Machine Learning-Driven Virtual Bidding With Electricity Market Efficiency Analysis

This paper develops a machine learning-driven portfolio optimization framework for virtual bidding in electricity markets considering both risk constraint and price sensitivity. The algorithmic trading strategy is developed from the perspective of proprietary firm to maximize profit. A recurrent neural network-based Locational Marginal Price (LMP) spread forecast model by leveraging inter-hour dependencies market clearing algorithm. LMP sensitivity with respect net bids modeled as monotonic function proposed constrained gradient boosting tree. We leverage bid evaluate profitability efficiency U.S. wholesale markets. comprehensive empirical analysis on PJM, ISO-NE, CAISO indicates that explicitly outperforms one neglects Sharpe ratio portfolios all three are much higher than S&P 500 index. It was also shown CAISO's two-settlement system lower PJM ISO-NE.