Vanalysis

Inspiration

Esports teams drown in raw match data but still struggle to turn it into clear coaching decisions. We wanted to build a Moneyball-style assistant that connects micro mistakes to macro outcomes and tells a readable story that coaches and players can act on quickly.

What it does

Vanalysis ingests GRID central-data and statistics feeds to produce:

• Player micro analytics (performance, consistency, impact)
• Match and team macro reviews (win conditions, strategic breakdowns)
• Narrative summaries with story beats and key players
• Scouting and meta snapshots (stats-only sources)
• An AI coaching panel where users can drag analysis cards into a chat for contextual advice

How we built it

• Backend: FastAPI services that integrate GRID feeds, compute analytics, and assemble narrative frameworks.
• AI Coaching: Google ADK agents with a safe fallback path when rate limits or model errors occur.
• Frontend: React + TypeScript with MVVM hooks, a dashboard tab system, and drag-and-drop context cards.
• Resilience: Data gating and metadata fallbacks ensure the UI stays useful even when some feeds are unavailable.

Challenges we ran into

• Live data feed permissions were not available, which removed positions and real-time events.
• GraphQL schemas differed across feeds and required careful validation.
• Rate limits and incomplete statistics caused empty states that needed graceful handling.
• Ensuring the UI remained clear and inclusive while still dense with analytics.

Accomplishments that we're proud of

• A narrative-first dashboard that ties player metrics to team outcomes.
• Team-scope analytics that can switch between match and tournament contexts.
• Drag-and-drop coaching workflow to chain insights into actionable advice.
• Robust fallbacks that keep the product functional without live data.

What we learned

• Data reliability beats data volume; every metric needs provenance and fallbacks.
• Clear framing and readable UI matter as much as analytics depth.
• Rate limits and schema drift are inevitable—design for them early.

What's next for Vanalysis

• Add live event and position data when access is granted.
• Expand tactical visualizations with map overlays and sequence timelines.
• Improve caching and batching to reduce API load.
• Add evaluation loops so coaches can score the usefulness of insights.
• Build exportable reports for teams, tournaments, and scouting prep.

Built With

• googld-adk
• jetbrains
• llm
• python
• typescript