AI Trader for Prediction Markets

An AI-powered trading agent for prediction markets that leverages LLMs to create and execute trading strategies based on social network signals and on-chain analysis.

Features

Multi-Platform Support: Trade on Polymarket and Kalshi prediction markets
Social Signal Analysis: Track KOL mentions, sentiment, and trending tokens
LLM-Powered Strategies: Uses Google Gemini 3.0 Flash for intelligent analysis
UnifAI Integration: Dynamic tool discovery and agent-to-agent communication
Web Interface: Simple chat-based frontend for trading queries
Moltbot Skills: Packaged as reusable skills for AI agents

Quick Start

Prerequisites

Python 3.10+
UnifAI API key (for social signals and Polymarket)
Google API key (for Gemini LLM)

Installation

# Clone the repository
git clone https://github.com/zbruceli/trading.git
cd trading
 
# Create virtual environment
python -m venv venv
source venv/bin/activate
 
# Install dependencies
pip install -r requirements.txt

Environment Variables

export UNIFAI_AGENT_API_KEY="your-unifai-key"
export GOOGLE_API_KEY="your-google-key"
export LLM_MODEL="gemini/gemini-3-flash-preview"

Running

# Run the trading agent demo
python -m src.agents.trading_agent --demo
 
# Interactive mode
python -m src.agents.trading_agent
 
# Start web interface
uvicorn src.api.server:app --port 8080

Usage

Trading Agent

from src.agents import TradingAgent
 
agent = TradingAgent()
 
# Analyze a token with price + social + news signals
analysis = await agent.analyze_token("SOL")
 
# Get trending tokens from KOL discussions
trending = await agent.get_trending_signals()
 
# Natural language queries
response = await agent.chat("Get ETH price and recent news")

Kalshi Markets

from src.markets import KalshiClient
 
client = KalshiClient()
 
# Get Fed interest rate markets
fed_markets = await client.get_fed_markets(limit=10)
 
# Search markets
results = await client.search_markets("bitcoin", limit=5)

from src.signals import SocialSignalProcessor
 
processor = SocialSignalProcessor()
 
# Get token sentiment
sentiment = await processor.get_token_sentiment("ETH")
 
# Get trending tokens from KOLs
trending = await processor.get_trending_tokens(time_window="24h")

Prediction Market Integrations

Platform	Integration	Market Types
Polymarket	UnifAI SDK	Crypto, politics, sports
Kalshi	Direct API	Economics, politics, events

Project Structure

trading/
├── src/
│   ├── agents/        # Trading agents
│   ├── api/           # Web API & frontend
│   ├── markets/       # Market clients (Kalshi, Polymarket)
│   ├── signals/       # Social signal processors
│   └── strategies/    # Trading strategies
├── skills/            # Moltbot skill definitions
└── tests/

Moltbot Skills

Pre-packaged skills for AI agent platforms:

prediction-trader - Cross-platform trading assistant
kalshi-trader - Kalshi market queries
polymarket-trader - Polymarket integration
social-signals - Social signal analysis

See CLAUDE.md for detailed skill documentation.

Technology Stack

LLM: Google Gemini 3.0 Flash (via LiteLLM)
Agent Framework: UnifAI SDK
Skills Platform: Moltbot (AgentSkills-compatible)
Language: Python 3.10+

License

MIT

UnifAI Trading Suite