pith. machine review for the scientific record. sign in

arxiv: 2605.01178 · v1 · submitted 2026-05-02 · 🧮 math.OC · q-fin.MF

Recognition: unknown

Modeling Stochastic Multi-Agent Interaction in Intraday Battery Energy Storage Dispatch with Market Power

Ruimeng Hu , Mike Ludkovski , Hezhong Zhang
Authors on Pith no claims yet

Pith reviewed 2026-05-09 15:16 UTC · model grok-4.3

classification 🧮 math.OC q-fin.MF
keywords stochastic differential gameNash equilibriumRiccati equationsbattery energy storageintraday dispatchmarket powerlinear-quadratic gamecompetitive interaction
0
0 comments X

The pith

Battery storage operators reach Nash equilibrium in intraday dispatch by solving a system of Riccati equations when prices respond linearly to total charging.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Multiple battery energy storage system operators compete to maximize arbitrage revenue by adjusting their charging rates, with the resulting electricity price depending on the total charging across all operators. The paper models this as a finite-player linear-quadratic stochastic differential game driven by a shared noise process. It derives semi-explicit equilibrium strategies and prices using a system of Riccati equations, covering both cases where operators have different parameters and the simplified case where they are identical. The framework then quantifies competitive effects such as the impact of new entrants, the value of collusion, and the influence of large players. It also examines the limit as the number of operators becomes large.

Core claim

The Nash equilibrium of the finite-player linear-quadratic differential game with a shared stochastic driver is characterized by semi-explicit representations of equilibrium feedback controls and equilibrium prices. These are obtained for both the general heterogeneous BESS setting and the simplified homogeneous setting through a system of Riccati equations. The resulting model then supports analysis of marginal externalities from additional entrants, gains from coordination, market power of large operators, supply effects of hybrid BESS, and the large-population asymptotic regime.

What carries the argument

A system of coupled Riccati equations that solves the finite-player linear-quadratic stochastic differential game and yields the equilibrium charging feedback rules and price process.

Load-bearing premise

The electricity price is a linear function of aggregate charging rates plus one shared stochastic driver, and the overall interaction remains linear-quadratic.

What would settle it

Real-world charging decisions of multiple BESS operators deviate from the Riccati-derived feedback controls when observed prices are approximately linear in total charging activity.

Figures

Figures reproduced from arXiv: 2605.01178 by Hezhong Zhang, Mike Ludkovski, Ruimeng Hu.

Figure 1
Figure 1. Figure 1: Finite-player coupling structure between individual controls view at source ↗
Figure 2
Figure 2. Figure 2: A market with N = 8 homogeneous BESS operators, with parameters in view at source ↗
Figure 3
Figure 3. Figure 3: Left panel: E[ˆαt] and E[Sˆ t]. Middle and Right: Effect of c2 and c3 on average total charging and discharging TC (29) and average range of SOC TS (30), holding c3 = 0 and c2 = 0 respectively. Price sensitivity c1. The price impact c1 affects the sensitivity of the price to the total supply. Higher c1 strengthens both the feedback effect of the individual α i t (which contributes a quadratic cost c1(α i t… view at source ↗
Figure 4
Figure 4. Figure 4: Left panel: effect of price impact c1 on total charge/discharge TC (29) and on Range of SOC. Right: Effect of ρ on average Std( ˆα i t ) and Std(Pˆi t ). We use the base setting of Section 4.1 with N = 8 homogeneous agents. Proposition 4.1. Let {g1(t), g2(t), g3(t), g4(t)} be the functions defined in Theorem 3.3 and set g˜(t) := g2(t) + (N − 1)g3(t). Under the homogeneous-agent setting, the exogenous produ… view at source ↗
Figure 5
Figure 5. Figure 5: Market composition experiments. Top row: supply effect conveyed through increasing the number view at source ↗
Figure 6
Figure 6. Figure 6: Effect of added hybrid supply. Left panel: average control of operators E view at source ↗
Figure 7
Figure 7. Figure 7: Left panel: average control of operators E view at source ↗
Figure 8
Figure 8. Figure 8: Distributions across 500 heterogeneous markets with view at source ↗
Figure 9
Figure 9. Figure 9: Illustration of Major-Minor market compositions. view at source ↗
Figure 10
Figure 10. Figure 10: Average dispatch rate (left panel), per-unit revenue & cost (middle panel) and net profit (right view at source ↗
Figure 11
Figure 11. Figure 11: Average dispatch rate (left panel), per-unit revenue & cost (middle panel) and net profit (right panel) of Major and Minor operators. We consider a market consisting of N = 32 BESS units, with RM = 1 Major operator controlling M = 20 units and Rm = (N − M)/m Minor operators controlling m units each. The left panel is normalized such that ‘1’ corresponds to the average maximum dispatch E[maxt≤T ◦ |αˆ (M) t… view at source ↗
read the original abstract

We develop a stochastic game-theoretic model for intraday dispatch of grid-scale battery energy storage systems (BESSs). We assume that each BESS operator competitively manages her state-of-charge to maximize energy arbitrage revenues, driven by the endogenized electricity price that depends on the sum of the charging rates. We characterize the Nash equilibrium of the resulting finite-player linear-quadratic differential game with a shared stochastic driver, obtaining semi-explicit representations of equilibrium feedback controls and equilibrium prices both in the general heterogeneous and the simplified homogeneous BESS setting, via a system of Riccati equations. We then analyze competitive effects, including the marginal externality of additional BESS entering the market, the benefit of coordination and the corresponding market power of large operators, and supply effects from hybrid-type BESSs. We further study the asymptotic regime as the number of agents grows large. Our model provides a quantitative testbed to study the impact of decentralized BESS deployment on the grid and the resulting reduction in daily price spreads.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper develops a stochastic game-theoretic model for intraday dispatch of grid-scale battery energy storage systems (BESSs). Each operator maximizes arbitrage revenues under an endogenized electricity price that is linear in the aggregate charging rate plus a shared stochastic driver. The interaction is cast as a finite-player linear-quadratic differential game; the Nash equilibrium is characterized by semi-explicit feedback controls and prices obtained from a system of Riccati equations, both in the general heterogeneous case and the simplified homogeneous case. The manuscript then examines competitive effects (marginal externality of entry, coordination benefits, market power of large operators, hybrid-BESS supply effects) and the large-population asymptotic regime, positioning the model as a quantitative testbed for decentralized BESS impacts on the grid and daily price spreads.

Significance. If the Riccati system is globally well-posed, the work supplies a tractable, semi-explicit framework for quantifying market power and competitive externalities in BESS deployment—an area of growing practical importance. The reduction to a coupled Riccati system for heterogeneous agents and the subsequent large-N analysis are technically attractive features that enable concrete comparative statics and asymptotic predictions without requiring fully numerical solution of the game at every step.

major comments (2)
  1. [Section 3 (equilibrium characterization via Riccati system)] The central claim that semi-explicit equilibrium representations are obtained via a system of Riccati equations for heterogeneous agents rests on the global existence, uniqueness, and positive-semidefiniteness of the solutions to the fully coupled backward ODE system whose coefficients depend on all agents’ individual capacities, costs, and efficiencies. Standard linear-quadratic theory guarantees only local existence; the manuscript provides neither a priori bounds nor verifiable conditions (e.g., smallness of heterogeneity or sufficiently strong terminal penalties) that would preclude finite-time blow-up or loss of definiteness over the intraday horizon. This issue is load-bearing for every subsequent result that invokes the equilibrium feedback laws.
  2. [Section 4 (competitive-effects and asymptotic analysis)] All comparative-static and asymptotic results in Section 4 (marginal externality of additional BESS, benefit of coordination, market power, hybrid-BESS supply effects, and large-N limit) presuppose that the Riccati system admits a unique global solution for the parameter regimes under consideration. Without such a guarantee, the claimed quantitative testbed properties cannot be asserted uniformly.
minor comments (2)
  1. [Abstract] The abstract states that representations are obtained “via a system of Riccati equations” but does not indicate the dimension of the system or any solvability conditions; a brief clarifying sentence would help readers assess the scope of the result.
  2. [Section 2 (model formulation)] Notation for the shared stochastic driver and the individual BESS parameters (capacities, efficiencies, cost coefficients) is introduced piecemeal; a consolidated table of symbols at the beginning of the model section would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful and constructive report. The two major comments both concern the global well-posedness of the coupled Riccati system that underlies the equilibrium characterization. We address each point below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Section 3 (equilibrium characterization via Riccati system)] The central claim that semi-explicit equilibrium representations are obtained via a system of Riccati equations for heterogeneous agents rests on the global existence, uniqueness, and positive-semidefiniteness of the solutions to the fully coupled backward ODE system whose coefficients depend on all agents’ individual capacities, costs, and efficiencies. Standard linear-quadratic theory guarantees only local existence; the manuscript provides neither a priori bounds nor verifiable conditions (e.g., smallness of heterogeneity or sufficiently strong terminal penalties) that would preclude finite-time blow-up or loss of definiteness over the intraday horizon. This issue is load-bearing for every subsequent result that invokes the equilibrium feedback laws.

    Authors: We agree that the manuscript invokes the coupled Riccati system without supplying explicit a priori bounds or sufficient conditions that guarantee global existence and positive-semidefiniteness on the finite horizon for arbitrary heterogeneity. Standard LQ theory indeed yields only local solutions. In the revision we will add an appendix that derives verifiable sufficient conditions (bounds on the heterogeneity parameters together with sufficiently strong terminal penalties) under which the backward ODE system admits a unique global solution that remains positive semidefinite. For the homogeneous case we will also supply an explicit global-existence argument. These additions will make the equilibrium feedback laws rigorous for the parameter regimes used in the numerical illustrations. revision: yes

  2. Referee: [Section 4 (competitive-effects and asymptotic analysis)] All comparative-static and asymptotic results in Section 4 (marginal externality of additional BESS, benefit of coordination, market power, hybrid-BESS supply effects, and large-N limit) presuppose that the Riccati system admits a unique global solution for the parameter regimes under consideration. Without such a guarantee, the claimed quantitative testbed properties cannot be asserted uniformly.

    Authors: We concur that every result in Section 4 rests on the existence of the equilibrium derived in Section 3. Once the sufficient conditions for global well-posedness are stated in the revision, the comparative-static and large-N statements will be explicitly qualified to hold under those conditions. We will also insert a short clarifying paragraph in the introduction and conclusion that delineates the scope of the quantitative testbed, thereby preventing any overstatement of uniformity. revision: yes

Circularity Check

0 steps flagged

No circularity: equilibrium derived from model primitives via standard Riccati reduction

full rationale

The paper posits a linear-quadratic stochastic differential game whose price is linear in aggregate control plus exogenous noise. The claimed semi-explicit Nash equilibrium is obtained by substituting the standard quadratic value-function ansatz into the coupled HJB equations, which produces a closed system of Riccati ODEs whose coefficients are explicit functions of the primitive parameters (capacities, costs, efficiencies). This reduction is the direct, non-circular consequence of the LQ structure assumed at the outset; no parameter is fitted to data and then re-labeled as a prediction, no self-citation supplies a uniqueness theorem, and no ansatz is imported from prior work by the same authors. The derivation therefore remains self-contained against the stated model primitives.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The model rests on standard domain assumptions from stochastic control rather than new postulates; no invented physical entities are introduced. Free parameters are the usual cost, capacity, and volatility coefficients typical of LQ games.

free parameters (2)
  • BESS-specific cost and capacity coefficients
    Input parameters to the linear-quadratic payoff functions that define the game.
  • Volatility and drift parameters of the shared stochastic driver
    Coefficients of the exogenous price shock process.
axioms (2)
  • domain assumption Electricity price depends linearly on the sum of all agents' charging rates plus an exogenous stochastic process
    Core modeling choice that endogenizes price and creates strategic interaction.
  • domain assumption The multi-agent interaction constitutes a linear-quadratic differential game
    Enables closed-form Riccati representation of the Nash equilibrium.

pith-pipeline@v0.9.0 · 5477 in / 1529 out tokens · 72791 ms · 2026-05-09T15:16:10.942005+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

24 extracted references · 4 canonical work pages

  1. [1]

    An extended mean field game for storage in smart grids

    Cl \'e mence Alasseur, Imen Ben Taher, and Anis Matoussi. An extended mean field game for storage in smart grids. Journal of Optimization Theory and Applications, 184 0 (2): 0 644--670, 2020

    2020

  2. [2]

    Intraday battery dispatch for hybrid renewable energy assets

    Thiha Aung and Mike Ludkovski. Intraday battery dispatch for hybrid renewable energy assets. arXiv preprint arXiv:2503.12305, 2025

    work page arXiv 2025

  3. [3]

    Storage games

    Sergei Balakin and Guillaume Roger. Storage games. The RAND Journal of Economics, Early View; to Appear, 2025

    2025

  4. [4]

    A stochastic optimal control solution to the energy management of a microgrid with storage and renewables

    Andrea Belloni, Luigi Piroddi, and Maria Prandini. A stochastic optimal control solution to the energy management of a microgrid with storage and renewables. In 2016 American Control Conference (ACC), pages 2340--2345. IEEE, 2016

    2016

  5. [5]

    A S tackelberg mean field game for green regulator with a large number of prosumers

    Maxim Bichuch, G \"o k c e Dayan kl , and Mathieu Lauri \`e re. A S tackelberg mean field game for green regulator with a large number of prosumers. In 2025 American Control Conference (ACC), pages 4129--4134. IEEE, 2025

    2025

  6. [6]

    Soaking up the sun: Battery investment, renewable energy, and market equilibrium

    R Andrew Butters, Jackson Dorsey, and Gautam Gowrisankaran. Soaking up the sun: Battery investment, renewable energy, and market equilibrium. Econometrica, 93 0 (3): 0 891--927, 2025

    2025

  7. [7]

    2024 special report on battery storage

    CAISO. 2024 special report on battery storage. Technical report, California Independent System Operator, May 29 2025. Available at https://www.caiso.com/documents/2024-special-report-on-battery-storage-may-29-2025.pdf

    2024

  8. [8]

    Price impact and long-term profitability of energy storage

    Roxana Dumitrescu, Redouane Silvente, and Peter Tankov. Price impact and long-term profitability of energy storage. arXiv preprint arXiv:2410.12495, 2024

    work page arXiv 2024

  9. [9]

    Trading in residential energy systems with storage: a kinetic mean-field approach

    Margherita Fabini, Andrea Pascucci, and Alessio Rondelli. Trading in residential energy systems with storage: a kinetic mean-field approach. arXiv preprint arXiv:2603.00713, 2026

    work page arXiv 2026

  10. [10]

    Price formation and optimal trading in intraday electricity markets with a major player

    Olivier F \'e ron, Peter Tankov, and Laura Tinsi. Price formation and optimal trading in intraday electricity markets with a major player. Risks, 8 0 (4): 0 133, 2020

    2020

  11. [11]

    Price formation and optimal trading in intraday electricity markets

    Olivier F \'e ron, Peter Tankov, and Laura Tinsi. Price formation and optimal trading in intraday electricity markets. Mathematics and Financial Economics, 16 0 (2): 0 205--237, 2022

    2022

  12. [12]

    Equilibrium price formation with a major player and its mean field limit

    Masaaki Fujii and Akihiko Takahashi. Equilibrium price formation with a major player and its mean field limit. ESAIM: Control, Optimisation and Calculus of Variations, 28: 0 21, 2022

    2022

  13. [13]

    Extended McKean-Vlasov optimal stochastic control applied to smart grid management

    Emmanuel Gobet and Maxime Grangereau. Extended McKean-Vlasov optimal stochastic control applied to smart grid management. ESAIM: Control, Optimisation and Calculus of Variations, 28: 0 40, 2022

    2022

  14. [14]

    Federated stochastic control of numerous heterogeneous energy storage systems

    Emmanuel Gobet and Maxime Grangereau. Federated stochastic control of numerous heterogeneous energy storage systems. Journal of Optimization Theory and Applications, 208 0 (3): 0 107, 2026

    2026

  15. [15]

    Deep fictitious play for finding M arkovian N ash equilibrium in multi-agent games

    Jiequn Han and Ruimeng Hu. Deep fictitious play for finding M arkovian N ash equilibrium in multi-agent games. In Mathematical and scientific machine learning, pages 221--245. PMLR, 2020

    2020

  16. [16]

    CAISO battery fleet crosses 15 GW after record 4.7 GW year

    Logan Hotz. CAISO battery fleet crosses 15 GW after record 4.7 GW year. Technical report, Modo Energy Research Blog, March 24 2026. Available at https://modoenergy.com/research/en/caiso-battery-fleet-2025-q4-15gw-interconnection-queue-standalone-colocated

    2026

  17. [17]

    Deep fictitious play for stochastic differential games

    Ruimeng Hu. Deep fictitious play for stochastic differential games. Communications in Mathematical Sciences, 19 0 (2): 0 325--353, 2021

    2021

  18. [18]

    A mean field game for capacity expansion modeling

    Emma Hubert, Dimitrios Lolas, and Ronnie Sircar. A mean field game for capacity expansion modeling. arXiv preprint arXiv:2507.10604, 2025

    work page arXiv 2025

  19. [19]

    Economics of grid-scale energy storage in wholesale electricity markets

    \"O mer Karaduman. Economics of grid-scale energy storage in wholesale electricity markets. Technical report, CEEPR Working Paper 2021-005, March 2023, 2023

    2021

  20. [20]

    Estimating the congestion benefits of batteries on electricity grids when network connections are unobserved

    A Justin Kirkpatrick. Estimating the congestion benefits of batteries on electricity grids when network connections are unobserved. Technical report, Working Paper, Michigan State U, 2025. https://www.justinkirkpatrick.com/Papers/kirkpatrickEnergyStorageJan2025.pdf

    2025

  21. [21]

    Large-scale battery storage, short-term market outcomes, and arbitrage

    Stefan Lamp and Mario Samano. Large-scale battery storage, short-term market outcomes, and arbitrage. Energy Economics, 107: 0 105786, 2022

    2022

  22. [22]

    On the existence of solutions to coupled matrix R iccati differential equations in linear quadratic N ash games

    G Papavassilopoulos and J Cruz. On the existence of solutions to coupled matrix R iccati differential equations in linear quadratic N ash games. IEEE Transactions on Automatic Control, 24 0 (1): 0 127--129, 2003

    2003

  23. [23]

    Continuous-time stochastic control and optimization with financial applications, volume 61

    Huy \^e n Pham. Continuous-time stochastic control and optimization with financial applications, volume 61. Springer Science & Business Media, 2009

    2009

  24. [24]

    Deep reinforcement learning for economic battery dispatch: A comprehensive comparison of algorithms and experiment design choices

    Manuel Sage and Yaoyao Fiona Zhao. Deep reinforcement learning for economic battery dispatch: A comprehensive comparison of algorithms and experiment design choices. Journal of Energy Storage, 115: 0 115428, 2025

    2025