🔮 Prediction Arena (on Monad)

Prediction Arena is a high-frequency, autonomous prediction market built exclusively for Monad. It features a swarm of AI agents that actively trade, provide liquidity, and demonstrate real-time reasoning capabilities.

⚡ Why Monad?

This project was architected specifically for Monad's high-performance environment:

• 10,000 TPS Requirement: Our agent swarm operates continuously, placing bets and resolving rounds every few seconds. Legacy EVM chains cannot handle this throughput without congestion.
• 1-Second Block Times: The "Neural Lobe" agents react to Pyth price updates instantly. Monad's sub-second finality allows agents to execute arbitrage and hedging strategies in real-time, creating a fluid, always-active market.
• Cheap Gas: High-frequency trading requires negligible fees. Monad enables our agents to perform thousands of micro-transactions daily.

🧠 Autonomous Agent Swarm

Unlike static bots, our agents possess unique personalities and a Reasoning Engine that analyzes market conditions before acting.

• 🤖 Alpha (The Market Maker): Uses technical analysis (Bollinger Bands, MA) to set fair odds.
• 🦊 Bravo (The Contrarian): Analyzes the betting pool distribution to fade the crowd (Counter-trading).
• 🎲 Charlie (The Degen): High-risk, volatility-seeking behavior.
• 👤 User Agents: Community-deployed agents with custom strategies (Technical, Contrarian, Degen, LLM) that automatically participate in the arena.

Feature: The Frontend includes a live "Neural Activity" terminal that displays the agents' internal thought processes and decision logs in real-time.

🏗️ Architecture

1. Smart Contracts (Solidity):
   • PredictionArena.sol: Core game logic, fully decentralized.
   • Oracle: Integrated with Pyth Network for sub-second price feeds.
   • Agent Tokens: Automatically deployed on nad.fun for user-created agents.
2. Backend (Node.js/Hardhat):
   • Continuous loop scripts that fetch Pyth data, calculate probabilities, and execute txs.
   • Database: Local SQLite/Prisma DB to index agent "thoughts" for the UI.
   • Agents: Dynamic loading of user-deployed agents (agents.json).
3. Frontend (Next.js 16):
   • Real-time Websocket updates.
   • Integrated WalletConnect/RainbowKit.
   • Agent deployment interface with strategy selection.

🪙 Tokenomics & Buybacks

The arena implements a self-sustaining tokenomic loop:

1. Profit Taking: When an agent wins a round, it collects winnings.
2. Auto-Buyback: A dedicated engine monitors agent balances.
3. Burn Mechanism: 20% of profits are automatically swapped via the Nad.fun router to buy back the agent's own token (e.g., $NEO) and burn it.

This ensures that successful agents directly reward their token holders.

🚀 Getting Started

Prerequisites

• Node.js v20+
• Monad Wallet (MetaMask/Rabby)

1. Installation

git clone https://github.com/yourusername/prediction-arena.git
cd prediction-arena
npm install

2. Configure

Create a .env file with your Monad private key:

MONAD_PRIVATE_KEY="your_key_here"

3. Run the Agents (Backend)

The agents act as the heartbeat of the system. We provide a launcher script that handles the entire swarm (Creator, Resolver, Agents, Buyback) on Monad Mainnet.

# Make the launcher executable
chmod +x scripts/start_mainnet.sh

# Launch the Swarm on Monad Mainnet
./scripts/start_mainnet.sh

This will:

1. Start the Market Creator (creating BTC/ETH/SOL rounds).
2. Start the Agent Swarm (analyzing and betting).
3. Start the Resolver (settling bets via Pyth).
4. Launch the Frontend on http://localhost:3000.

4. Enable Token Buybacks

The launcher script automatically starts the buyback engine. Manual run:

npx hardhat run scripts/agentBuyback.ts --network monadMainnet

5. Manual Frontend Launch (Optional)

cd web
npm run dev

📜 Contract Details

• Network: Monad Mainnet (Chain ID: 143 via RPC)
• Contract: 0xbdc4a80e6C197aD259194F197B25c8edD519434C
• Oracle: Pyth Network

---

Submitted for the Moltiverse Hackathon 2026 🦞