Price estimation by random walk mean reversion in day-ahead electricity market

Abstract

The fluctuations in the Day-ahead electricity market (DAM) prices generally follow the periodic demand variance. However, the observations of historical data also reveal large jumps or spikes that might be defined as the randomness of the price formation. In this study, the Turkish DAM price is treated as a discrete signal to follow a methodology of utilizing the frequency-domain of the signal and a Fourier Transform analysis is employed for the separation of the random electricity price variations from the periodic ones. The volatility of the random electricity price variation is then used for the estimation of DAM price for a future day based on a random walk with a mean reversion method. The promising results are then compared with a multiple linear regression method from the literature to quantify the effectiveness of the proposed model for price forecasting for a Day-ahead electricity market.