulation	days 13–16	~0 kg (rising)
Luteal	days 16–28	+retention (≈ +0.5 kg)	water retention — heaviest readings

User-facing onboarding defaults (simplified, relative to baseline, all editable): Luteal +0.5, Menstruation +0.3, Follicular 0, Ovulation 0. retention is the peak amplitude (default 0.5 kg). The slight framing difference (cosine-around-mean vs. baseline-relative phase numbers) is intentional: the curve drives the math; the phase numbers are the human-editable view.

7.2 How it feeds logic ✅

Projection / buffer — §4.4. The weigh-in-day offset sets cycleBuffer and trueTarget.
Plan line — undulates by cycleOffsetKg(day) so the planned scale reading rises in luteal and dips in follicular, landing on bodyTarget.
Trend interpretation — the Cycle screen annotates the current phase ("expected +0.5 kg hold — read the trend, not the scale") instead of a "behind" verdict.
Adaptive weekly goal — §9: nextWeekGoalKg −= Δphase_nextWeek.

7.3 Personalisation from logged data ✅ (designed) — Learned from your data screen

The per-phase amplitude is a starting estimate, then learned per athlete. Once ~2–3 cycles are logged, fit each athlete's real luteal amplitude from the cyclic residual of their weigh-ins (the swing left after removing the downward cut trend) and recommend an update (e.g. default +0.5 → measured +0.8). The athlete accepts or keeps the default — never silent.

🟡 Caveats to honour in the model: (1) needs ≥2–3 cycles; (2) the cut trend confounds the cyclic signal — must separate trend from residual; (3) round conservatively and label it a personalised estimate, not medical advice.

7.4 Absent / irregular cycles ✅ (designed) — Cycle Regularity screen (Refine step 4)

Hard camps and aggressive cuts frequently disrupt or stop periods. Captured as Regular / Irregular / Not currently:

Regular → full cycle-aware math (above).
Irregular → down-weight date-based phase prediction; lean on logged-data personalisation (§7.3).
Not currently (amenorrhoea) → cycle math off, and surface a gentle, non-diagnostic note: a lost period can signal low energy availability / RED-S (bone, recovery, power cost) — suggest a check-in with coach/doctor. Never diagnose.

✅ The prototype now uses the evidence-based default (luteal peak +0.5, follicular −0.5, amplitude retention = 0.5), personalised from data and always overridable. This resolves the old +2.0 / +1.0 / −1.0 discrepancy and the F8 menstruation-note contradiction (menstruation now carries a small positive lingering hold, consistent with the "may rise" note).

7b. Weigh-in time & consistency ✅ (implemented)

The daily fasted weigh-in is only comparable day-to-day if it's taken at a consistent time. The app treats consistency as the priority and first-thing-fasted as the accuracy ideal.

7b.1 Preference (`weighInTime`, `weighInTargetHour`)

Set at onboarding (GS_02_WeighInTimePreference); editable any time from Plan → Daily weigh-in, or the "Change time" action on the Trend-tab time-drift card.
🟡 Rule: switching the preferred window (e.g. morning → evening) is allowed — what breaks the signal is scattering, not the hour itself. First-thing-fasted is flagged as most accurate, but a consistent evening weigh-in is fine. When the athlete changes the window, the app should re-baseline the "comparable window" band to the new time rather than warn.

7b.2 Per-entry time ✅

Every entry stores its time of day, not just the date. The Trend-tab log shows the time under each weigh-in.
Off-window flag: a reading more than ±1 hour from weighInTargetHour is marked off-window (noisier — water/food/activity drift). Surfaced as a red time label in the log and a red dot on the time-drift scatter (Trend tab).

7b.3 Double weigh-ins ✅ (canonical rule)

If the athlete logs more than once on the same day, the earliest fasted reading is canonical — it's the one used for the trend, projection and verdict. Later same-day readings are kept and shown as context ("2nd reading · not used"), greyed, never feeding the trend.
🔴 Confirm: should a later reading ever replace the canonical one (e.g. the first was a mis-entry)? Current rule keeps earliest and lets the user delete the wrong one.

8. Fight-week cut strategies ✅

Source: CutStrategy. These are fight-week tactics, separate from the daily camp cut rate.

Strategy	Effect	Recommendation
Low Fibre Diet	Flush gut, ~1% of scale weight	Recommended
Low Sodium Diet	2–3 days, ~0.5–1% water weight	Recommended
Sweat Out	Sauna/sweat	⚠️ Never more than 5% of bodyweight
Low Carb	Glycogen depletion	❌ Not recommended — 24 h to reload, fuels performance
Water Load	Manipulate water, temp. ~0.5% drop	❌ Not recommended for female athletes

🟡 Selected strategies define the fight-week portion of the cut (the water/manipulation buffer), which combines with the camp's true-weight loss to reach scaleTarget. Total water-cut should respect the per-strategy ceilings (esp. Sweat Out ≤5%) and the re-weigh cap (§3.1).

9. Adaptive weekly goal & the tracking cascade

✅ Documented as a flow ("Log weight → adaptive weekly goal updates → feedback"; "Edit Plan → recalculated targets") and shown per-week in Weeks (Check-ins x/7, Tracking Weight, Weekly Loss %, Average weight loss kg).

9.1 The tracking cascade ✅ (CONFIRMED — evidence-based, sports-dietitian + research review)

Principle: daily scale weight is mostly water/gut noise and is stressful for athletes to interpret. We never judge a cut on a raw daily reading. We escalate through timescales, and every comparison is against the cycle-adjusted expected line (§7), never a raw straight line.

The evidence that sets the thresholds. Within-subject day-to-day body-mass SD is ~0.5 % of bodyweight under standardised weighing (≈0.4 kg at 78 kg), and larger free-living (2-wk change SD ≈1.1–1.3 kg; ~630 kcal/day intake swing ≈ 1 kg). Averaging only cuts noise by √n, so:

Window	residual noise (1 SD)	true loss in window @0.5–1 %/wk
1 day	0.4–0.8 kg	0.06–0.11 kg
3-day avg	0.23–0.46 kg	0.17–0.34 kg
7-day avg	0.15–0.30 kg	0.39–0.78 kg

→ A 3-day average's own noise is as big as the signal AND bigger than a ±200 g band, so a ±200 g pass/fail on 3 days misclassifies ~⅓–½ of the time. The signal only clears the noise at the 7-day rate (and 7-day-vs-prior-7-day). Therefore:

Step 0 — Displayed trend weight. The only weight the athlete sees as a verdict is a 7-day-equivalent smoothing (EWMA, smooth()), never the raw daily number. Raw daily is logged, charted as dots, but carries no verdict.

Step 1 — Daily silent gate. If today's reading is within ±200 g of the cycle-adjusted expected weight → silently "fine", stop. Off it → escalate silently (no alarm, no red).

Step 2 — 3-day provisional filter. 3-day trailing average within ±0.4 kg (≈0.5 %, matched to its real noise — not 200 g) of the cycle-adjusted expected weight → treat as noise, on track. Else → go to the rate check.

Step 3 — 7-day rate verdict. Compare the last-7-day vs prior-7-day true-weight rate (cycle-water removed). Requires ≥4 logged days per window; with <14 days of data, fall back to an all-data regression. This rate drives the on-track / cutting-fast / behind verdict.

Symmetry. Being under the line (losing too fast) flags the same way as being over — but the response differs: too-fast → add food back; behind → tighten. (Voice per §11.)

9.2 Two independent re-plan triggers ✅

A "Change your plan" flow launches when either:

(A) Projection miss — the projected scale weight (trend + weigh-in-day cycle water, §4.4) misses the limit, confirmed across 2 readings (hysteresis — never one noisy window).
(B) Safety ceiling — measured rate > 1 %/wk, independent of whether weight will still be made. Making weight is not the only goal; not wrecking performance/health is the other.

Anti-flapping: separate enter/exit thresholds and a 2-reading confirmation before the flow is offered, so the athlete is never bounced in and out of "off track".

9.3 Weekly goal recalculation ✅

Each week, re-derive the remaining required pace from the current trend so the plan self-corrects: `` weeksLeft = daysToWeighIn / 7 requiredWeekly = (trendingToday − trueTarget) / weeksLeft // trueTarget already cycle-buffered (§4.4) nextWeekGoalKg = clamp(requiredWeekly, 0, currentWeight × 1%) // never exceed the 1%/wk ceiling nextWeekGoalKg −= Δphase_nextWeek // cycle offset (§7) ` If requiredWeekly` would exceed the 1 %/wk ceiling, don't silently cap and carry on — fire re-plan trigger (A). Missed check-ins: a window with <4 logs holds the prior verdict rather than guessing.

9.4 Cycle amplitude default ✅

Population default water amplitude = 0.5 kg (evidence: ~0.45–0.5 kg average premenstrual gain, fluid not fat; many women see none, some up to ~1.4–2.3 kg). Individualised from the athlete's own logged residuals (§15.4), learned upward or downward. Never assume a large swing for everyone.

9.5 Two-tier detection — drift-watch (early) + confirmed verdict (late) ✅ (CONFIRMED — evidence + detection-theory review)

The problem the dietitian raised: a 7-day rate is statistically sound but slow — for a fighter on a tight deadline, "by day 7 it's too late." Lost runway can't be recovered.

The resolution (not a binary): a 7-day window is rolling — it updates every day, not a checkpoint you wait for; and we layer a faster early-warning tier on top. Detection theory is clear there's an unavoidable trade between detection delay and false-alarm rate, so we place that trade deliberately: early + soft (high sensitivity, off the alarm channel) → confirmed + hard (high specificity, drives the re-plan flow).

Tier	Method	Speed	Surfaced as	Can fire re-plan?
Trend	EWMA (`smooth()`) — recent days weighted heaviest	bends in ~2–3 days	the only weight shown	no
Drift-watch	CUSUM of daily deviation vs cycle-adjusted line (`driftWatch()`)	catches consistent drift in ~2–3 days	quiet, non-red note	no (soft only)
Confirmed verdict	7-day rolling rate + daily projection (§9.1, §9.2)	confirmed, low false-alarm	the verdict + Change-plan	yes

Why EWMA, not a flat average. An EWMA "will detect shifts of 0.5σ to 2σ much faster than a simple chart with the same sample size" because recent data is weighted heaviest. Responsiveness is tunable via the smoothing factor (β high = smoother/slower, β low = faster/noisier).

Why CUSUM for the early tier. A one-sided cumulative sum S = max(0, S + (deviation − k)) accumulates small same-direction deviations "to achieve the effect of amplification," making a persistent drift visible "much faster" than a windowed average — while the slack k (~0.15 kg/day) absorbs ordinary water noise so a single bad-water morning never trips it. It is soft-only: it opens a watch, never the re-plan flow.

Deadline-tightening (context-driven). The CUSUM decision limit h shrinks as fight day nears (h = 0.85 → 0.65 → 0.45 kg at >28 / ≤28 / ≤14 days out). A fighter's costs are asymmetric — a missed "you're behind" is far worse than an extra soft nudge — so we buy faster detection late, when lost runway is most expensive, at the price of a few more (still non-alarming) watches.

Two protections that matter more than raw detection speed:

Don't plan to the wire — planned rate is capped under the 1 %/wk ceiling with margin (§14.5), so a few slow days don't immediately become a crisis.
Projection always visible — the cycle-adjusted cone updates daily, so trajectory-to-fight-day is on screen continuously, not revealed at a checkpoint.

Symmetry. Drift-watch runs two-sided: behind (losing too slow) and fast (losing too quick) trip independently, each with its own supportive response (§11).

10. Key user flows ✅

The app is paywalled end to end — there is no free tier and no Pro tier. First-run is therefore one funnel with one hinge: the paywall. Before it, the job is convert; after it, activate. (The earlier Free-Onboarding / Pro-Goal-Setting tier split is retired — those tiers don't exist.) See OnWeight - Onboarding & Goal Setting Flows.html for the strategy + visual map (the prior tier-based map is preserved as …Flows v1.html); live screens are in OnWeight - All Screens.html.

Flow 1 · Onboarding (pre-paywall) — prove value, earn the subscription (eyebrow "OnWeight Setup"). Six high-signal questions, each reshaping the plan; account & permissions deferred.

Launch → Sport → (Fight date | No fight) → Biological sex → Current weight → Goal weight → Cut rate → Building your camp → Coach's Read → Plan Preview · Paywall

Each pre-paywall question must change the plan or be cut. The Coach's Read is the engineered "aha" before price; the paywall delivers a plan the athlete watched build, headline echoing their own numbers. The hinge: subscribe → create account to save the camp → setup. A hard paywall has no free-home fallback — declining ends the session, so re-engagement (reminder / email / win-back) is the only recovery path and must be built. (Trial vs hard-paid: still open — recommend a free trial; longer trials convert materially better.)

Flow 2 · Set up camp (post-paywall) — precision, then a first win (eyebrow "Set up your camp", no tier pill). Friction acceptable; the new-user experience runs to the first logged weigh-in. `Weigh-in time → Morning reminder (push opt-in, deferred) → [if female & opted-in] Cycle on? → Period date → Cycle length → Phase weight changes → Confirm competition rules (equipment, equipment weight, re-weigh-in + buffer, hydration clause/cap) → Cut strategies → Check details → Refining plan →

Choose plan shape (Tapered | Linear) → Log first weigh-in → Home`

Cut strategies moved here (fight-week detail, not a conversion lever). Account creation happens at purchase; the notification permission is deferred to Step 2. The cycle sub-flow is gated on biologicalSex = female (captured pre-paywall) and the opt-in; otherwise it skips straight to competition rules. Every cycle screen is independently skippable. Ends on the first weigh-in — the activation milestone that predicts retention.

Log weight (Home quick action or Logbook) Enter weight → pick date → Save → recompute trend + projection → adaptive weekly goal update → Feedback screen (On Track / Too Fast / Slow / Off-track CTA)

Edit Plan Change fight date / weigh-in / division / equipment / cycle settings → recalculated targets

Browse Learn → watch video / read article → (optional) save.

View Strategy Guidance → what's generally recommended / NOT for your sport & weigh-in timing.

11. Prototype reconciliation (what to change)

When wiring the prototype to this spec:

Pace states → computed. Replace the four hand-picked PACE_STATES presets with a classifier that takes one weight series + cutRate and derives the gauge zone via §6 thresholds.
Cycle defaults — ✅ resolved. The build now runs the evidence-based default retention = 0.5 kg (§7/§14.3), personalised from logged data and always overridable. The old +2.0/−1.0 guesses and the "§7 is OPEN" hold are retired.
Add onboarding decisions that actually feed logic: cutRate (§5) and per-sport rules (§2.2) driving bodyTarget (§3).
Projection — ✅ cycle-aware scale projection + projection cone added (§4.4). Optionally also switch the central line to plan-relative.
Adaptive goal — implement §9 once the recalculation rule is confirmed.
New Trend-tab instruments — ✅ projection cone (§4.4), cut burndown (kg-to-go vs days-to-go), weigh-in time-drift scatter (§7b), and the cycle-adjusted plan chart (§7.2). All theme/tone-aware.

12. Open questions — ✅ now resolved in §14

These were parked; all six are now answered against the literature + governing-body rules in §14. The values there are evidence-based defaults, still user-overridable.

✅ Gauge thresholds (§6 → §14.1) — two signals (symmetric pace gauge + projection cone); 1%/wk ceiling.
✅ Adaptive weekly-goal rule (§9 → §14.2) — re-derived weekly off current weight; total-cut capped.
✅ Menstrual-cycle values (§7 → §14.3) — 0.5 kg default (0–2.5 range); peak shifts to late luteal.
✅ Division limits (→ §14.4) — per-sport class table; athlete picks a division.
✅ Same-day weigh-in sports (→ §14.5) — yes, water cutting disabled; distinct "walk-it-down" plan.
✅ Feasibility warning (→ §14.6) — required-rate vs 1%/wk ceiling, with three coach-toned outs.

13. ⚠️ Rules that don't fully hold up (flagged for review)

Things that are internally inconsistent, ambiguous, or look risky as written. None are blocking, but they should be resolved before this logic is trusted in a build.

F1 · The gauge can't carry two kinds of "bad" on one axis. ✅ Resolved (§14.1) — split into a symmetric pace gauge (too slow ↔ on-plan ↔ too fast, hard red at ≥1%/wk) plus a separate projection signal for "behind / off-track."
F2 · Two different windows for the same idea. ✅ Resolved (§14.1) — the 7-day change is the canonical current rate (gauge + weekly goal); the 5-day EWMA stays for the display line only.
F3 · "% of body weight" — of which weight? ✅ Resolved (§14.1) — recompute the weekly kg target off current weight each week (literature definition; self-conserves as the athlete leans out).
F4 · Linear projection is twitchy. ✅ Addressed — the Trend tab now shows a projection cone (§4.4) instead of a single line, making the uncertainty (and the cycle-water risk on the upper edge) explicit. The central estimate is still current-pace linear; plan-relative is optional.
F5 · Camp cut + fight-week cut aren't reconciled against a total ceiling. ✅ Resolved (§14.2/ 14.5) — the two are summed and checked against acute ≤5 % body-mass and the re-weigh/rehydration cap; the plan won't schedule more water than the window can return.
F6 · Same-day weigh-in breaks the model. ✅ Resolved (§14.5) — same-day sports use a distinct "walk-it-down" plan type (no fight-week water buffer; trueTarget = scaleTarget), selected automatically by weigh-in timing.
F7 · No-fight users have no deadline, but pace needs one. ✅ Resolved (§16) — open-ended users run a distinct Maintenance mode: no countdown, no projection cone. Instead a maintenance band (self-set, with an evidence-based recommendation) + a readiness readout (weeks to a safe cut).
F8 · Cycle menstruation note contradicts its value. ✅ Resolved — §7 now uses one smooth cosine curve (follicular low / luteal peak) at a single retention amplitude (0.5 kg), replacing the contradictory per-phase constants. Still an unvalidated amplitude.
F9 · Re-weigh-in "5%". ✅ Resolved (§14.4) — it's 5 % of the division limit, weighed in gear (IJF random morning control; WT similar). Rehydration cap = 1.05 × divisionLimit in gi.
F10 · Equipment allowance direction. ✅ Resolved (§14.4) — the division limit is the scale limit; gear sports subtract equipmentWeight to get the body target, underwear-only sports use it directly.
F11 · Hydration-clause fields are dead by default. Every default sport has hydration clause = "no" (§2.2), so hydrationCap never engages. Fine to keep for other promotions, but confirm it's intentionally unused rather than missing data.

14. Research-backed resolutions ✅ (June 2026)

The open questions in §12 and several flags in §13 are resolved below against sports-science literature and governing-body rules. Numbers are defaults, not medical advice — every value stays user-overridable, but these are the evidence-based starting points.

Sources (combat-sport weight-cutting science): Reale, Slater & Burke, Acute-weight-loss / rapid-weight-loss reviews (IJSNEM); Ruiz-Castellano et al. 2021, Achieving an Optimal Fat Loss Phase in Resistance-Trained Athletes (Nutrients); Hagmar et al. & Stachenfeld, menstrual-cycle fluid balance; ACSM/IOC weight-management position stands; governing bodies: IJF SOR, World Taekwondo, IBJJF, UFC / ABC Unified Rules, IBA / USA Boxing.

14.1 Safe cut-rate & the gauge (resolves Q-gauge, F1, F2, F3)

Evidence: a loss rate of 0.5–1.0 % of body mass per week preserves fat-free mass; ~0.5 %/wk is optimal for retaining muscle/strength, while sustained loss > 1 %/wk reliably costs lean mass and performance. This validates the §5 table and fixes 1.0 %/wk as a hard safety ceiling.
F3 — % of which weight: recompute the weekly kg target off current weight each week (the literature defines the rate this way; it also self-conserves — the kg target shrinks as the athlete leans out). So plannedWeeklyLossKg = currentTrendWeight × cutRate%.
F2 — one window: the 7-day change is the canonical "current rate" that drives the gauge and weekly goal. The display trend line may stay a 5-day EWMA, but the rate quoted anywhere = 7-day.
F1 — the gauge needs two signals, not one axis. Ship two:
1. Pace — a symmetric gauge of 7-day loss rate: too slow ← on-plan → too fast, with a hard

red zone at ≥ 1.0 %/wk.

Projection — will the cone (§4.4) land at/under the limit by weigh-in? (the "behind / off-track" signal). "Dangerously behind" and "dangerously fast" are different signals and must not share one bar.
Thresholds (defaults): let r = 7-day loss %bw/wk, target = chosen cutRate.
On track: target−0.15 ≤ r ≤ target+0.20 and projection cone clears the limit.
Too fast (amber): target+0.20 < r < 1.0.
Unsafe fast (red): r ≥ 1.0 (the ceiling — muscle-loss zone).
Behind (amber): projection misses but the gap is still closable by date at ≤ 1.0 %/wk.
Off-track (red): gap cannot be closed by the date at ≤ 1.0 %/wk → feasibility CTA (§14.6).

14.2 Adaptive weekly goal & total-cut ceiling (resolves Q-adaptive, F5)

Re-derive each week from required pace, clamped to the ceiling, minus the cycle offset: `` requiredWeekly = (trendingToday − trueTarget) / weeksLeft nextWeekGoalKg = clamp(requiredWeekly, 0, currentTrendWeight × 1.0%) − Δphase_nextWeek ``
Missed check-ins: never punish — recompute from the last known trend point and re-pace. A missed week just steepens requiredWeekly (and may trip the feasibility CTA).
F5 — total cut is now bounded. Camp true-weight loss plus the fight-week water cut are summed and checked against two caps: (a) acute fight-week water loss ≤ 5 % of body mass (see 14.5), and (b) the re-weigh / rehydration cap (14.4). The plan refuses to schedule more water than the rehydration window can return.

14.3 Menstrual-cycle defaults (resolves Q-cycle)

Evidence: most menstruating athletes retain fluid in the late-luteal → first 1–2 days of menses window and are lightest in the mid-follicular phase; typical swing ~0.5–1.0 kg, individual outliers to ~2–2.3 kg. The vast majority experience some perimenstrual water retention.
Decision: keep the smooth cosine (§7.1) at a default amplitude of 0.5 kg, user-adjustable 0–2.5 kg; contraceptive / absent-cycle users set 0. (This supersedes the earlier 1.0 kg figure: the build runs FC.cycle.retention = 0.5 — see §7/§9.4 — because the average cyclic swing is ~0.45 kg, fluid not fat, and many athletes see none. 1.0 kg was the original guess. ⚠️ Note one build inconsistency to fix: the Learned-from-your-data card still labels the population "typical" as +1.0 kg (LEARNED_CYCLE.populationPeakKg) while the default amplitude is 0.5 — align these.) Refinement: shift the modelled peak to late luteal / menses onset (it currently peaks ~day 24; move toward day 26–28) and the trough to mid-follicular (~day 8–10). The shape is validated; only the peak position needs the nudge.

14.4 Division limits & equipment direction (resolves Q-limits, F9, F10)

Q-limits: ship a per-sport class table of official division limits; the athlete picks a division (autofills scaleTarget) instead of typing a blind goal weight. Maintain one table per org (Boxing, MMA/UFC, IBJJF, IJF, WT, Muay Thai).
F10 — direction is settled: the published division limit is the scale limit.
- Weighed in gear (judo/BJJ gi, TKD): the limit is the clothed weight → `bodyTarget =

divisionLimit − equipmentWeight` (the body must come in lighter).

Underwear-only (boxing, MMA, Muay Thai): bodyTarget ≈ divisionLimit.
F9 — the "5 %" is settled: for Judo (IJF) the official weigh-in is the evening before; a random control on the competition morning requires the athlete to be ≤ +5 % of the category limit, weighed in judogi. So the rehydration cap = 1.05 × divisionLimit (in gear), i.e. 5 % of the division limit, not body weight. World Taekwondo uses the same day-before + same-day random-control concept where applied. This caps how hard the athlete may dehydrate (they must be able to rebound under the cap).

14.5 Same-day weigh-in needs a different plan type (resolves Q-sameday, F6)

Evidence: rapid (water) weight loss only avoids a performance penalty when there is a rehydration window (broadly: ≤ 5 % of body mass, regained over ~24 h). Cuts > 5 % and/or with no rehydration window measurably impair performance and stress the kidneys.
Decision — weigh-in timing sets the max safe fight-week water cut: | Weigh-in timing | Sports | Max planned fight-week water cut | |---|---|---| | Same day | amateur boxing / Muay Thai | ~0 % — water cutting disabled; the gap must be all true weight | | Within ~2 h | IBJJF | minimal (conservative only) | | Day before (~24–32 h) | MMA, TKD, Judo, pro boxing/MT | up to ~5 %, bounded by the re-weigh cap (14.4) |
F6 — distinct plan model: same-day sports use a "walk-it-down" plan — no fight-week water buffer, trueTarget = scaleTarget, the whole cut is true-weight loss across camp at ≤ 1 %/wk. This is a different plan type from the camp-cut-plus-water model, selected automatically by weigh-in timing.

14.6 Feasibility warning (resolves Q-feasibility)

At onboarding and on every plan edit: `` requiredAvgWeekly = (currentWeight − bodyTarget) / weeksToWeighIn // off current weight infeasible = requiredAvgWeekly > currentWeight × 1.0% // can't be done safely veryHighCut = (currentWeight − bodyTarget) / currentWeight > 0.10 // >10% total — flag risk ``

If infeasible: coach-toned, not alarmist — offer three outs: move up a division, move the date, or start now at the ≤ 1 %/wk ceiling and reassess. Never silently plan an unsafe rate.
If very-high-cut: surface a health note even when technically feasible by date.

15. Cycle intelligence 🟡 (researched design — June 2026)

The cycle work is bigger than one water-offset number. It spans who has a cycle to model, a health safeguard, where you are right now, symptom logging, learning from past months, and education. Built on the menstrual-cycle science (§7/§14.3) plus the sources below.

Sources: Elliott-Sale et al. (cycle methodology); Menstrual Cycle and Hormonal Contraceptives in Female Athletes: Should Symptoms and Nutrition Matter More than Cycle Phase? (Nutrients 2026, doi 10.3390/nu18071144); IOC RED-S consensus (2014/2018/2023); ACSM Female Athlete Triad; PMC9724109 (RED-S female athlete); PMC7937612 / PMC8281678 (MC & contraception vs strength).

Guiding principle (from the 2026 review): individual, logged symptom & weight patterns should outweigh textbook phase math. Generic phase recommendations are applied **cautiously, and only once a consistent personal pattern is documented.* This is the spine of §15.4 (learning).

15.1 Cycle mode — three states, set at onboarding & editable

The phase model only applies to one of these. Captured at GS_09A, editable in Cycle settings.

Mode	Who	What the app does
Natural cycle	Eumenorrheic, no hormonal contraception	Full phase model (§7.1): water offset, undulating plan, buffer. Refined by learning (§15.4).
Hormonal contraception	Pill / patch / ring / hormonal IUD / implant / injection	No natural-phase model — HC suppresses endogenous fluctuation. Default flat (0 offset). If the method has a withdrawal/placebo week the user can flag it; otherwise rely on logged symptoms only. Effects are modest & highly individual — never assume.
Absent / irregular	No period, post-menopause, pregnancy, or irregular	No phase model. If "absent" is unexpected → RED-S safeguard (§15.2). Irregular cycles fall back to symptom + learned patterns, not calendar prediction.

🟡 Default offset for HC and Absent modes is 0 kg until the athlete's own data says otherwise.

15.2 The RED-S safeguard ⚠️ (health-critical — non-negotiable)

Combat & weight-category sports carry elevated RED-S risk (low energy availability), and losing your period is a primary warning sign, not a convenience for the model.

Detection: secondary amenorrhea = 3+ consecutive missed expected cycles; oligomenorrhea = cycle length > 45 days. The app already tracks period dates, so it can detect both.
Behaviour when detected:
1. Never treat a missing period as "0 offset, carry on." Surface a calm, non-judgemental health

note: a lost/irregular period during a hard cut can signal low energy availability (RED-S).

Recommend a check-in with a doctor / sports physician. Provide an education piece (§15.5).
Soften, don't push: if RED-S risk is flagged, the app must not suggest a faster cut or a bigger deficit; bias toward the conservative end and the feasibility outs (§14.6).
🔴 Product decision needed: how forcefully to intervene (gentle nudge vs. hard gate that pauses aggressive-cut features until acknowledged). Recommend at least a one-time acknowledgement gate.
⚠️ This is a wellness signal, not a diagnosis — wording must avoid alarm and avoid medical claims.

15.3 "Where am I?" — phase indicator + symptom logging

Phase indicator: show current phase, cycle day (day X / length), and the plain-language read ("late luteal — expect to hold ~+1 kg water"). One glance, interpretation-first.
Symptom check-in (optional, fast): let the athlete log bleeding (start/flow), bloating, cramps, cravings, mood, energy, sleep. Two jobs:
1. Confirm/flag the period — a logged bleed start re-anchors the cycle calendar (more accurate

than pure prediction) and resets the amenorrhea counter.

Feed learning (§15.4) — symptoms + the weight swing on those days teach the personal model.
Predicted vs confirmed: calendar prediction is a guess; a logged bleed is truth. Always prefer confirmed. Surface "period expected ~Fri" as a soft prediction the athlete can confirm or correct.

15.4 Learning — get smarter each month 🟡 (the headline feature)

Rather than holding the §7 constants forever, personalise from the athlete's own logged history (weight swings aligned to confirmed cycle days, plus symptoms). This is exactly what the 2026 review argues for.

`` For each completed, confirmed cycle: alignedSwing[phase] = mean(trend-detrended weight) within that phase // remove the cut trend first learnedOffsetKg(phase) = weightedMean(alignedSwing[phase] over last N cycles) // recent cycles weighted higher learnedPeakDay = cycle day of the athlete's observed water maximum confidence = f(number of confirmed cycles, consistency of pattern) ``

Cold start: use §7 population defaults (1.0 kg luteal peak). As confirmed cycles accumulate, blend toward the learned values, surfacing confidence ("based on your last 3 cycles…").
Detrending is essential: the athlete is losing weight on a cut, so the cycle swing must be measured as the residual around the trend, not raw weight.
What gets smarter: the per-phase amplitude, the peak day/timing, and which symptoms reliably precede a water hold for this athlete → better buffer (§14.4), better plan undulation, better "expected, not behind" messaging.
Always overridable & transparent: show what was learned and let the athlete correct it. Low confidence → stay near population defaults and say so.
🔴 Decide N (cycles in the window), the confidence function, and the cold-start→learned blend curve.

15.5 Education pieces (Learn tab)

Short, coach-toned, sport-specific. Each ties to a moment in the app:

"Why the scale lies in your luteal phase" — water vs fat; read the trend, not the day. (Linked from the Cycle screen & a luteal-phase stall.)
"Losing your period isn't a flex" — RED-S in plain language, why it matters for power/bone/long-term, when to see a doctor. (Linked from §15.2.)
"On the pill? Here's what changes" — why the phase model is off, what to watch instead. (Linked for HC mode.)
"Timing your cut around your cycle" — using the buffer & timing strip (§14.4, cycle×fight-date) strategically.
🔴 Confirm tone/length and whether these are articles, short videos, or both (the Learn tab supports either).

15.6 What to build (prototype)

Cycle mode selector (natural / contraception / absent) driving the engine (§15.1).
Phase indicator card + symptom check-in sheet (§15.3).
RED-S safeguard state on the Cycle screen + its education piece (§15.2/§15.5).
"Learned from your cycles" surface — learned offset/peak with confidence, vs population default (§15.4).
Learn-tab education entries (§15.5).

16. Maintenance / no-deadline mode 🟡 (researched design — June 2026) — resolves F7

When there's no fight date, the countdown engine (projection cone, pace gauge, adaptive goal) has nothing to anchor to. Maintenance mode replaces it. Weight-only — this app is about making weight on the scale, so no body-composition input here.

Sources: ISSN 2024 position stand, Nutrition and weight-cut strategies for MMA & combat sports (PMC11894756); Weight cycling in combat sports: 25 years of evidence (BMC, PMC8670259); Weight loss in combat sports (J Int Soc Sports Nutr, 1550-2783-9-52); Physiological Perturbations: Weight Cycling & Metabolic Function (PMC10890020).

16.1 Three no-deadline users (all supported)

User	Has division?	Maintenance target
Off-season	yes	Recommend a band from the division; they confirm/adjust.
No fight booked	yes	Same as off-season — stay fight-ready.
General / no division	no	Fully self-set goal weight or range; no division math.

16.2 The maintenance band — self-set, with a recommendation + reasoning

Evidence: off-camp, athletes should sit ~12–15 % above the division limit (ISSN 2024) — high enough to train/build, low enough that the next cut stays achievable at ≤ 1 %/wk. A standing buffer is healthy: 2–3 % RWL has insignificant drawbacks.
Decision (✅): the target is self-set, but the app recommends a band and explains why (e.g. "most fighters sit 8–15 % over their class between camps so the next cut isn't a crash — for you that's 80–86 kg"). Recommendation is derived from the division; the athlete owns the number.
States: in band (good) · drifting high (gentle nudge — the next cut is growing) · near/above the re-entry ceiling (the next safe cut would exceed what a sensible timeline allows).

16.3 What replaces gauge + projection (no date) — band adherence + readiness

Both signals, no countdown:

Band adherence — where today's trend sits in the maintenance band.
Readiness — "from today, a safe cut to [division] takes ~X weeks at ≤ 1 %/wk." `` readinessWeeks = max(0, (currentTrend − bodyTarget) / (currentTrend × 1%)) `` Always tells the athlete the standing cost of getting back to weight — the no-deadline analogue of the projection.

16.4 Anti-cycling guardrail — gentle nudges + education (✅)

Evidence: repeated balloon-and-recut (some judo athletes 10+ cycles/year) carries serious chronic health consequences. Off-camp guidance: avoid gaining fat, keep training, don't let the walk-around weight run away.
Behaviour: gentle, educational nudges — never hard gates. If weight drifts toward/past the re-entry ceiling, surface a coaching nudge linked to an education piece ("why ballooning between camps makes the next cut harder & riskier"). Optionally track cuts-per-year as a soft signal.
No lower guardrail (✅): maintenance mode does not police being too lean (that's RED-S’s job in §15.2 when a cycle is tracked); maintenance only watches the upper drift.

16.5 Weigh-in cadence (✅)

Default stays daily, but maintenance offers a "drop to weekly" option — daily fasted weigh-ins with no deadline risk fatigue/obsession, so the athlete can relax cadence off-camp.

16.6 Transitions — manual, with prompts (✅)

Camp → maintenance and maintenance → camp are athlete-confirmed, not silent. The app prompts ("Your fight's done — switch to maintenance?" / "Booked a fight? Start a camp.") and flips only on confirmation, then auto-configures.
Carries over: walk-around baseline (seeds the band), division, and learned-cycle data (§15.4).

16.7 Learning opportunities (surface, don't lecture)

Maintenance is a teaching moment between camps. Where appropriate, attach short education: the maintenance-band reasoning (16.2), the anti-cycling piece (16.4), and "what your walk-around weight says about your division" — all in the calm, coach-toned Learn style (§15.5).

16.8 What to build (prototype)

Maintenance Today state (no countdown) — band adherence + readiness readout.
Maintenance band card with the self-set target + recommendation/reasoning.
Anti-cycling nudge (gentle, education-linked) when drifting high.
Cadence option (daily ↔ weekly) in settings.
Transition prompt (camp↔maintenance confirm).
Learn entries (16.7).

17. Per-sport rules & guidance — sourced (June 2026)

The experience should feel bespoke to each sport: the weigh-in timing, equipment, re-weigh rule and the performance reason behind the cut all differ. This section is the sourced source-of-truth.

17.1 Confidence legend

🟢 Verified this session against the governing body / peer-reviewed literature (cited).
🟡 Established structure — used only where a body has no universal standard (e.g. Muay Thai reference classes); figures there are illustrative, not authoritative.

17.2 Weigh-in & rules by sport

Sport	Weigh-in timing	Equipment	Re-weigh / hydration	Plan type	Conf.
Boxing	Amateur: same-day; Pro: day before	none (underwear/shorts)	amateur uses same-day weigh-ins to curb cycling	amateur → walk-it-down; pro → camp	🟢
MMA	Day before (a.m.); some promotions add hydration tests (ONE Championship)	none	regain typically 8–15 lb / ~4–8% bw; some orgs cap/ test hydration	camp (water cut viable)	🟢
Muay Thai	No universal standard — varies by sanctioning body; pros ~24h before, amateurs (IFMA) same-day	none (fight attire)	varies; ONE ≤5% regain	free-entry goal weight; same-day → walk-it-down	🟢
BJJ (IBJJF)	Same-day, in the gi, immediately before you compete	gi (weighed dressed)	single weigh-in, no tolerance, no re-weigh	minimal water cut	🟢
Taekwondo (WT)	Day before + random same-day control	dobok (gear)	random control ≤ +5 % of category limit → DQ (WT Competition Rules Art. 9 §2.2, confirmed Jun 2026)	camp	🟢
Judo (IJF)	Day before (since 2017) + random comp-day control	judogi (gear)	random control ≤ +5 % of category limit	camp	🟢
Wrestling (UWW)	Same-day, morning of competition, in singlet; multi-day events re-weigh each morning	none (singlet)	no tolerance; same-day → almost no rehydration window	walk-it-down	🟢

Sources: World Boxing Competition Rules (Nov 2024, in force 1 Jan 2025); amateur same-day weigh-in practice (World Boxing / Olympic). IJF 2017 rule change — day-before official weigh-in + random competition-day control allowing ≤5 % above the category upper limit (Ceylan/ScienceDirect S2211266925000830; PMC10484915). World Taekwondo Olympic categories (olympics.com; WT) + random weigh-in +5 % tolerance — confirmed Jun 2026 against WT Competition Rules Article 9 §2.2 ("the random weigh-in shall be conducted with plus 5% tolerance of the contestant's weight category"; random check on the morning of competition; plus-categories excluded; no 2nd attempt). UWW senior freestyle categories + same-day weigh-in (uww.org, re-confirmed Jun 2026). ISSN 2024 position stand (PMC11894756) for MMA regain & hydration.

17.3 Division tables (current best, with confidence)

🟢 Boxing (World Boxing Elite, from 1 Jan 2025) — Men (max kg): 50, 55, 60, 65, 70, 75, 80, 85, 90, +90. Women: 48, 51, 54, 57, 60, 65, 70, 75, 80, +80. (Pro boxing is a separate ~17-class, pound-based system with day-before weigh-ins — pull fresh if supporting pro.)
🟢 Taekwondo (WT Olympic) — Men: −58, −68, −80, +80. Women: −49, −57, −67, +67. (WT World Championships use 8 categories/gender — finer; confirm if targeting WCh not Olympics.)
🟢 Judo (IJF senior) — Men: −60, −66, −73, −81, −90, −100, +100. Women: −48, −52, −57, −63, −70, −78, +78.
🟢 MMA (ABC Unified, lb) — Flyweight 125, Bantam 135, Feather 145, Light 155, Welter 170, Middle 185, Light-Heavy 205, Heavy 265; Super-Heavy 265+. Women: Strawweight 115 (52.2 kg), Flyweight 125, Bantam 135, Feather 145. Day-before morning weigh-in (early format since 2016); 1-lb non-title allowance. (ABC Unified Rules PDF.)
🟢 BJJ (IBJJF, adult, WITH gi) — Men (gi-on, max kg): Galo 57.5, Pluma 64, Pena 70, Leve 76, Médio 82.3, Meio-Pesado 88.3, Pesado 94.3, Super-Pesado 100.5, Pesadíssimo +100.5. Women (gi-on): Galo 48.5, Pluma 53.5, Pena 58.5, Leve 64, Médio 69, Meio-Pesado 74, Pesado 79.3, Super-Pesado +79.3 (women's gi has no Pesadíssimo — tops out at Super-Pesado open). Single same-day weigh-in in the gi, no tolerance. (Both genders re-confirmed vs IBJJF Jun 2026: women's gi Rooster 48.5, Heavy 79.3.)
🟢 BJJ (IBJJF, adult, NO-GI) — Men (max kg): Galo 55.5, Pluma 61.5, Pena 67.5, Leve 73.5, Médio 79.5, Meio-Pesado 85.5, Pesado 91.5, Super-Pesado 97.5, Pesadíssimo +97.5. Women: Galo 46.5, Pluma 51.5, Pena 56.5, Leve 61.5, Médio 66.5, Meio-Pesado 71.5, Pesado 76.5, Super-Pesado +76.5. Weighed in fight kit (rashguard + shorts), ~2 kg under the gi ladder. (IBJJF no-gi table, re-confirmed Jun 2026.)
🟢 Wrestling (UWW senior freestyle) — Men (max kg): 57, 61, 65, 70, 74, 79, 86, 92, 97, 125. Women (freestyle): 50, 53, 55, 57, 59, 62, 65, 68, 72, 76. Same-day morning weigh-in, in singlet. (Olympics use the 6-class subset 57/65/74/86/97/125; app uses the full 10-class World-Champs ladder. Re-confirmed against uww.org, Jun 2026.)
🟢 Muay Thai — confirmed: no universal table. WBC Muaythai ≈17–19 boxing-style classes; Thai stadiums (Lumpinee/Rajadamnern) differ; IFMA amateur uses its own. Pros weigh ~24h before, amateurs (IFMA) same-day, in fight attire. → Use free-entered goal weight.

17.4 Performance & cut guidance per sport (the why, sourced)

General anchors (apply to all): safe loss 0.5–1.0 %/wk (§5/§14.1); acute fight-week water only where a rehydration window exists, ≤5 % bw (§14.5); combat & weight-class sports carry elevated RED-S / weight-cycling risk (§15.2). Per-sport emphasis:

MMA — longest rounds; as a bout passes ~4 min, aerobic contribution can exceed 70 % while anaerobic bursts decide exchanges (ISSN 2024). Day-before weigh-in makes a moderate water cut viable, but the cut must protect repeat-effort recovery across rounds — gradual descent, fuel the engine.
Boxing — amateur same-day weigh-in → no rehydration window → water cutting is unsafe; the whole gap must be true weight (walk-it-down). Pro day-before allows a moderate cut. Either way, cardio-over-rounds and a clear head (dehydration impairs cognition) are the priorities.
BJJ — weigh-in in the gi, minutes before you roll → almost no rehydration → conservative, true-weight cut only; a depleted grappler gasses out fast. Don't sacrifice grip/anaerobic capacity.
Judo / Taekwondo — day-before + the 5 % random control caps how hard you may dehydrate (you must rebound under the cap on comp morning). Explosive, repeated high-output efforts → preserve power; the 5 % cap is a built-in anti-cut guardrail to lean on.
Muay Thai — varies; where it's same-day, treat like amateur boxing (walk-it-down). Clinch & sustained pace reward an intact aerobic base.
Judo physiology note — judoka commonly lose 3–6 % fat mass in the final weeks, occasionally up to ~10 % body mass; combined high training load + RWL raises muscle-damage/injury risk (PMC5968945) — a reason to hold the gradual line.

17.5 Verification status

🟢 All seven sports verified against governing-body rules / peer-reviewed literature: Boxing (World Boxing 2025), MMA (ABC Unified Rules), BJJ gi + no-gi (IBJJF, men + women), Judo (IJF 2017 + 5% control), Wrestling (UWW senior freestyle), Taekwondo (WT Olympic + +5 % random control, confirmed vs WT Competition Rules Art. 9 §2.2), Muay Thai (confirmed no universal standard → free-entry). Implementation cross-check is logged in OnWeight - Rules Verification.md.

Open items for human sign-off:

🟡 Boxing category names — the app now uses kg-only labels (50 kg … +90 kg) instead of pro-style names, since World Boxing amateur categories are kg-designated and pro labels (cruiser/ heavy) don't match pro limits. Confirm official World Boxing category names if they exist.
🟡 Pro boxing (~17-class pound system) and WT World-Championship 8-class set are not modelled — pull fresh if those competition contexts are ever supported.

(Resolved Jun 2026: Taekwondo +5 % tolerance — confirmed vs WT Competition Rules Art. 9 §2.2. BJJ women's-gi rooster — confirmed 48.5 kg, Heavy 79.3 kg vs current IBJJF tables.)

18. Additions since the core flows were written (June 2026)

These areas were designed after §1–§17 and are live in OnWeight - All Screens.html. They don't change the cut logic above; they make the product launch-ready. Each is documented here so this file stays the single source of truth.

18.1 Lexicon — "goal" → "limit" (app-wide)

The athlete-facing word for the fight-day scale number is "limit" (the weight you must hit), not "goal weight." The onboarding screen reads "Your limit / The weight you must hit on the scale," and charts label the line LIMIT. Rationale: a fighter doesn't aspire to the number, they're bound by it — "limit" carries the consequence "goal" doesn't. Code/field names stay goalWeight, bodyTarget, scaleTarget (changing them risks regressions); only display copy uses "limit." Any new copy must follow this — see OnWeight - Copy Rules.html §5 lexicon.

18.2 Accounts & safety (`fc-account-safety.jsx`) — launch-blockers

The returning-user / legal-gate screens onboarding never covered:

Create account — social-first (Apple / Google) + email, with a consent line.
Verify email — 6-digit confirmation.
Safety disclaimer (required acknowledgement) — three plain commitments: no crash cuts ever, guidance not medical advice, listen to your body — gated behind an "I understand" checkbox. This is the product's front-line health gate and pairs with the §15.2 RED-S posture.
Age gate — date of birth; 18+ proceeds.
Age gate · minor — under-18 routes to a guardian-consent path (youth combat sport is real; the cut science and safety burden differ, so minors aren't given an adult aggressive-cut flow).

Placement: account creation happens at purchase (§10, Flow 1 hinge); the safety disclaimer and age gate are part of that account step. 🔴 Product decision: exact ordering of disclaimer vs. payment.

18.3 Legal & data rights (`fc-legal.jsx`)

In-app, readable Terms of Service and Privacy Policy (numbered sections, including a real Health & safety clause: "OnWeight is guidance, not medical advice… you use your plan at your own discretion"), plus data rights controls — export my data and delete my account/data. Needed for App Store / Play compliance and GDPR-style requests.

18.4 Imperial / metric parity (`fc-units.jsx` + `fc-imperial.jsx`)

A central OWUnits conversion + formatting engine so every weight renders in kg or lb from one setting — no hardcoded units. fc-imperial.jsx demonstrates the key screens in lb (US/UK athletes, and the lb-native division ladders e.g. MMA/ABC). Rule: never hardcode a unit string; format through OWUnits. Division tables stay canonical in their governing-body unit and convert for display.

18.5 Cut-strategy info sheets (`fc-strategy-info.jsx`) — restored from Figma `GS_06A_Info`

Each fight-week strategy (§8) now has a tap-through ⓘ explainer bottom sheet: Low Fibre · Low Sodium · Sweat Out · Low Carb · Water Load. Each sheet carries headline · how much can you lose? · how do you do it (food examples / dosages) · warning, with the safety ceilings stated inline (e.g. Sweat Out: never more than 5 % of body weight; stop if dizzy/faint). These restore content that had survived only as one-line card descriptions.

18.6 Widgets (`fc-widgets.jsx`) — post-launch retention surface

Home Screen (small + medium) and Lock Screen complications showing the cut at a glance — weight, on-track state, days/■ to weigh-in — without opening the app. Design-for-sign-off; not a logic change.

18.7 Documentation coverage note

The §10 flow maps predate 18.2–18.3. When the onboarding/camp flow diagrams are next revised, splice in: account creation + safety disclaimer + age gate at the paywall hinge, and a settings → legal / data-rights branch. Tracked here so the omission is explicit, not silent.

18.8 BFM design-audit changes (Jun 2026) — see `OnWeight - BFM Audit (read).html`

A start-to-finish Built for Mars critique drove a round of fixes; all are live in the prototype and All Screens (shared JSX) unless noted, with the source-of-truth and flow maps updated to match.

Onboarding / paywall (fc-onboarding.jsx, fc-screens-b.jsx):

Social proof, earlier + on the hinge — grounded in real, cited category research (no fake user counts; OnWeight is pre-launch). The paywall now carries a distinct full-bleed "proof band" (not the rounded card component — fixes the earlier card-on-card confusion): the sport-specific stat is the hero (e.g. 97% of MMA fighters cut to make weight), then a real, attributed fighter quote in serif italic (tap to cycle — FC_FIGHTER_QUOTES, balanced male/female: Robert Whittaker & Kayla Harrison, all verbatim), then the cross-sport safety/authority line as the quietest footnote. The same band is reused on the Coach's Read (compact variant, no quote) via a shared ProofBand component, so social proof reads consistently across the app. Sport content keyed in FC_SOCIAL_PROOF (fc-engine.jsx), falling back to the combat-wide 60–80% figure. Full sourcing in OnWeight - Social Proof Research.md.
Trial = lifeboat, honestly framed — 7-day free trial with "we'll remind you 2 days before your trial ends." (BFM recommendation: a 14-day trial converts better; final length is a business call.)
Paywall offer = 2–3 plans, annual default — annual pre-selected and priced as a low monthly equivalent with the saving shown; monthly kept (Anchoring + Default Bias + Framing). Prices placeholder.
Division picker — "Your limit" leads with a one-tap class picker (natural class starred, gear allowance applied); keypad kept as the custom fallback (see §14.4).

Activation (fc-empty.jsx + prototype routing):

First win before configuration — after purchase + account the athlete lands on the dashboard empty state; precision setup is the opt-in "Sharpen your plan" card, not a forced gauntlet (§10, Flow 2).

Daily loop (fc-screens-a.jsx):

Habit reward — a calm consistency strip on the post-log result ("every morning this week"), which rewards showing up without shaming pace.

Cycle (fc-engine.jsx):

Number alignment — LEARNED_CYCLE.populationPeakKg set to 0.5 kg so "typical" matches the evidence default (§14.3); the demo athlete's learned +1.4 stays, to show personalisation working.

Future-state concepts (fc-future.jsx, in the All Screens "Future · sandbox" section only — never wired from a live screen, per the launch-point rule):

Win-back (FutWinBack) — the re-engagement recovery the hard paywall depends on (§10).
Next camp, smarter (FutNextCampSmarter) — a missed weight closes on what the data says for next time (a fresh-start hook); supported, never judged.
Reminder re-ask + widget (FutReminderReask) — hardens the daily-loop trigger for push-declined users, with the Home Screen widget as a no-permission backup.
Coach view (FutCoachView) — a coach's read-only roster of their fighters (the accountability / social layer); the athlete still owns the plan.

Still open (business / content, not design): final trial length + price; annual-only vs annual+monthly; and replacing the labelled placeholders (social-proof fighters/quotes, win-back & widget pricing) with verified content before launch.

17.6 Changelog

Jun 2026 — Added Wrestling (UWW) + BJJ no-gi to the source-of-truth (were implemented but undocumented). Fixed Boxing in fc-engine.jsx: added the missing +90 open class (was 9 of 10 divisions) and switched to kg-only labels. Flagged Taekwondo +5 % tolerance for re-confirmation.
Jun 2026 — Women's divisions built for all seven sports in fc-engine.jsx (divisions.women): Boxing (World Boxing), MMA (UFC/ABC + Strawweight), BJJ gi + no-gi (IBJJF), Taekwondo (WT Olympic), Judo (IJF), Wrestling (UWW women's freestyle). Muay Thai women's set is illustrative (free-entry). OnbGoalWeight now takes a sex prop so the recommender returns the correct ladder; a female-athlete Goal Weight artboard was added to the canvas.
Jun 2026 — Resolved both prior open items with official sources: Taekwondo +5 % confirmed vs WT Competition Rules Art. 9 §2.2 (conf → verified); BJJ women's gi corrected to Rooster 48.5, Heavy 79.3 (was 48 / 79) per current IBJJF tables. Only product-decision items remain open (boxing category names; pro-boxing / WT-WCh sets not modelled).