Meteorologist / Data Scientist: Daily High-Temp Forecasting Model for Prediction Markets

PROJECT OVERVIEW We are a quantitative trading and market-making firm seeking an experienced meteorologist or atmospheric data scientist to help optimize and build the prediction engine behind an automated market-making system. The engine quotes daily high-temperature event contracts across multiple major U.S. cities (starting with Los Angeles/KLAX and expanding to 19 additional cities). The goal of this project is to take raw meteorological model data and turn it into highly accurate, localized, and calibrated temperature probability density curves to dynamically adjust bid/ask spreads. KEY RESPONSIBILITIES - Automated Bias Correction & MOS: Develop and implement statistical bias correction (Model Output Statistics) on historical GRIB2 grid cells versus National Weather Service (NWS) actuals. - Ensemble Ingest & Modeling: Design logic to ingest, parse, and weight live ensemble weather models (e.g., NOAA/GFS, ECMWF, and regional high-resolution models like HRRR). - Probability Density Estimation: Use ensemble standard deviations and historic performance metrics to build a continuous probability density curve for daily peak high temperatures at specific airport stations. - API Data Blending: Assist in structural logic for blending real-time METAR feed sequences, surface wind vectors, and local station data (e.g., Weather Underground PWS network) to catch intraday advection changes or sea breeze shifts. REQUIRED EXPERIENCE - Degree in Meteorology, Atmospheric Science, Climate Data Analytics, or Data Science with a heavy weather focus. - Proven experience working with raw gridded meteorological datasets (GRIB2, NetCDF) and NOAA/ECMWF AWS buckets. - Deep understanding of local microclimates (especially coastal airport advection patterns like KLAX). - Familiarity with Python data science ecosystems (Xarray, NumPy, Pandas, Scipy) or machine learning frameworks (PyTorch) applied to weather features. - Bonus: Previous experience with weather derivatives, event contracts (ForecastEx, Kalshi), or algorithmic trading environments. PROJECT TYPE & BUDGET Hourly / deliverable-based. $100-$200/hr depending on experience. Rate negotiable upward for an exceptional candidate; performance-linked upside available for someone who demonstrably improves forecast calibration. SCREENING QUESTIONS (please address in your proposal) 1. Describe your experience with statistical downscaling or Model Output Statistics (MOS) for specific station-level temperature forecasting. 2. Which weather ensemble models (e.g., GEFS, ECMWF EPS) do you have the most experience programmatically parsing, and what tools did you use? 3. How would you handle an abrupt intraday wind shift or advection change when calculating the probability of a specific 1-degree temperature bin hitting its max before the contract close?