Comparing Least Squares Monte Carlo with the Rolling Intrinsic Strategy, for Batteries trading on the Intraday Electricity Market

Abstract

In this thesis, we investigate how a Least Squares Monte Carlo (LSM) trading strategy compares to the rolling intrinsic (RI) strategy, for battery energy storage systems (BESS) trading on the intraday electricity market. We formulate the battery trading optimization problem as a stochastic optimal control problem and formulate each trading strategy within this framework. This work extends earlier works using the LSM method for battery trading optimization, incorporating operational cycle constraints, that enable realistic modeling of battery cycling limitations within the dynamic programming framework.

Using historical intraday price data from Nordpool and EPEX SPOT markets spanning 2023- 2025, our empirical analysis for the German market reveals that the rolling intrinsic strategy generally outperforms the LSM approach across all tested battery configurations and seasonal conditions. The RI strategy’s superior performance is attributed to its ability to exploit price volatility through re-trading of future contracts multiple times as market conditions evolve, capturing substantial extrinsic value despite its deterministic optimization framework. In contrast, the LSM method, which only commits to trades immediately before delivery, cannot capitalize on this re-trading opportunity.

While the rolling intrinsic strategy demonstrates superior performance, the LSM method pro- vides valuable forward-looking continuation values that capture the expected future opportunity value of different battery states. To leverage these insights, we introduce a hybrid trading strategy that enriches the rolling intrinsic approach by incorporating LSM-derived continuation values as a filter for proposed trades, helping to prevent commitments to positions that may be profitable in the short term but disadvantageous in the long term. The proposed hybrid strategy shows modest improvements over the pure rolling intrinsic approach, with performance gains of up to 15% in certain market conditions, demonstrating the value of incorporating forward-looking continuation values into the decision-making process. These improvements are mentioned as potential enhancements to current industry practices, where rolling intrinsic strategies are widely deployed due to their computational efficiency and interpretability.

This thesis was written in collaboration with Vattenfall, who supported this research and provided the data used in the experiments.