🔮 MetaPredict

The Prediction Market for Prediction Markets

Built for Chainlink Convergence Hackathon 2026

📹 Demo Video: [Coming Soon]

🎯 Overview

MetaPredict is a meta prediction market — users create and bet on prediction markets about prediction market platforms themselves. Questions like "Will Polymarket launch their own token?", "Will Kalshi exceed $500M monthly volume?", or "Which prediction market will launch a token first?" are settled automatically using Chainlink CRE workflows that fetch real data from Polymarket & Kalshi APIs, then use AI (Google Gemini) to determine outcomes.

Why it's unique: While prediction markets let you bet on real-world events, MetaPredict lets you bet on the prediction market industry itself — creating a recursive, meta layer on top of the ecosystem. Markets are settled with verifiable, AI-powered analysis of real platform data, not manual resolution.

Use cases:

Token launches & airdrops from prediction market platforms
Platform volume & growth milestones
Feature launches, partnerships, and regulatory outcomes
Cross-platform comparisons (Polymarket vs Kalshi vs Azuro)

🏗️ Architecture

                          ┌─────────────────────────────────────────────────────┐
                          │              Chainlink CRE Workflow                 │
                          │                                                     │
                          │  ┌──────────┐  ┌──────────┐  ┌──────────────────┐  │
                          │  │ Polymarket│  │  Kalshi   │  │ OpenRouter       │  │
                          │  │ Gamma API │  │   API     │  │ (Gemini 2.0     │  │
                          │  └────┬─────┘  └────┬──────┘  │  Flash + Search) │  │
                          │       │              │         └────────┬─────────┘  │
                          │       └──────┬───────┘                 │            │
                          │              ▼                         │            │
                          │     ┌────────────────┐                 │            │
                          │     │ Data Aggregation├────────────────┘            │
                          │     └───────┬────────┘                             │
                          │             │  AI determines winning outcome        │
                          └─────────────┼──────────────────────────────────────┘
                                        │
    ┌──────────┐    ┌──────────────┐    │    ┌───────────────────────────────┐
    │          │    │   Next.js    │    ▼    │     MetaPredictionMarket      │
    │  Users   ├───►│  Frontend    ├────────►│     Smart Contract (Base)     │
    │          │    │  (wagmi v2)  │◄────────│     USDC-based betting        │
    └──────────┘    └──────────────┘         └───────────────────────────────┘

Flow

Create — A user creates a prediction market with a question, deadline, and resolution source
Predict — Users bet USDC on their predicted outcome (Yes/No or multi-choice)
Trigger — After the deadline, anyone calls requestSettlement(), emitting a SettlementRequested event
Settle — Chainlink CRE picks up the event, fetches real data from Polymarket & Kalshi APIs, sends it to Google Gemini for AI analysis, and writes the result onchain
Claim — Winners claim their proportional share of the USDC pool (minus 2% protocol fee)

🔗 Chainlink CRE Integration

This is the core of MetaPredict. Chainlink CRE orchestrates the entire settlement pipeline — from event detection to AI analysis to onchain finalization.

Workflow Files

Main workflow: settlement-workflow/main.ts ⭐
Staging config: settlement-workflow/config.staging.json
Production config: settlement-workflow/config.production.json

CRE Workflow Steps

SettlementRequested event
        │
        ▼
┌─ EVM Log Trigger ──────────────────────────────┐
│  Listens for SettlementRequested(uint256)       │
│  on MetaPredictionMarket contract               │
└────────────────────┬───────────────────────────┘
                     ▼
┌─ EVM Read ─────────────────────────────────────┐
│  Reads market data from contract:               │
│  • question, options, resolutionSource          │
│  Routes to appropriate data sources             │
└────────────────────┬───────────────────────────┘
                     ▼
┌─ HTTP Fetch (conditional) ─────────────────────┐
│  Based on resolutionSource routing:             │
│  • Polymarket Gamma API → events & markets      │
│  • Kalshi API → markets & trades                │
│  • ai_search → skip API, use web search only    │
└────────────────────┬───────────────────────────┘
                     ▼
┌─ AI Analysis (OpenRouter + Gemini 2.0 Flash) ──┐
│  Sends market question + fetched data to AI     │
│  • ai_search mode: Gemini with web search       │
│    grounding for news/announcement questions     │
│  • data mode: Gemini analyzes API data to        │
│    determine if threshold/condition is met       │
│  Returns: winning outcome index (0-based)       │
└────────────────────┬───────────────────────────┘
                     ▼
┌─ Report & EVM Write ──────────────────────────┐
│  Encodes (marketId, winningOutcome) as report   │
│  Signed report delivered to contract via        │
│  Chainlink forwarder → onReport() callback      │
│  Market is settled onchain                      │
└────────────────────────────────────────────────┘

Key CRE Features Used

EVM Log Trigger — Automatically starts workflow when SettlementRequested is emitted
EVM Read — Reads market state directly from the contract
HTTP Compute — Fetches external API data with consensus aggregation
Custom Compute — AI-powered resolution logic with conditional data source routing
EVM Write — Settles the market onchain through a signed, verified report

🧠 Smart Resolution Source Routing

MetaPredict uses a dynamic routing system that selects the right data sources based on the question type. Each market has a resolutionSource field that determines how the CRE workflow fetches and analyzes data:

Resolution Source	Data Fetched	Best For
`ai_search`	AI web search (Gemini grounding)	News, announcements, token launches
`polymarket_volume`	Polymarket events & markets API	Volume-based questions
`polymarket_traders`	Polymarket trader & market data	User growth questions
`kalshi_markets`	Kalshi markets & trades API	Kalshi-specific questions
`polymarket_kalshi_data`	Both Polymarket + Kalshi APIs	Cross-platform comparisons

This makes the system modular — adding new data sources (e.g., Azuro, Drift) requires only adding a new route in the CRE workflow, not modifying the contract.

📊 Data Sources

Source	API	Data
Polymarket	Gamma API (`/events`, `/markets`)	Top events by volume, market prices, liquidity, 24h volume
Kalshi	REST API (`/markets`, `/trades`)	Open markets, recent trades, open interest, volume
OpenRouter	Chat Completions API	Google Gemini 2.0 Flash with web search grounding

🖥️ Frontend

Next.js 16 + wagmi v2 + RainbowKit for wallet connection
Dark cyberpunk UI with glass morphism, glow effects, gradient accents
Features:
- Browse & filter markets (active, settled, pending)
- Create binary or multi-choice markets with resolution source selection
- Place USDC predictions with automatic approval flow
- Settlement status indicator with visual timeline
- Claim winnings after settlement
- Responsive design

📝 Smart Contracts

Contract: contracts/evm/src/MetaPredictionMarket.sol

Deployed Addresses

Network	Contract	USDC
Base Sepolia	`0xB5d15e699606feF7b93b29C04c489DCe16236673`	`0x036CbD53842c5426634e7929541eC2318f3dCF7e`
Tenderly Virtual TestNet	`0x3CeE04482438d2474e364aAC75DA54156535be8a`	`0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913`

Key Functions

Function	Description
`createBinaryMarket(question, deadline, resolutionSource)`	Create a Yes/No market
`createMultiChoiceMarket(question, options, deadline, resolutionSource)`	Create a multi-choice market (2–6 options)
`predict(marketId, outcome, amount)`	Bet USDC on an outcome
`requestSettlement(marketId)`	Trigger CRE settlement (after deadline)
`onReport(metadata, report)`	CRE callback — settles market onchain
`claimWinnings(marketId)`	Claim proportional USDC payout

USDC-based — all bets and payouts in USDC (6 decimals)
2% protocol fee on settled market pools
Chainlink CRE forwarder — only the authorized forwarder can call onReport()

🧪 Tenderly Virtual TestNet

Why Tenderly?

Testing CRE workflows against real Base mainnet state — real USDC contract, real protocol data — without spending real funds. This validates that our settlement workflow works correctly with production data before mainnet deployment.

How It Solves Problems

Virtual TestNets eliminate the gap between testnet and mainnet behavior. Our CRE workflow fetches real Polymarket & Kalshi data and settles markets on a fork of Base mainnet, proving the system works with real-world state. The USDC contract at 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913 is the actual Base mainnet USDC — balances are set via tenderly_setErc20Balance for testing.

Deployment Proof

Explorer →

Action	Transaction Hash
Deploy Contract	`0xdcc77a2072c8d91bd9cd96875e17021128540677b61cf95bf81cf24fb559eabe`
Create Market 0 (`ai_search`)	`0xaf37106653e1d66aa60cbc2c03e4b726c6ad5ddca58264a0a7a6420299915ffd`
Create Market 1 (`polymarket_volume`)	`0x8afdc04ab2fc893b1f6711b671129f935d5f468306e1b68f6cf5bed6427cad5f`
Predict YES on Market 0 (1 USDC)	`0x521af484201707d7ce49bd454cd253416a1addf4fd7aeff4b2a46e1678960902`
Predict NO on Market 1 (2 USDC)	`0xcea75feb454b0c7c0f1787504ffb8c133f77339e79653f321c705e90f0298ad9`

Automated Deployment

# Deploy to Tenderly Virtual TestNet with one command
./scripts/deploy-tenderly.sh

The script mints USDC, deploys the contract, creates sample markets, and places test predictions — all automated.

🚀 Getting Started

Prerequisites

Foundry (forge, cast)
Bun (for CRE workflow)
Node.js 18+ (for frontend)
Chainlink CRE CLI

Setup

# Clone
git clone https://github.com/ArgonStark/MetaPredict.git
cd MetaPredict

# Smart Contracts
cd contracts/evm
forge build

# CRE Workflow
cd ../../settlement-workflow
bun install

# Simulate CRE workflow
cre workflow simulate settlement-workflow --target staging-settings

# Frontend
cd ../frontend
npm install
npm run dev

Environment Variables

Create a .env file in the project root:

CRE_ETH_PRIVATE_KEY=0x...       # Deployer private key
GEMINI_API_KEY_VAR=...           # Google Gemini API key (via OpenRouter)
OPENROUTER_API_KEY_VAR=...       # OpenRouter API key

📁 Project Structure

MetaPredict/
├── contracts/evm/                # Solidity smart contracts (Foundry)
│   ├── src/
│   │   ├── MetaPredictionMarket.sol   # Main contract
│   │   └── interfaces/IReceiver.sol   # CRE receiver interface
│   └── foundry.toml
├── settlement-workflow/          # Chainlink CRE workflow
│   ├── main.ts                   # Workflow logic (⭐ MAIN CRE FILE)
│   ├── config.staging.json       # CRE staging config
│   └── config.production.json    # CRE production config
├── frontend/                     # Next.js frontend
│   └── src/
│       ├── app/                  # Pages (home, markets, market detail, create)
│       ├── components/           # UI components
│       ├── hooks/                # wagmi hooks (useMarkets, useMarket, usePrediction)
│       ├── config/               # Contract addresses, ABI, chain config
│       └── providers/            # Web3Provider (wagmi + RainbowKit)
├── scripts/
│   └── deploy-tenderly.sh        # Automated Tenderly deployment
└── README.md

🔑 Key Files for Chainlink CRE

File	Description
`settlement-workflow/main.ts`	CRE workflow — EVM triggers, HTTP fetches, AI integration, EVM write
`settlement-workflow/config.staging.json`	CRE workflow configuration (Base Sepolia)
`settlement-workflow/config.production.json`	CRE production configuration
`contracts/evm/src/MetaPredictionMarket.sol`	Smart contract with `onReport()` CRE forwarder integration

🏆 Hackathon Tracks

Track	How MetaPredict Fits
Prediction Markets	AI-powered prediction market with automated settlement via real platform data
CRE & AI	Full CRE workflow: EVM trigger → HTTP fetch → AI analysis (Gemini) → EVM write
Tenderly Virtual TestNets	Deployed & tested on Virtual TestNet against real Base mainnet state

👥 Team

Built by ArgonStark

📄 License

MIT

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
contracts/evm		contracts/evm
frontend		frontend
scripts		scripts
settlement-workflow		settlement-workflow
.gitignore		.gitignore
README.md		README.md
project.yaml		project.yaml
secrets.yaml		secrets.yaml

Navigation Menu

Repository files navigation

🔮 MetaPredict

🎯 Overview

🏗️ Architecture

Flow

🔗 Chainlink CRE Integration

Workflow Files

CRE Workflow Steps

Key CRE Features Used

🧠 Smart Resolution Source Routing

📊 Data Sources

🖥️ Frontend

📝 Smart Contracts

Deployed Addresses

Key Functions

🧪 Tenderly Virtual TestNet

Why Tenderly?

How It Solves Problems

Deployment Proof

Automated Deployment

🚀 Getting Started

Prerequisites

Setup

Environment Variables

📁 Project Structure

🔑 Key Files for Chainlink CRE

🏆 Hackathon Tracks

👥 Team

📄 License

About

Releases

Packages

Contributors

Languages

Folders and files

Latest commit

History

Repository files navigation

🔮 MetaPredict

🎯 Overview

🏗️ Architecture

Flow

🔗 Chainlink CRE Integration

Workflow Files

CRE Workflow Steps

Key CRE Features Used

🧠 Smart Resolution Source Routing

📊 Data Sources

🖥️ Frontend

📝 Smart Contracts

Deployed Addresses

Key Functions

🧪 Tenderly Virtual TestNet

Why Tenderly?

How It Solves Problems

Deployment Proof

Automated Deployment

🚀 Getting Started

Prerequisites

Setup

Environment Variables

📁 Project Structure

🔑 Key Files for Chainlink CRE

🏆 Hackathon Tracks

👥 Team

📄 License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages