Modeling electricity markets with hidden Markov model

This paper proposes to model the movements of electricity markets as partially observable Markov processes driven by underlying economic forces. An electricity market is modeled as a dynamic system evolving over time according to Markov processes. At any time interval, the electricity market can be in one state and transition to another state in the next time interval. This paper models the states of an electricity market as partially observable, while each state has incomplete observations such as market-clearing price and quantity. The true market states are hidden from a market participant behind the incomplete observation. The hidden Markov model (HMM) is of a more fundamental approach and focuses on capturing the interaction of supply and demand forces on electricity markets. Such an approach is appropriate because the simultaneous production and consumption of electricity eliminates the storage sector, while limited transmission networks segment electricity markets. This model is shown to be able to link the fundamental drivers to the price behaviors; therefore, it provides forecast power for mid-term and long-term price movements. This work applies HMM to historical data from New York independent system operator (NYISO), and examples are given to illustrate the forecast power of HMM.