Bet-Studio AI Match Insights

AI-powered football insights using Chrome's Gemini Nano. Analyzes matches locally with Prompt & Summarizer APIs. Private, fast, hybrid architecture.

Comment

Inspiration

Traditional sports analytics show only match statistics. Bet-Studio wanted to solve a harder problem: correlating live sports metrics with market dynamics to detect mispriced opportunities in real-time. Current solutions require expensive server infrastructure and raise privacy concerns. Chrome's Gemini Nano enables instant, private, multi-dimensional analysis impossible before.

What it does

Bet-Studio AI Match Insights uses Chrome's Prompt API and Summarizer API to perform hybrid sports + market analysis:

Unique Value Proposition:

  • Analyzes both sports metrics AND market data (odds movements, volume, momentum)
  • Detects value opportunities where match performance diverges from market pricing
  • Works in prematch AND live scenarios
  • Surfaces only high-confidence matches (AI-filtered top opportunities)
  • 100% private on-device processing

How it works:

  1. Fetches match data + market data from bet-studio.com API
  2. Prompt API analyzes multi-dimensional patterns (xG, possession, shots, corners, odds movements, market volume)
  3. Summarizer API generates "Why NOW?" insights explaining the opportunity
  4. Displays top 3-6 matches in interactive flip cards
  5. Hybrid fallback ensures always-working UX

Key Benefits:

  • ⚡ 200ms analysis vs 2-3s server (10-15x faster)
  • 🔐 100% private - sensitive market data stays on-device
  • 💰 Zero AI costs - no server compute
  • 🎯 AI-curated selection (not all matches, only opportunities)
  • 🔄 Works prematch + live

How we built it

Tech Stack:

  • Vanilla JavaScript, HTML5, CSS3
  • Prompt API - Multi-dimensional analysis (sports + market correlation)
  • Summarizer API - Concise opportunity explanations
  • Service Worker for real-time updates
  • Hybrid architecture (on-device primary, server fallback)

AI Analysis Pipeline:

  1. Service Worker receives match + market data
  2. Prompt API prompt: "Analyze this match considering: [sports metrics] AND [market data: odds, volume, momentum]. Score opportunity 0-100."
  3. Filters matches >75% confidence
  4. Summarizer API creates "Why NOW?" summary
  5. Frontend displays top opportunities in flip cards

Innovation: Most AI sports tools analyze ONLY match stats. We correlate sports performance + market behavior to find mispriced opportunities - this is computationally expensive and privacy-sensitive, perfect for on-device AI.

Challenges we ran into

  1. Complex Prompting - Teaching Prompt API to correlate 2 data dimensions (sports + market) → iterative prompt engineering
  2. AI Availability - Gemini Nano not everywhere → hybrid architecture
  3. Real-time Performance - Processing market + sports data <200ms → optimized prompts + caching
  4. Filtering Logic - Showing only high-confidence matches → AI-powered threshold system

Accomplishments that we're proud of

  • First sports extension correlating market + sports data on-device
  • 10-15x faster than server (200ms vs 2-3s)
  • 100% private - sensitive market analysis never leaves device
  • Zero cost - complex AI processing without server fees
  • AI-curated - shows only high-value opportunities, not noise
  • Hybrid - works everywhere with graceful degradation

What we learned

  • Chrome AI enables complex multi-dimensional analysis previously server-only
  • Prompt engineering critical for correlating diverse data types
  • On-device AI perfect for privacy-sensitive financial/market data
  • Hybrid architecture essential for real-world reliability
  • Users prefer AI-curated insights over data overload

What's next for Bet-Studio AI Match Insights

  • Translation API for multi-language market insights
  • Writer API for detailed opportunity reports
  • Real-time alerts using Alarms API (threshold-based notifications)
  • More sports (basketball, tennis) + market analysis
  • Historical pattern recognition (learning from past opportunities)
  • Chrome Web Store publication

Built With

Share this project:

Updates

Submission history