EventGraph
Unified Intelligence for Prediction Markets
Demo
About
The problem it solves **The Problem EventGraph Solves** Prediction markets are growing rapidly, but liquidity and pricing are fragmented across multiple platforms like Polymarket, Kalshi, Limitless, and others. Today: * The same event trades at different prices on different venues * There is no unified event-level view across platforms * Traders must manually compare prices across tabs * Arbitrage and inefficiencies are hard to detect in real time * Liquidity is siloed and opaque This fragmentation creates: * Missed opportunities * Poor execution decisions * Higher slippage risk * Inefficient capital allocation **What EventGraph Enables** EventGraph aggregates prediction markets across venues into a unified event graph. Users can: * View the same event across multiple platforms in one interface * Compare prices, spreads, and liquidity instantly * Detect cross-venue inefficiencies * Identify arbitrage and sentiment divergence * Make safer, data-driven execution decisions Instead of manually checking multiple platforms, traders get a normalized, real-time intelligence layer. **How It Makes Existing Tasks Easier & Safer** Without EventGraph: * Open multiple tabs * Manually compare prices * Risk buying at worse prices * Miss cross-venue spreads * Have no structured view of global liquidity With EventGraph: * One dashboard * Cross-platform price normalization * Clear spread visibility * Liquidity transparency * Reduced execution risk **Why This Matters** As prediction markets scale onchain, fragmentation increases. EventGraph acts as the coordination and intelligence layer that makes markets more efficient, transparent, and capital-efficient. This improves: * Trader outcomes * Market efficiency * Liquidity discovery * Onchain market usability Use of AI tools and agents EventGraph uses AI to solve fragmentation across prediction markets. LLM-based semantic matching is used to identify when differently worded contracts across platforms refer to the same real-world event. This enables accurate cross-venue aggregation beyond simple string matching. AI-assisted logic is also applied to filter false arbitrage signals by accounting for liquidity depth, slippage, and latency differences between APIs. The system is structured around lightweight monitoring agents that: Continuously fetch and normalize market data Evaluate cross-venue spreads Apply execution-aware filters Surface high-confidence opportunities This architecture lays the foundation for autonomous execution agents that can eventually transact and self-sustain onchain.
How it's made
Challenges we ran into **Cross-Venue Market Normalization** Each prediction market platform structures contracts differently — naming conventions, price formats (cents vs probability), liquidity reporting, and contract logic vary across Polymarket, Kalshi, etc. The same event often appeared slightly different across platforms, making naive aggregation impossible. How I solved it: I built a normalization layer that: * Converts all pricing into unified probability format * Maps contracts to canonical event IDs * Uses fuzzy matching + metadata heuristics to align identical events * Standardizes liquidity and volume metrics This allowed me to create a true event-level graph instead of just merging raw market feeds.