Paper: arXiv 2510.03209
Authors: Yiming Zhang, Wolfgang Ridinger, David Wozabal
Abstract
As renewable energy integration increases supply variability, battery energy storage systems (BESS) present a viable solution for balancing supply and demand. This paper proposes a novel approach for optimizing battery BESS participation in multiple electricity markets. We develop a joint bidding strategy that combines participation in the primary frequency reserve market with continuous trading in the intraday market, addressing a gap in the extant literature which typically considers these markets in isolation or simplifies the continuous nature of intraday trading. Our approach utilizes a mixed integer linear programming implementation of the rolling intrinsic algorithm for intraday decisions and state of charge recovery, alongside a learned classifier strategy (LCS) that determines optimal capacity allocation between markets. A comprehensive out-of-sample backtest over more than one year of historical German market data validates our approach: The LCS increases overall profits by over 4% compared to the best-performing static strategy and by more than 3% over a naive dynamic benchmark. Crucially, our method closes the gap to a theoretical perfect foresight strategy to just 4%, demonstrating the effectiveness of dynamic, learning-based allocation in a complex, multi-market environment.
Complexity vs Empirical Score
- Math Complexity: 7.0/10
- Empirical Rigor: 8.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematical modeling, including mixed-integer linear programming (MILP) and a rolling intrinsic algorithm, which represents significant mathematical complexity. It is heavily data-driven, featuring a comprehensive out-of-sample backtest over more than one year of real market data with specific profit improvements and a comparison to a theoretical perfect foresight benchmark, indicating high empirical rigor.
Research Flowchart
flowchart TD A["Research Goal: Optimize BESS Participation in Joint Intraday & Frequency Reserve Markets"] --> B["Methodology: Learning Classifier Strategy + Rolling Intrinsic Algorithm"] B --> C["Data: 1+ Year of Historical German Market Data"] C --> D["Process: Mixed Integer Linear Programming & State of Charge Recovery"] D --> E["Outcomes: Backtested via Out-of-Sample Simulation"] E --> F["Findings: +4% vs Best Static Strategy, +3% vs Dynamic Benchmark, 4% vs Perfect Foresight"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a joint bidding strategy that combines participation in the primary frequency reserve market with continuous trading in the intraday market, using a mixed‑integer linear programming rolling intrinsic algorithm and a learned classifier to allocate capacity.
Source sentence
We develop a joint bidding strategy that combines participation in the primary frequency reserve market with continuous trading in the intraday market
Assumes 5
- Zero‑price bids are assumed in the pay‑as‑cleared FCR auction. market
Source sentence
our model assumes zero-price bids in the pay-as-cleared FCR auction and uses the suboptimal myopic RI for intraday trading.
- The intraday market is modeled as an order‑book based continuous trading environment. market
Source sentence
We propose a mixed-integer linear programming version of the intrinsic policy that can handle orderbook-based continuous trading.
- A small, curated set of FCR strategies is used for classification. model
Source sentence
A primary constraint is the reliance on a small and curated set of FCR strategies.
- Rolling intrinsic algorithm is myopic (suboptimal) for intraday trading. model
Source sentence
uses the suboptimal myopic RI for intraday trading.
Show 1 more
- Training data window of 240 days is sufficient for daily model updates. execution
Source sentence
Each day, a training window of N = 240 days is used.
Data
2023-01 – 2024-09 · German FCR and intraday electricity markets · other · daily
- German intraday and frequency containment reserve market data market-prices unknown
Source sentence
Using detailed historical data from these two markets, our results show that using our algorithm to combine FCR and IDM trading returns up to 4% more revenue than delivering FCR only
Costs
Not modeled
Source sentence
our model assumes zero-price bids in the pay-as-cleared FCR auction and uses the suboptimal myopic RI for intraday trading.
Finds 4
- The learned classifier strategy (LCS) increases overall profits by over 4% compared to the best‑performing static strategy and by more than 3% over a naive dynamic benchmark. performance
Source sentence
The LCS increases overall profits by over 4% compared to the best‑performing static strategy and by more than 3% over a naive dynamic benchmark.
- The joint strategy is only 4% worse than a clairvoyant (perfect‑foresight) strategy. performance
Source sentence
our method closes the gap to a theoretical perfect foresight strategy to just 4%
- Combining FCR and IDM trading returns up to 4% more revenue than delivering FCR only and delivers 96% more than trading in the continuous intraday market only. performance
Source sentence
our results show that using our algorithm to combine FCR and IDM trading returns up to 4% more revenue than delivering FCR only and delivers 96% more than trading in the continuous intraday market only
- The 'Only FCR' strategy was the single best choice on 37.1% of all days, highlighting the dominance of the FCR market. performance
Source sentence
the ’Only FCR’ strategy proved to be the single best choice on a remarkable 37.1% of all days
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 29 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.