This document presents a semi-parametric time series model using autocopulas. It models the time series of the product of daily natural gas price deviations and temperature deviations from normal levels. The model uses a non-parametric empirical autocopula to capture dependence between time points, combined with a parametric normal inverse Gaussian distribution for the time-varying marginals to account for seasonality in the data. The autocopula allows modeling of tail dependence observed in the data. The seasonal pattern in the data is captured by making the parameter δ of the NIG distribution time-dependent.