Battery Bidding under Price Uncertainty in Wholesale Electricity Markets

Battle Count: 68

While focused on electricity markets rather than traditional financial markets, the paper shares deep methodological connections with quantitative trading: mean-CVaR optimization (analogous to portfolio risk management), scenario-based stochastic programming, bid curve construction (analogous to order book strategies), and the interplay between uncertainty and risk-adjusted decision-making. The LP reformulation technique and dual variable interpretation are directly transferable to algorithmic execution and market-making contexts. The withholding behavior analysis parallels strategic order placement in limit order books.

The LP reformulation is computationally tractable (solves in seconds vs. hours for MILP). Implementation requires: (1) scenario generation via Gaussian Process with exponential kernel, (2) construction of precomputed clearing indicator matrices, (3) LP formulation with standard constraints. The main complexity lies in correctly constructing the clearing pattern matrices and handling the bid price discretization. Standard LP solvers (Gurobi, CPLEX) suffice. The practical LP variant further simplifies by fixing hourly modes (charge/discharge/idle).

The paper provides detailed mathematical formulations, algorithm descriptions, and parameter specifications (battery configuration, scenario generation via Gaussian Process with exponential kernel, CAISO node SLATE_7_N004 data). However, no code repository is mentioned. CAISO DAM price data is publicly available. The LP reformulation is fully specified and implementable with standard solvers.