prediction-market-arbitrage_skill

Overview

This skill orchestrates three installed tools—topic-monitor, polymarket-odds, and simmer-weather—into a repeatable arbitrage detection workflow for prediction markets. It watches for high-confidence breaking signals about an event, looks up live Polymarket probabilities, computes a confidence-vs-price delta, and issues alert-first recommendations. The skill frames decisions, enforces freshness and normalization rules, and can produce a guarded execution plan without auto-trading.

How this skill works

The meta-skill drives a heartbeat: topic-monitor supplies scored news signals, polymarket-odds returns normalized market yes-probabilities, and simmer-weather provides the execution pattern reference. It normalizes all numeric values to decimals, rejects stale signals/market snapshots, computes delta = news_confidence - market_probability, and applies threshold rules (min_news_confidence and min_delta). On trigger it emits a structured alert record and, if requested, a stepwise execution-plan that remains dry-run until explicit user consent.

When to use it

• Monitoring potential news-vs-market mispricing for specific event hypotheses (e.g., CEO resignation).
• When you need a clear LM playbook combining signal discovery and live market checks.
• Before deciding manual or automated trades to ensure signal freshness and safety.
• When you require an alert-first workflow with optional guarded execution guidance.
• When you must enforce normalization and staleness rules across tools.

Best practices

• Install and verify topic-monitor, polymarket-odds, and simmer-weather before running; abort on missing commands.
• Configure topic-monitor topic_id and monitoring cadence once; run external heartbeat every 5 minutes.
• Always convert percentages to decimals before comparison and store timestamps in ISO format.
• Reject signals older than 30 minutes or market snapshots older than 5 minutes to avoid stale decisions.
• Use execution_mode=alert-only by default; require explicit opt-in for any live trading step.

Example use cases

• Detect a high-confidence rumor that a CEO will resign, compare to Polymarket price, and alert if the market lags.
• Run continuous monitoring for an M&A rumor set up via topic-monitor and notify traders with a delta-based recommendation.
• Provide a compliance-safe execution-plan showing dry-run steps via simmer-weather when delta threshold is met.
• Escalate conflicting or low-confidence signals for secondary verification before issuing any trade suggestion.