BowlGuard

AI-powered workload management for cricket fast bowlers — tracks bowling load, calculates injury risk using ACWR sports science, and delivers Claude AI recommendations.

---

What It Does

BowlGuard helps cricket coaches and physiotherapists manage fast bowler workloads scientifically. It implements the ACWR (Acute:Chronic Workload Ratio) — a validated sports science model from the British Journal of Sports Medicine — and layers Claude AI on top to give personalized, context-aware training recommendations.

No backend required. Runs entirely in the browser with localStorage persistence.

---

Features

Feature	Details
ACWR Engine	Calculates Acute:Chronic Workload Ratio from session history
Injury Risk Zones	Color-coded dashboard: Undertrained / Sweet Spot / Caution / High Risk
Claude AI Recommendations	AI analysis of workload trend with actionable coaching advice
Rule-Based Fallback	Works fully offline without an API key
Session Logging	Log bowling sessions with balls bowled, intensity, date
History View	Full session history with ACWR trend
Rate Limiting	Backend proxy with per-IP rate limiting (10 req/min)
Zero Dependencies	Pure HTML/CSS/JavaScript — no npm, no build step

---

ACWR Sports Science

ACWR = This week's balls bowled / 4-week rolling average

< 0.8   → Undertrained   (yellow) — increase load gradually
0.8–1.3 → Sweet Spot     (green)  — optimal training zone
1.3–1.5 → Caution        (yellow) — monitor closely
> 1.5   → High Risk      (red)    — injury risk, reduce load

Based on: Hulin et al. (2014), British Journal of Sports Medicine — the gold standard for cricket fast bowler workload management.

---

Architecture

Browser (Vanilla JS)
├── store.js      → localStorage data layer
├── acwr.js       → ACWR sports science calculation engine
├── ai.js         → Claude API client + rule-based fallback
├── dashboard.js  → Risk zone rendering + trend charts
├── log.js        → Session logging form
├── history.js    → Session history list
└── app.js        → Main controller

Optional: Python server.py → Proxies Claude API calls (CORS + rate limiting)

---

Quick Start

Option 1: Direct (no server needed)

# Just open in browser
open index.html
# Or: right-click index.html → Open with Live Server (VS Code)

Option 2: With Claude AI (full features)

# Start the proxy server
export ANTHROPIC_API_KEY=your_key_here
python server.py
# Open http://localhost:8000

---

Enable Claude AI Recommendations

By default, BowlGuard uses rule-based recommendations (works offline).

To enable Claude AI:

1. Get an API key at console.anthropic.com
2. Set env var: export ANTHROPIC_API_KEY=your_key
3. Run python server.py — it proxies requests with rate limiting

Never hardcode API keys in frontend JavaScript. The server.py proxy keeps your key server-side.

---

Project Structure

bowlguard-app/
├── index.html          # Main app (3 screens: Dashboard, Log, History)
├── server.py           # Python CORS proxy with rate limiting
├── css/
│   ├── styles.css      # Base styles, CSS variables, nav
│   ├── dashboard.css   # ACWR risk zone dashboard
│   ├── log.css         # Session logging form
│   ├── history.css     # History view
│   └── ai.css          # AI recommendation card
├── js/
│   ├── store.js        # localStorage data layer
│   ├── acwr.js         # ACWR calculation engine
│   ├── ai.js           # Claude API + rule-based fallback
│   ├── dashboard.js    # Dashboard rendering
│   ├── log.js          # Session logging
│   ├── history.js      # History list
│   └── app.js          # Main controller
└── assets/

---

Built By

Apuroop Yarabarla — AI/ML Engineer & AI Product Owner