diff --git a/Runs/2026-09-23-sprint.md b/Runs/2026-09-23-sprint.md new file mode 100644 index 0000000..2771391 --- /dev/null +++ b/Runs/2026-09-23-sprint.md @@ -0,0 +1,64 @@ +--- +date: 2026-09-23 +type: run +tags: [run, afternoon, sprint, max-hr-test, under-read] +--- + +# Run — 2026-09-23 (max-HR sprint test) + +**1.39 km in 15:13 — mostly a walk/jog at 10'57" avg, plus an all-out ~100 m sprint (burst 2'34") to find max HR. The watch read 165 bpm max — almost certainly an under-read: he nearly threw up, and the sensor cadence-locked (cadence maxed at 241 spm).** + +## Data + +| Metric | Value | +|--------|-------| +| Distance | 1.39 km | +| Duration | 15:13 | +| Avg pace | 10'57" /km (walk/jog warmup) | +| Fastest pace | 2'34" /km (the sprint burst) | +| Avg HR | 128 bpm | +| Max HR | **165 bpm (under-read — see verdict)** | +| Cadence | 115 spm (max 241 — sensor locked to arm swing) | +| Stride | 79 cm · 1,764 steps | + +| Metric | Value | +|--------|-------| +| VO₂Max | 46 ml/kg/min | +| Recovery time | 8 h | +| Total / active calories | 139 / 116 kcal | + +**HR zones:** Extreme 0 · Anaerobic 1 min · Aerobic 3 min · Fat-burning 7 min · Warm-up 1 min. + +## Sleep — Night before + +**8h 20min** · 00:26→09:00 · REM: 11% · Deep: 30% · HRV: 126 · Resting HR: 49 · Score: 87/100 + +> See [[Sleep/2026-09-23]]. + +## How it felt + +> "Tried to measure my max HR with an all-out sprint but it did not measure well. HR only went up to ~165, which is not real — I could not stand still and almost threw up after the ~100 m sprint." + +## Verdict + +**The 165 is not your max HR — and the watch itself is the proof.** Three things stacked up to under-read it: + +1. **The sprint is too short for the *heart* to peak.** Heart rate physically lags effort by ~20–30 seconds. A 100 m sprint is ~15 seconds — it's over before your HR even begins to climb toward its true ceiling. Even a perfect sensor would read the *recovery* peak, not the sprint peak. +2. **The optical sensor lagged and smoothed it.** Wrist HR (light-based) averages over a ~10–20s window. A 15-second spike gets flattened out, so the "165" is closer to "where HR was ~30s after you stopped" than "where it was mid-sprint." +3. **The smoking gun: cadence maxed at 241 spm.** During a max sprint the arm swing is violent, and the wrist sensor "cadence-locks" — it mistakes your arm cadence for your pulse and reads the wrong signal. A 241 spm cadence reading next to a 165 bpm "max" is the sensor telling you it lost the pulse and grabbed the motion instead. + +You're 23 and you just nearly vomited from a 100 m effort — your true max is almost certainly **in the 185–200 range**, not 165. + +**How to actually measure it (for real):** + +1. **Warm up properly** — 15–20 min easy, so the heart is already responsive before the test. +2. **Make the peak effort long enough to register** — a single 15s sprint can never do it. The standard field test: after the warmup, do **3 hard uphill repeats of ~2–3 min** (or 400–600 m all-out) with ~1–2 min recovery, finishing with one final **30–60s all-out** burst. The highest number in that last minute is your max. +3. **A chest strap is the real answer.** For max-HR and sprint work, the wrist optical sensor fundamentally can't track fast changes — a chest strap (ECG) will catch the actual peak. If max HR matters to you (it sets your training zones), it's worth borrowing/grabbing one for the test. + +**But — don't over-index on this right now.** For setting training zones, your watch's *current* estimates (and especially your lactate-threshold HR, which the watch tracks better than raw max) are more useful than a one-off max. The useful numbers from today are already in the log: 143 bpm avg at 7'03"/km on the 2 km leg, VO₂Max stable at 46. You don't need a perfect max-HR number to keep training well this month. + +## Connected + +- [[Sleep/2026-09-23]] (the night before) +- [[Runs/2026-09-23]] (the 2.01 km warmup leg ~30 min earlier) +- [[Half-marathon-plan]] (the block this max-HR test was meant to calibrate) diff --git a/Runs/2026-09-23.md b/Runs/2026-09-23.md new file mode 100644 index 0000000..20b43fc --- /dev/null +++ b/Runs/2026-09-23.md @@ -0,0 +1,55 @@ +--- +date: 2026-09-23 +type: run +tags: [run, afternoon, easy, back-from-rest] +--- + +# Run — 2026-09-23 + +**2.01 km in 14:11 — 7'03" avg at 143 bpm (max 163). A short, clean negative split (7'52" → 6'15") two days after the 10 km PR, with HR recovery a healthy 24 bpm in 2 min. The warmup leg before the max-HR sprint attempt.** + +## Data + +| Metric | Value | +|--------|-------| +| Distance | 2.01 km | +| Duration | 14:11 | +| Avg pace | 7'03" /km | +| Fastest pace | 6'15" /km (km 2) · burst 5'45" | +| Avg HR | 143 bpm | +| Max HR | 163 bpm | +| Cadence | 163 spm (max 178) | +| Stride | 87 cm · 2,318 steps | + +| Metric | Value | +|--------|-------| +| HR recovery (2min) | 24 bpm (dropped 24, 158→110) | +| VO₂Max | 46 ml/kg/min (held, "average") | +| Recovery time | 5 h | +| Total / active calories | 165 / 143 kcal | + +**HR zones:** Extreme 0 · Anaerobic 2 min · Aerobic 8 min · Fat-burning 2 min · Warm-up <1 min. **Splits:** 7'52" → 6'15" + 0.01 km in 0'04". Running form: ground contact 320 ms (Good), balance L 49.9 / R 50.1 (clean). + +## Sleep — Night before + +**8h 20min** · 00:26→09:00 · REM: 11% · Deep: **30%** · HRV: **126** · Resting HR: 49 · Score: 87/100 + +> Best stage balance in the recent log — deep back to 30%, HRV steady at 126. See [[Sleep/2026-09-23]]. + +## How it felt + +> No narrative given for this leg — it reads as the easy warmup run ahead of the sprint test. + +## Verdict + +**This is the "prove the engine is intact" run, and it does.** Two days after the 10.01 km PR, back on the legs with a short 2 km at 7'03" — a touch slower than Sep 18's 6'04" but this is deliberately short and easy, not a race. The clean **negative split** (7'52" → 6'15") says pacing and legs are both fine; no fade, no stiffness carried over from the long run. + +The controls are all where they should be: **HR recovery 24 bpm** (158→110 in the 2-min window) is a strong number, **VO₂Max held at 46**, and balance is dead clean (49.9/50.1). The ankle that was the whole story on Sep 20 gets a silent pass here — no complaint, no stop. + +The only thing this run *isn't* is a training stimulus. 2 km at 143 bpm avg is recovery-range territory, so treat it as exactly that: legs moved, engine confirmed, no load banked that needs a recovery bill. The real event of the day was the sprint attempt that followed — see [[Runs/2026-09-23-sprint]]. + +## Connected + +- [[Sleep/2026-09-23]] (the strong night this was built on) +- [[Runs/2026-09-23-sprint]] (the max-HR sprint attempt ~30 min later) +- [[Runs/2026-09-20]] (the 10 km PR this run recovers from) diff --git a/Sleep/2026-09-23.md b/Sleep/2026-09-23.md new file mode 100644 index 0000000..721402b --- /dev/null +++ b/Sleep/2026-09-23.md @@ -0,0 +1,52 @@ +--- +date: 2026-09-23 +type: sleep +tags: [sleep, recovery, solid-night, naps] +--- + +# Sleep — 2026-09-23 + +**8h 20min** · 00:26→09:00 · REM: 11% · Deep: **30%** · HRV: **126** · Resting HR: 49 · Woke: 2× · Score: 87/100 + +> A genuinely strong night — the best stage balance in the recent log. Deep sleep is back to 30% (vs 19% on the Sep 20 repayment night), REM recovered to 11%, and HRV holds at 126ms with RHR at 49. The two afternoon naps (46 min) suggest the weekend debt is still being collected, but the night itself did its job. + +**Interpretation:** The headline is the architecture. Deep sleep **30% (2h 31min)** — dead centre of the 20–60% reference and the best absolute deep since the early-September entries — plus REM **11% (54min)**, comfortably normal (and up from 7% on Sep 20). Light sleep 59% is still flagged "High" (ref <55%) but it's a small overshoot, not the 74% light-dilution of the repayment night. The stage balance finally looks like a healthy night rather than a recovery compromise. + +The autonomic side held perfectly steady: **HRV 126ms** — identical to Sep 20's baseline and dead centre of his 110–135 sober range — with **RHR 49 bpm**, his excellent low. Bed **00:26** is post-midnight, and per the bedtime-ceiling effect that's the ~120–130ms band, so 126 is right on target: no hidden suppression, nothing artificial propping the number up. + +The one soft spot the watch flagged is **2 wake-ups** (continuity 70, ref 0–1 times) — the boilerplate "repeated waking → vicious cycle" warning fired. In context it's noise: 2× is a hair over the 0–1 reference, continuity 70 is still inside normal, and the 87-point score lands at the 98th percentile. Not a real problem — worth a note only if it becomes 3×+ on consecutive nights. + +The **two afternoon naps** (29 min 14:47–15:16, 17 min 15:30–15:47 = 46 min) on top of an 8h20 night are the interesting signal. A full night *plus* 46 min of napping points to residual debt still being repaid from the weekend (the 1.5-bottle wine night Sep 19 + the 10 km PR run Sep 20), not a fresh problem — the body is collecting, and a home-office Wednesday afternoon gave it the room to. Watch whether the naps shrink once the debt clears. + +| Metric | Value | Reference | +|--------|-------|-----------| +| Night sleep | 8h 20min | 6-10h (Normal) | +| Bed → Rise | 00:26 → 09:00 | — | +| Deep sleep | **30% (2h 31min)** | 20-60% (Normal) | +| Light sleep | **59% (4h 55min)** | <55% (**High**, mild) | +| REM sleep | **11% (54min)** | 10-30% (Normal) | +| Deep sleep continuity | 70 points | 70-100 (Normal) | +| Times woke | 2 times | 0-1 (**High**, mild) | +| Breathing quality | 98 points | 70-100 (Normal) | +| HRV | 126ms | 78-137ms (Within range) | +| Resting HR | 49 bpm | 47-57 bpm (Within range, low) | +| SpO2 | 98% | 97-99% (Within range) | +| Respiratory rate | 14 brpm | 13-16 brpm (Within range) | +| Naps | 29 min + 17 min = 46 min | self-reported ~ | +| **Total sleep** | **9h 06min** (8h20 night + 46min naps) | — | +| Score | 87 / 100 | Better than 98% of users | + +### HRV / RHR progression + +| Date | HRV | RHR | Context | +|------|-----|-----|---------| +| Sep 18 | 127ms | **47** | First big sleep post-quit, bed 23:07 | +| Sep 19 | **63ms** | 57 | 1.5 bottles wine, bed 01:36, 5h30 | +| Sep 20 | 126ms | 49 | Repayment, bed 22:33, 9h41 | +| **Sep 23** | **126ms** | **49** | Solid, bed 00:26, deep 30%, +46min naps | + +## Connected + +- [[Runs/2026-09-23]] (the 2.01 km run this night fed) +- [[Runs/2026-09-23-sprint]] (the max-HR sprint attempt later that afternoon) +- [[Sleep/2026-09-20]] (the repayment night — deep 19% then, 30% now) diff --git a/sportnote-roadmap.md b/sportnote-roadmap.md new file mode 100644 index 0000000..c590fdb --- /dev/null +++ b/sportnote-roadmap.md @@ -0,0 +1,96 @@ +# SportNote — Tech Stack & Roadmap + +> Decisions from 2026-09-20 (stack pivot + product planning session with Hermes). +> Companion to [[sportnote-blueprint]] (the target experience). + +--- + +## 1. Tech Stack Decision — Pivot to Python + +| Layer | Was | Now | +|---|---|---| +| Backend | Spring Boot | **FastAPI** | +| ORM | JPA/Hibernate | **SQLModel** (SQLAlchemy + Pydantic) | +| DB | SQLite/Turso | **SQLite → Postgres later** | +| Migrations | — | **Alembic** | +| Stats | — | **pandas + scikit-learn** (polars later) | +| Agent | Spring AI (idea) | **direct LLM SDK** (LangChain only if needed) | +| Frontend | Angular (idea) | **PWA from the start** (framework TBD, Svelte/React) | +| Dev/CI | Jenkins (work) | **GitHub Actions** + Docker | + +**Why the pivot:** statistics + agentic AI are the *core* of the product, not a "later" feature. Python owns that ecosystem (pandas, scikit-learn, LangChain, MCP). One language across app + analysis + agent beats a Spring app + separate Python service. + +--- + +## 2. Product Identity + +- **"Not a tracker — an AI agent for health & sports."** +- "The equivalent of talking to Hermes about my running, but simpler with a UI." +- **No wearable required** — but **brand-agnostic screenshot OCR** is the moat (any watch → universal DB). +- **Self-report ("how did it feel?") is load-bearing** — the one universal, device-free input, and what makes raw numbers meaningful. +- **Goals are data**, not just a page — the coach consumes them as input. +- **PWA from the start** — no native Android app. + +--- + +## 3. The 6 Checkpoints + +| # | Checkpoint | Note | +|---|---|---| +| 1 | **Backend + universal DB + endpoints** | schema includes `run`, `sleep`, **and `goal`** | +| 2 | **OCR ingestion (Huawei first)** | bar = reliable on *my* screenshots, not "perfected" | +| 3 | **Manual-entry UI (PWA)** | self-report required, installable, works with no wearable | +| 4 | **Statistics** | the "visible number" | +| 5 | **Goals experience** | set + track — coach needs them as input | +| 6 | **AI coach** | reads stats + goals + self-report | + +Through-line: #1 contract → #2 proves the moat → #3 makes it usable → #4–6 are the product. + +--- + +## 4. Universal Schema Tiers + +| Tier | Data | Flag | +|---|---|---| +| **0 — Core** | date, duration, distance → **pace (derived)** | required | +| **1 — Most important** | avg HR, max HR, **HR recovery (2-min)**, sleep (HRV/RHR/REM%/deep%/total), **self-report** | required for self-report, optional wearable | +| **2 — Extra** | cadence, HR zones, pace-by-km (splits) | optional | +| **3 — Skip** | stride, ground contact, vertical oscillation, balance L/R, training load, recovery-time estimate, elevation | store nothing | + +**Rule:** *"Would this change my advice?"* — if no, skip it. + +**Correction vs earlier assumption:** sleep is NOT "extra" — it's the single most important modifier of performance. Every run log cross-references the night before. + +--- + +## 5. Key Architectural Insights + +1. **Coach = stats-computed + prompt-injected** — never free-range tool-calling, especially with a 9B local model. Backend crunches numbers with pandas → hands the model a pre-digested summary → model reasons over it. +2. **OCR's hard part is layout→schema mapping**, not OCR itself. Layered: easyocr (read text) → normalization/parse (field aliases) → vision model only as fallback. +3. **Goals are input to the coach** — "training toward Nov 29 half, 10km PR, ankle recovering" changes every recommendation. +4. **Manual entry and OCR write to the same endpoint** — the universal schema is the contract both paths hit. + +--- + +## 6. Local Dev Setup (as of 2026-09-20) + +- **Hardware:** Fedora, RX 9060 XT 16GB +- **Local model:** Qwen 3.5 9B via LM Studio (fast on the AMD card) +- **Harness:** OpenCode, pointed at LM Studio's OpenAI-compatible API (`http://127.0.0.1:1234/v1`) +- **Division of labor:** local Qwen = *pair programmer* (helps, doesn't write for me) · Hermes = *PM/thinker* (specs, review, architecture) · Gutca = *owner* (understands everything) +- **Project home:** `~/Projects/sportnote` on Fedora, pushed to GitHub + +--- + +## 7. Current State + +- ✅ Phase 0 skeleton: FastAPI + SQLModel + SQLite, `run`/`sleep`/`weight` tables, Alembic wired, pushed to GitHub. +- ⏳ Checkpoint 1: define the universal schema (the contract), then build endpoints. + +--- + +## Related + +- [[sportnote-blueprint]] — the target experience (conversation architecture, OCR pipeline, onboarding interview) +- [[Half-marathon-plan]] — current training context +- [[Health/sleep-system]] — sleep data context