Inspiration
Watching Cloud9 compete at Worlds, I noticed analysts spending hours manually reviewing opponent VODs and spreadsheets before each match. Professional esports teams pay analysts $50k-100k annually to do what should be automated. I wanted to democratize competitive intelligence - giving amateur teams the same data-driven insights that only top-tier organizations can afford.
What it does
Cloud9 Scouting Agent is an AI-powered intelligence platform that transforms raw match data into actionable competitive insights in seconds. Coaches select an opponent team and instantly receive:
- Smart Ban Recommendations: Top 3 champions to ban based on win rate and pick frequency
- One-Trick Detection: Identifies players who favor specific champions (>60% play rate)
- Aggression Profiling: Calculates early-game aggression through First Blood statistics
- Performance Trends: Visual K/D ratio analysis showing team form
- Exportable Reports: Downloadable scouting reports for team strategy sessions
How we built it
Python + Streamlit: Rapid UI development with a dark esports aesthetic.
Pandas: Efficient data manipulation and statistical analysis.
Mock Data Architecture: Built a realistic simulation layer while awaiting GRID API access, ensuring seamless integration later.
Analytics Engine: Developed three core algorithms (ban priority scoring, one-trick detection, aggression metrics).
Modular Design: Clean separation between data layer, analytics, and presentation for easy API integration.
Challenges we ran into
API Timing: GRID API key approval pending, so I architected a robust mock data system that mirrors real API responses.
Algorithm Tuning: Balancing win rate vs pick rate for ban priority required multiple iterations to match real analyst decisions.
UI/UX: Creating an "esports broadcast" aesthetic in Streamlit while maintaining data clarity.
Data Realism: Generating mock data that reflects actual League of Legends meta and player behavior patterns.
Accomplishments that we're proud of
The production-ready architecture is built with real API integration in mind, requiring just one function swap to go live. The ban priority algorithm intelligently combines multiple metrics rather than relying on raw stats alone. The dark mode UI with gradient cards and interactive charts rivals commercial tools in polish. The complete feature set spans from data ingestion to exportable reports in a single-file application, and the entire MVP was built in under 48 hours as a fully functional prototype ready for immediate testing.
What we learned
Domain knowledge matters tremendously. Understanding League of Legends meta and esports workflows was crucial for building relevant features. Building realistic simulations accelerates development when external APIs aren't ready. Streamlit proved powerful for rapid prototyping without sacrificing professional appearance. Translating analyst intuition into algorithms requires iterative refinement and deep
What's next for Cloud9 Scouting Agent
The immediate priority is GRID API integration to connect live match data for real-time scouting. Advanced analytics features will include draft phase prediction to forecast what opponents will pick, jungle pathing heatmaps, vision control scoring, and team fight positioning analysis. Historical comparison will let teams see how opponents have evolved over time. Multi-team analysis will enable side-by-side opponent comparisons. AI prediction models will calculate match outcome probability using machine learning. A mobile app will give coaches insights on-the-go during tournaments. The business model follows a freemium approach with basic reports free and advanced analytics available through subscription.
Log in or sign up for Devpost to join the conversation.