Online Learning-Based Optimal Bidding Approach for FTR Market Participants

We consider the problem of optimal bidding and portfolio optimization for bidders in the financial transmission rights (FTR) auction market. Based on the price-taker assumption, each FTR market participant aims to maximize the profit, which is the difference between the clearing price and FTR revenue. However, both clearing price and the FTR revenue are random and unknown. An online learning methodology is proposed to learn optimal bidding by updating its policy with the newest observations of clearing results. With bidding prices derived by the online learning algorithm, a budget-constrained portfolio optimization problem is solved to distribute the budget among profitable FTRs. Compared to other state-of-the-art online learning approaches, the proposed tree-based bid searching (TBS) algorithm converges faster to the optimal bidding price and has favourable linearithmic time complexity.