In this talk, we discuss a new approach to demand forecasting in supply chains. Demand forecasting is an inevitable task in supply chain management. Due to the endogenous and exogenous factors impact a supply chain, the regime of the supply chain may vary significantly. Such changes in the regime can bring a high volatility to demand time series and consequently, a single statistical model may not suffice to forecast the demand with a desirable level of precision. We develop a nexus between stochastic processes and statistical models to forecast the demand in supply chains with regime switching. The preliminary results on real world time series data sets are promising.