AI-powered analysis for prediction markets with on-chain signal publishing and cross-platform trading.
Fourcast helps traders find edge in prediction markets through:
- AI-powered analysis - Venice AI (Llama 3.3 70B) with 200+ ML models across crypto, sports, politics & more
- Multi-platform trading - Polymarket & Kalshi integration with live odds
- On-chain signals - Publish verifiable predictions to Movement/Aptos blockchain
- DeFi arbitrage - Detect price discrepancies across platforms
- Live Market Analysis - AI-generated predictions with confidence scores (HIGH/MEDIUM/LOW)
- ML-Backed Forecasts - SynthData integration for crypto/equity price predictions (BTC, ETH, SOL, SPY, NVDA, etc.)
- Sports & Events Intelligence - Weather-aware analysis for sports and event markets
- Cross-Platform Arbitrage - Find mispriced odds between Polymarket and Kalshi
- Direct Trading - Place orders on Polymarket/Kalshi from the interface
- Signal Publishing - Publish predictions on-chain (Movement testnet)
- Reputation System - Track win rates, Brier scores, and accuracy streaks
- Tipping - Earn APT for high-quality signals
- Leaderboards - Compete with other analysts
- Unified API - Access markets, signals, and AI analysis
- SDK - Build custom signal publishers and domain analyzers
- Modular Architecture - Extend with new data sources and chains
graph LR
A[Polymarket/Kalshi] -->|Live Odds| B(AI Analysis Engine)
C[Weather API] -->|Real-time Data| B
D[SynthData] -->|ML Forecasts| B
B -->|Signals| E[SQLite Database]
B -->|Publish| F[Movement Blockchain]
E --> G[Frontend Dashboard]
F -->|Verify| G
G -->|Trade| A
- Frontend: Next.js 15, React 19, Tailwind CSS, Three.js
- Backend: Node.js, SQLite (Turso), Redis
- AI: Venice AI (Llama 3.3 70B) with Edge Search
- Blockchains:
- Movement/Aptos (signal publishing, tipping)
- BNB, Polygon, Arbitrum (trading contracts)
- Data Sources: Polymarket, Kalshi, Open-Meteo, SynthData
git clone https://github.com/thisyearnofear/fourcast.git
cd fourcast
npm installThe pre-commit hook will install automatically to protect against committing secrets.
Copy .env.local.example to .env.local:
# Required: AI & Weather
VENICE_API_KEY=your_venice_api_key
NEXT_PUBLIC_WEATHER_API_KEY=your_weather_api_key
# Optional: Movement (signal publishing)
NEXT_PUBLIC_APTOS_NETWORK=custom
NEXT_PUBLIC_APTOS_NODE_URL=https://testnet.movementnetwork.xyz/v1
NEXT_PUBLIC_MOVEMENT_MODULE_ADDRESS=0x...
# Optional: Polymarket Trading
POLYMARKET_PRIVATE_KEY=your_private_key
# Optional: Kalshi Trading
KALSHI_API_KEY=your_api_keySee Setup Guide for complete configuration.
npm run devNavigate to http://localhost:3000
fourcast/
├── app/ # Next.js pages and components
├── components/ # React components (UI, 3D, trading)
├── services/ # Backend services (AI, markets, chains)
├── sdk/ # Signal SDK for developers
├── move/ # Move contracts (signal registry)
└── docs/ # Documentation
| Platform | Purpose | Status |
|---|---|---|
| Polymarket | Live odds, trading | ✅ Live |
| Kalshi | Live odds, trading | ✅ Live |
| Movement/Aptos | Signal publishing, tipping | ✅ Testnet |
| BNB Chain | Trading contracts | 🧪 Beta |
| Polygon | Trading contracts | 🧪 Beta |
| Arbitrum | Trading contracts | 🧪 Beta |
- Features - What works today vs. future roadmap
- Setup Guide - Installation, environment, wallets
- Architecture - System design and data flows
- API Reference - All endpoints and schemas
- Deployment - Production deployment guide
We welcome contributions! Areas of interest:
- New data sources (sentiment, on-chain metrics)
- Additional prediction market platforms
- EVM chain integrations
- UI/UX improvements
MIT
Fourcast the future. We're building quantitative prediction intelligence for the decentralized web — combining ML forecasts, AI reasoning, and real-time data to create verifiable, composable signals that traders and developers can trust.
Today: AI analysis across crypto, sports & politics; signal publishing; cross-platform trading
Future: Premium signals, developer ecosystem, expanded data integrations