Quantifying Weather-Driven Price Dynamics in Sri Lankan Tea Auctions
ORIGINAL / Forecasting Weather-Driven Price Dynamics Across Sri Lankan Tea Market Catalogues
This study constructs a novel dataset combining weather data and tea auction prices, and uses Granger causality and machine learning models to reveal differential sensitivity of tea catalogues to weather, offering a more precise forecasting framework.
01 ABSTRACT
Based on 105 weekly broker reports from late 2023 to 2026 and regional weather data, the study analyzes price dynamics of four main tea catalogues. Results show market dynamics are primary, but weather matters: Low Grown tea is sensitive to precipitation and sunshine, while Off-Grade and Dust respond to temperature. Catalogue-specific modeling outperforms unified, with LightGBM best for three catalogues.
02 KEY FINDINGS
- Novel dataset combining 105 weekly broker reports and weather data
- Low Grown tea significantly sensitive to precipitation and sunshine at 1-3 week lags (p<0.05)
- Off-Grade and Dust show significant responses to temperature variations
- Catalogue-specific modeling outperforms unified; LightGBM best for three out of four
AI GENERATED SUMMARY / DISCOVERED BY ARXIV CS.LG · ARXIV Q-FIN