Electricity Price Prediction Method with Multi- Time-Segmented Models Based on Load Factor

In the context of spot trading in the electricity market, accurate electricity price forecasting is crucial for market participants to formulate optimal trading strategies. Electricity price fluctuations are driven by supply and demand dynamics, and price behaviors exhibit distinct volatility characteristics across different load segments. This paper proposes a load factor-based multi-model forecasting approach that segments the day into specific time periods to enhance the accuracy and reliability of day-ahead multi-step price predictions. Firstly, the relationship between day-ahead electricity prices and bidding space is analyzed from the perspective of power producers' bidding strategies, anchored in the electricity price formation mechanism. Secondly, leveraging price autocorrelation analysis and load factor indicators, various time segmentation methods are proposed and applied to the dataset. Finally, using actual data from Guangdong province, the proposed method is compared with traditional time series forecasting methods under different time segmentation strategies. The study demonstrates that the load-segmented multi-model forecasting approach significantly improves the accuracy and reliability of electricity price predictions while providing robust economic interpretability.