weather-arbitrage_skill

Overview

This skill is a weather prediction-market arbitrage assistant that exploits the information gap between federal-grade forecasts (NOAA/ECMWF) and retail market pricing. It offers two modes: a recommended NOAA arbitrage mode that requires no forecasting, and a multi-source temperature-prediction mode for higher-risk opportunities. The system is optimized for fast scans, continuous monitoring, backtesting, and small-stake, high-frequency execution.

How this skill works

The skill compares authoritative numerical weather model outputs (NOAA, ECMWF, GFS, ICON, GEM) with market prices on weather prediction platforms to find mispriced contracts. In NOAA arbitrage mode it flags opportunities where federal forecast confidence strongly disagrees with market odds; in temperature-prediction mode it computes a weighted ensemble forecast and estimates edge vs market odds. Commands support single scans, continuous watches, city-level analysis, simulation, and backtests.

When to use it

• When markets show large gaps between NOAA/federal forecasts and retail odds
• For short-term 24–48 hour events where model confidence is high
• When you want low-complexity, low-risk arbitrage rather than directional forecasting
• To run automated scans or a continuous watch for fast entry/exit signals
• When backtesting strategy performance on historical data

Best practices

• Prioritize NOAA arbitrage mode for lower risk and higher hit rate
• Use confidence thresholds: NOAA confidence >85% before acting
• Limit single-bet exposure (recommend ≤ $2 for NOAA mode; follow mode rules for prediction mode)
• Run continuous monitoring every 1–3 minutes for time-sensitive markets
• Always validate signals with backtests and simulate before increasing stake

Example use cases

• Scan Polymarket and similar platforms to buy low-priced contracts when NOAA implies near-certain outcomes
• Run a 2-minute watch process to capture fleeting arbitrage windows around updated federal forecasts
• Analyze a city (e.g., Chicago) with multi-model ensemble to identify temperature edges above 20%
• Backtest the temperature-prediction ladder strategy across 2024 historical data to validate ROI
• Simulate single trades to confirm execution rules and position sizing before real funds