A Data-Driven Convergence Bidding Strategy Based on Reverse Engineering of Market Participants’ Performance: A Case of California ISO

Convergence bidding, a.k.a., virtual has been widely adopted in wholesale electricity markets recent years. It provides opportunities for market participants to arbitrage on the difference between day-ahead locational marginal prices and real-time prices. Given fact that convergence bids (CBs) have a significant impact operation of markets, it is important understand how strategically select their CBs real-world. We address this open problem with focus operated by California ISO. In regard, we use publicly available data learn, characterize, evaluate different types bidding strategies are currently used participants. Our analysis includes developing data-driven reverse engineering method apply three years real-world data. involves feature selection density-based clustering. results identifying main clusters CB ISO market. Different characteristics performance each cluster analyzed. Interestingly, unmask common strategy does not match any existing strategic methods literature. Next, build upon lessons learned from propose new can significantly outperform them. bidding. The steps: net profit maximization capturing price spikes, dynamic node labeling, algorithm. show through case studies annual most lucrative increase over 40% if proposed used.