Nexus Prophet – AI Draft Draft Analyzer

Nexus Prophet is an AI-powered League of Legends draft analyzer that predicts win probability from 5v5 team comps using a ML model and a sleek cyberpunk web dashboard.

Comment

💡 Inspiration

As eSports data becomes more complex, the gap between "casual play" and "pro analysis" widens. We wanted to bridge this gap by building Nexus Prophet-an AI-powered drafting assistant that uses Logistic Regression to predict match outcomes based on champion composition. Our goal was to answer the question: "Can a simple AI model identify winning synergies from professional match data?"

🤖 What it does

Nexus Prophet is a full-stack drafting dashboard where users can simulate a 5v5 League of Legends draft.

  1. Drafting UI: Users select 5 champions for "Team Order" (Blue) and "Team Chaos" (Red).
  2. AI Prediction: The app sends the composition to our Python backend.
  3. Real-Time Analysis: The model calculates the win probability using a trained Logistic Regression classifier.
  4. Result: The UI displays a dynamic "Win Probability" bar, visualizing the AI's confidence.

⚙️ How we built it

We adopted a "Vibecoding" methodology-leveraging AI agents to accelerate development.

  • Backend: Built with Python and Flask. We used scikit-learn to train a Logistic Regression model on match data.
  • Frontend: A React (Vite) dashboard with a Cyberpunk/eSports theme, featuring neon accents and responsive design.
  • Data Pipeline: We integrated the GRID Open Access GraphQL API to fetch professional match series data.
  • AI Assistance: We used JetBrains AI Assistant (Junie) to generate boilerplate code for the API fetchers and React components, allowing us to focus on architecture and logic.

🧠 The Math Behind the Magic

We use a Logistic Regression model where the probability ( P ) of Team A winning is given by the sigmoid function:

$$ P(y=1|x) = \frac{1}{1 + e^{-(\beta_0 + \beta_1 x_1 + ... + \beta_n x_n)}} $$

Where:

  • ( x ) represents the feature vector (one-hot encoded champions).
  • ( \beta ) represents the learned weights for each champion's impact on victory.

🚧 Challenges we faced

  • API Authentication: Connecting to the GRID Open Access API presented strict permission scopes (UNAUTHENTICATED errors). We built a robust Hybrid Data Pipeline that attempts to fetch real data but gracefully falls back to synthetic training data if the API connection times out.
  • Time Pressure: Building a full-stack app in a short timeframe meant we had to prioritize core features over perfection.

🏅 Accomplishments that we're proud of

  • Successfully integrating a Machine Learning model into a live web app.
  • Creating a "Fail-Safe" architecture that ensures the app always runs, even if the data source fails.
  • Designing a slick, professional UI/UX that fits the Cloud9 eSports aesthetic.

🚀 What's next for Nexus Prophet

  • Deep Learning: Upgrading from Logistic Regression to a Neural Network for capturing complex champion synergies.
  • Live Draft Integration: Connecting to the live League Client API to suggest picks in real-time.
  • More Games: expanding support to VALORANT and Dota 2.

Built With

Share this project:

Updates