Inspiration
In professional League of Legends, 2025 and 2026 marked a revolutionary competitive shift: the global adoption of Fearless Draft across the LCK Cup, LEC, LTA, and LPL. Under Fearless Draft, once a champion is selected in earlier games of a Best-of-3 or Best-of-5 series, that champion is permanently locked out for subsequent games.
Traditional static draft notebooks and spreadsheets collapse under Fearless rules. Coaching staff and analysts must simultaneously monitor player comfort pools across 10 roles, forecast pool depletion up to Game 5, identify opponent choke points, and react to real-time pick/ban timers in under 30 seconds.
We built the Cloud9 Fearless Draft & Macro Scouting Engine as a pro-grade scouting and real-time draft assistant tailored directly for Cloud9 Kia.
What It Does
- Real-time Pro Fearless State Machine: Enforces the official 20-step competitive draft sequence (10 bans and 10 picks) across Bo3 and Bo5 matches. Supports both Hard Fearless (mutual lockout: champion played by any team is banned for all remaining games) and Soft Fearless (team-specific lockout), providing instant automated rejection (
FearlessLockoutError) if an illegal locked champion is selected. - Official GRID Match Data Ingestion: Indexes 745 professional LoL series and 45 global teams from GRID esports data (
meta/), extracting detailed player mastery profiles and champion pool depths for Cloud9 (Blaber, Thanatos, Loki, APA, Zven, Tactical, Vulcan) and opponents. - Opponent Choke Point & Vulnerability Detection: Analyzes opponent champion pools per role. When an opposing player is down to $\le 2$ comfort champions in subsequent games, the engine flags a high-priority Choke Point Warning, recommending target bans to suffocate the opponent's strategy.
- AI & Statistical Recommendation Engine: Scores available champions for the active turn using a multi-objective formula:
$$\text{Score} = w_1 \cdot \text{Mastery}(p, c) + w_2 \cdot \text{Synergy}(c, \text{Allies}) - w_3 \cdot \text{Counter}(c, \text{Enemies}) + w_4 \cdot \text{ChokeBonus}$$
- Evaluates player pro mastery (win rate + sample size).
- Computes pairwise champion synergy from historical winning rosters.
- Computes head-to-head counter pick differentials against already-locked enemy champions.
- Interactive Web Canvas: A dark-themed, low-latency dashboard featuring Blue/Red side rosters, visual turn banners, recommendations with one-click lock-in, and a live Fearless series lockout ledger.
How We Built It
- JetBrains PyCharm & Python: Engineered using Python 3.12 with async
aiohttpfor high-throughput JSON REST endpoints and static UI delivery. - Strict Data Segregation: In full compliance with Cloud9 and GRID licensing requirements, zero proprietary match data is committed to the repository; the engine ingests data via an external directory pointer (
STL_DATA_DIR). - Automated Verification: Backed by a comprehensive 11/11 automated pytest suite covering data loading, state machine transitions, illegal pick rejections, and HTTP REST routes.
What's Next
- Real-time coach audio cues and integration with Cloud9 stage headsets during draft phases.
- Real-time live game telemetry stream ingestion for in-game macro win probability forecasting.
Built With
- aiohttp
- data-science
- esports-analytics
- grid-data
- jetbrains-pycharm
- league-of-legends
- python
- typescript
