Multidimensional Portfolio Optimization in the Turkish Electricity Market: A PGP-MVS Approach

This study examines the optimization of electricity generation portfolios in the Turkish electricity market (EPIAS), which is characterized by high price volatility. Previous studies in the literature generally rely on the ‘Mean-Variance’ (MV) model; however, sudden price jumps and asymmetric return structures in the electricity market do not meet the normal distribution principle, which is the basic assumption of the MV model. Therefore, skewness moment has been included in the MV model, and the Mean-Variance-Skewness (MVS) model, which does not assume normal distribution, has been created. In this study, Mean-Variance-Skewness (MVS) is used with hourly data for the period 2024-2025 to analyze how electricity producers take advantage of asymmetric opportunities in the market, and the optimization process is carried out using the Polynomial Goal Programming (PGP) method. In the application phase, a three-dimensional analysis surface showing the balance between return, risk, and asymmetric opportunities is first created based on 100 different production preference scenarios. Then, optimal electricity generation portfolios are created on an hourly basis according to eight different strategic preference scenarios determined within the MVS-PGP framework. Based on the results of these optimal generation portfolios, it was concluded that electricity generation in the evening hours is indispensable for portfolio stability. On the other hand, it was determined that electricity producers aiming to capture asymmetric profit opportunities should shift their electricity production to midday hours when solar energy is abundant. Finally, these eight strategic portfolios were analyzed based on financial performance metrics. As a result of this empirical evaluation, it was observed that the MVS model produces financially superior and more efficient results compared to the traditional MV model.