Neural and Time-Series Approaches for Pricing Weather Derivatives: Performance and Regime Adaptation Using Satellite Data
Marco Hening Tallarico, Pablo Olivares
TL;DR
The paper addresses weather-derivative pricing under regime heterogeneity by leveraging NASA POWER satellite data (1981–2023) for Toronto and Chicago. It jointly assesses a temperature model based on harmonic regression with ARMA residuals and a neural-network forecast, and a precipitation model built on a compound Poisson–Gamma process estimated via maximum likelihood and a CNN trained on synthetic rainfall sequences to capture regime shifts; December pricing is used to illustrate regime-adaptive improvements. Key contributions include region-wide satellite data usage, regime-aware parameter estimation for precipitation via a CNN, and closed-form strangle pricing enabled by the Gamma-sum property, yielding valuations that differ meaningfully from Historic Burn Approach and time-series benchmarks. The results suggest that regime-adaptive ML approaches can improve WD pricing accuracy in practice, while highlighting the trade-offs between estimator precision and the ability to capture nonlinear, regime-dependent climate dynamics; the NASA POWER data prove to be a practical substitute for sparse station data in this financial-forecasting context.
Abstract
This paper studies pricing of weather-derivative (WD) contracts on temperature and precipitation. For temperature-linked strangles in Toronto and Chicago, we benchmark a harmonic-regression/ARMA model against a feed-forward neural network (NN), finding that the NN reduces out-of-sample mean-squared error (MSE) and materially shifts December fair values relative to both the time-series model and the industry-standard Historic Burn Approach (HBA). For precipitation, we employ a compound Poisson--Gamma framework: shape and scale parameters are estimated via maximum likelihood estimation (MLE) and via a convolutional neural network (CNN) trained on 30-day rainfall sequences spanning multiple seasons. The CNN adaptively learns season-specific $(α,β)$ mappings, thereby capturing heterogeneity across regimes that static i.i.d.\ fits miss. At valuation, we assume days are i.i.d.\ $Γ(\hatα,\hatβ)$ within each regime and apply a mean-count approximation (replacing the Poisson count by its mean ($n\hatλ$) to derive closed-form strangle prices. Exploratory analysis of 1981--2023 NASA POWER data confirms pronounced seasonal heterogeneity in $(α,β)$ between summer and winter, demonstrating that static global fits are inadequate. Back-testing on Toronto and Chicago grids shows that our regime-adaptive CNN yields competitive valuations and underscores how model choice can shift strangle prices. Payoffs are evaluated analytically when possible and by simulation elsewhere, enabling a like-for-like comparison of forecasting and valuation methods.