Catastrophe & Extreme Risk Trader

This is a template. The default signal is keyword-based market discovery combined with conviction-based sizing and catastrophe_bias() — two structural edges that work without any external API. The skill handles all the plumbing (market discovery, trade execution, safeguards). Your agent provides the alpha.

Strategy Overview

Catastrophe prediction markets are uniquely mispriced because retail traders anchor on the most recent vivid disaster rather than historical base rates. The availability heuristic is the dominant pricing force: the first named storm of the season spikes subsequent storm markets 20–40%, even when NOAA's seasonal forecast hasn't changed by a single storm. After a major wildfire, every "will X state break a fire record?" market overshoots. After a quiet start to a season, markets underprice the base rate. Two structural edges compound:

1. Availability bias correction — NOAA, NHC, NIFC, and USGS publish decades of calibrated base rates. Named Atlantic storm counts have 40+ years of forecasting data. Global temperature records are measured to ±0.01°C simultaneously by three independent agencies. The edge is in knowing these numbers when retail is trading on vibes and recent memory.

2. Seasonal data quality timing — The signal is only actionable when models are actively running. During hurricane peak season (Aug–Oct), NHC issues advisories every 6 hours and model ensembles update in real time. A named-storm-count market in February is priced on stale pre-season data; the same market in September is priced against daily NHC output. The edge doubles when real-time data is flowing.

Signal Logic

Default Signal: Conviction-Based Sizing with Catastrophe Bias

1. Discover active catastrophe and extreme weather markets on Polymarket
2. Compute base conviction from distance to threshold (0% at boundary → 100% at p=0/p=1)
3. Apply catastrophe_bias() — hazard type data quality × seasonal calendar timing
4. Size = max(MIN_TRADE, conviction × bias × MAX_POSITION) — capped at MAX_POSITION
5. Skip markets with spread > MAX_SPREAD or fewer than MIN_DAYS to resolution

Catastrophe Bias (built-in, no API required)

Factor 1 — Hazard Type / Data Quality

The Availability Bias Rule — The first major event of a season creates a retail pricing spike that is almost always an overreaction. The NOAA seasonal forecast before and after that first storm is essentially unchanged, but the market price jumps 20–40%. Fading these spikes — or, better, entering before them — is the core mechanism of the named storm edge. The base rate, not the headline, is the signal.

The Earthquake Exemption — Unlike weather hazards, earthquakes have no seasonal signal and no meaningful short-term forecasting capability. USGS can give you a 1-in-500 annual probability for a M7+ event in a specific fault system. They cannot tell you if it will happen in Q3. Trade earthquake markets at maximum caution (0.80x), and tsunami/volcanic markets at the floor (0.75x).

Factor 2 — Seasonal Calendar Timing

Combined Examples

Keywords Monitored

hurricane, tropical storm, category 5, Atlantic season, named storm,
tornado, wildfire, acres burned, fire season, earthquake, magnitude,
tsunami, volcanic eruption, 100-year flood, FEMA, disaster declaration,
billion-dollar disaster, polar vortex, bomb cyclone, derecho, heat dome,
hottest year, warmest year, temperature record, blizzard, ice storm,
above-normal season, NOAA, NHC, Category 4, major hurricane

Remix Signal Ideas

• NOAA National Hurricane Center: Named-storm seasonal forecast gives a directly tradeable probability for storm-count markets — compare NOAA's official probability to Polymarket price; the lag after a quiet start to the season can be 15–25%
• NIFC Wildfire Statistics: Year-to-date acres burned vs 10-year average — when YTD is tracking 40% above average by July, "above-normal fire season" markets are structurally underpriced
• USGS Earthquake Hazards API: Real-time seismic data M2.5+ globally — for post-earthquake aftershock markets, the USGS Omori decay law gives probability estimates of M6+ aftershocks within days of a major event
• Berkeley Earth / NASA GISS: Annual global temperature anomaly updated monthly — when October anomaly is already 0.2°C above the prior record, "will 2026 be hottest year?" is a near-certainty the market underprices

Safety & Execution Mode

The skill defaults to paper trading (venue="sim"). Real trades only with --live flag.

autostart: false and cron: null — nothing runs automatically until you configure it in Simmer UI.

Required Credentials

Tunables (Risk Parameters)

All declared as tunables in clawhub.json and adjustable from the Simmer UI.

Dependency

simmer-sdk by Simmer Markets (SpartanLabsXyz)

• PyPI: https://pypi.org/project/simmer-sdk/
• GitHub: https://github.com/SpartanLabsXyz/simmer-sdk