polymarket-correlation

What This Skill Does

The polymarket-correlation skill acts as a sophisticated arbitrage engine for prediction markets. It leverages the OpenClaw agent architecture to monitor, compare, and analyze pairs of Polymarket markets in real-time. By identifying discrepancies in the pricing of related events, the skill detects when one market's probability fails to align with historical correlations or thematic associations relative to another market.

It performs this by calculating expected outcomes against actual market prices. For instance, if historical data suggests that a Fed rate cut has a 70% correlation with an S&P 500 rally, but the current market prices imply only a 35% chance of a rally following a rate cut, the skill flags this as a potential mispricing. It provides actionable signals (BUY_YES, BUY_NO, or HOLD) along with confidence scores, helping AI agents execute trades based on data-driven statistical advantages.

Installation

To integrate this skill into your agent, use the OpenClaw package manager. Open your terminal and run:

clawhub install openclaw/skills/skills/sbaker5/polyedge

Ensure you have configured your environment variables if you intend to utilize the API endpoint features. The installation automatically pulls the necessary dependencies, including the Polymarket API client and the correlation pattern logic.

Use Cases

• Cross-Market Arbitrage: Automatically identifying when two interdependent event markets are priced inconsistently.
• Hedge Optimization: Finding market pairs where a hedge position is currently underpriced due to market sentiment bias.
• Trend Analysis: Detecting when geopolitical or economic market movement is lagging behind the realized probability of related future events.
• Automated Research: Allowing agents to cross-reference news sentiment against market probability shifts.

Example Prompts

1. "Check the correlation between the market 'will-fed-cut-rates-in-july' and 'sp500-end-of-year-performance' and tell me if there is a buy signal."
2. "Analyze the current pricing discrepancy between the Ukraine ceasefire market and the Taiwan invasion market and report the confidence level."
3. "Is there an arbitrage opportunity between these two markets: russia-ukraine-ceasefire-before-gta-vi-554 and will-china-invades-taiwan-before-gta-vi-716?"

Tips & Limitations

• Historical Patterns: The engine is only as good as the patterns in src/patterns.py. You can significantly improve accuracy by adding specific conditional probability data for your target domains.
• Thresholds: Keep in mind that 'low' confidence signals (12% threshold) carry higher risk. For automated trading, prioritize 'high' confidence signals (5% threshold).
• Limitations: This tool does not factor in liquidity, slippage, or transaction fees. These costs can erode the thin margins found in arbitrage. Always perform manual due diligence before executing large-scale positions based on automated signals.