Product-Differentiated Energy Markets

• Product-Differentiated Real-Time Energy Markets are trading systems that explicitly price electricity by physical, temporal, and contractual attributes rather than as a uniform commodity.
• They employ diverse clearing architectures—such as limit order books, peer-to-peer bilateral trading, and DSO-centric models—to ensure the market integrates physical constraints and network feasibility.
• Advanced pricing and settlement mechanisms, including nodal and temporal marginal pricing, incentivize efficient investments and enable optimal co-optimization of energy and grid-support services.

Searching arXiv for recent and foundational papers on product-differentiated real-time energy markets. First, searching for limit-order-book and transactive energy market mechanisms. A product-differentiated real-time energy market is a market design in which electricity is cleared and settled with explicit differentiation across product attributes rather than as a single undifferentiated commodity. In the cited literature, these attributes include location, time, source, bilateral trade criteria, real and reactive power, ramping capability, and multi-product bundles of electricity and grid-support services. The resulting designs appear in transactive energy limit order books, DSO-centric retail markets with nodal marginal prices, peer-to-peer bilateral trading frameworks, combinatorial local energy markets, and analytical competitive-equilibrium models that treat solar energy and grid energy as different products (Sreekumar et al., 2023, Haider et al., 2021, Sorin et al., 2018, Singhal et al., 31 Oct 2025, Davoudi et al., 8 Sep 2025).

1. Product dimensions and market definition

The literature operationalizes product differentiation along several distinct axes. In transactive retail markets, differentiation is supported by location through Distribution Locational Marginal Pricing, time through bids that vary across time intervals, and service attributes through revenue requirements and flexible output conditions (Huang et al., 2020). In DSO-centric retail markets, agents receive separate marginal prices for real power and reactive power, so grid-support services such as voltage regulation are priced as distinct products rather than embedded implicitly in a single energy price (Haider et al., 2021). In peer-to-peer formulations, bilateral trading costs encode criteria such as distance, green source, emissions, and social value, so the product is differentiated at the trade pair level rather than only at the node or market-wide level (Sorin et al., 2018).

In other formulations, product differentiation is tied directly to source or contractual structure. The distributed solar investment model distinguishes solar electricity from grid electricity, with buyers exhibiting a heterogeneous solar premium and the utility acting as a backstop at fixed price $\pi_u$ (Davoudi et al., 8 Sep 2025). The transactive-energy limit order book allows participants to submit limit orders containing not only price and quantity but potentially additional product attributes such as time or resource type, including examples such as renewable and time-of-use differentiated energy (Sreekumar et al., 2023). Multi-product local energy market designs broaden the product set further to electricity and grid-support services such as flexibility, frequency regulation, and reactive power, and allow participants to express preferences over bundles rather than isolated commodities (Singhal et al., 31 Oct 2025).

Differentiation axis	Representative mechanism	Source
Location	Node-level DLMP or d-LMP	(Huang et al., 2020, Haider et al., 2021)
Time	Time-varying interval bids	(Huang et al., 2020)
Source	Solar versus grid; green-only attributes	(Davoudi et al., 8 Sep 2025, Sreekumar et al., 2023)
Electrical service	Separate prices for real and reactive power	(Haider et al., 2021)
Flexibility	Energy plus ramping or flexibility options	(Chen et al., 17 Jun 2026, Spyrou et al., 2024)
Bilateral preference	Distance, emissions, social value	(Sorin et al., 2018)
Multi-product bundle	Package queries over electricity and grid-support products	(Singhal et al., 31 Oct 2025)

A recurring implication is that “product-differentiated” does not denote a single market rule. It denotes a family of market constructions in which heterogeneous physical, temporal, environmental, or contractual attributes are made explicit in bidding, clearing, and settlement.

2. Clearing architectures

One prominent architecture is the real-time limit order book for transactive energy systems. Participants post buy or sell limit orders, with buy orders ranked by descending price and sell orders by ascending price. Matching occurs whenever the highest bid price is at least the lowest ask price, and matched quantities are filled up to $\min(q_i^b,q_j^s)$; with attribute-based differentiation, matching can be constrained further by compatibility conditions such as

$$p_i^b \geq p_j^s \quad \text{and} \quad \text{attributes}_i=\text{attributes}_j .$$

The mechanism is designed for retail real-time electricity price discovery and includes support for discovering prices arising from wholesale electricity markets, distribution system asset constraints, distributed energy resource constraints, and consumer willingness to consume or produce at a reservation price (Sreekumar et al., 2023).

A second architecture is peer-to-peer multi-bilateral clearing. In the Multi-Bilateral Economic Dispatch formulation, each bilateral trade variable $P_{nm}$ is coupled to its counterpart through the reciprocity constraint

$P_{nm}+P_{mn}=0,$

and the shadow prices of these reciprocity constraints, $\lambda_{nm}$, become trade-specific prices. Product differentiation enters through the bilateral trading cost

$$\widetilde{C}_n(\boldsymbol{p}_n)=\sum_{m\in\omega_n} c_{nm}|P_{nm}|,\qquad c_{nm}=\sum_{g\in\mathcal{G}} c_n^g \gamma_{nm}^g,$$

which allows each peer to value distance, green source, or other criteria on a bilateral basis (Sorin et al., 2018).

A third architecture is the DSO-centric distributed retail market. Here the market-clearing problem is an optimal power flow solved by the Proximal Atomic Coordination algorithm, and the distribution-level Locational Marginal Price is determined by the dual variables of the nodal power-balance constraints. Agents exchange only necessary local variables with direct neighbors, and the market is described as operating in real time at intervals shorter than wholesale clearing, with 1 minute suggested in the summary (Haider et al., 2021).

A fourth architecture is the combinatorial clock exchange for multi-product local energy markets. At each iteration, the operator posts product prices, each participant returns its preferred feasible bundle by solving

$$\boldsymbol{x}_i^\star(\boldsymbol{\lambda})=\arg\max_{\boldsymbol{x}_i\in\mathcal{X}_i}\Big[v_i(\boldsymbol{x}_i)-\langle \boldsymbol{\lambda},\boldsymbol{x}_i\rangle\Big],$$

and prices are updated until approximate clearing. The bid language is a package query rather than a detailed supply or demand curve, and numerical simulations report convergence to clearing prices in approximately 15 clock iterations (Singhal et al., 31 Oct 2025).

These architectures differ in coordination topology—centralized operator, DSO-mediated distributed optimization, bilateral reciprocity, or iterative posted prices—but they share the property that differentiated products are present already in the clearing mechanism rather than introduced only ex post in settlement.

3. Pricing and settlement

In network-aware distribution markets, differentiated pricing is typically expressed through nodal marginal prices. In the DSO-dominated transactive retail framework, the time-varying Distribution Locational Marginal Price is the dual variable of the active power balance constraint for each node and timeslot, and it reflects local grid states, marginal operating cost, and network constraints. In the DSO-centric retail market, settlement is performed at nodal d-LMPs, with loads paying and generators being remunerated at their node’s marginal price; separate marginal prices $\mu_j^P$ and $\mu_j^Q$ are used for real and reactive power (Huang et al., 2020, Haider et al., 2021).

Intertemporal resources require a different pricing logic. For energy storage resources in a rolling-window real-time market, the literature shows that almost all uniform pricing schemes, including standard LMP, result in lost opportunity costs that require out-of-the-market settlements. Temporal Locational Marginal Pricing is proposed as an in-market discriminative scheme in which the storage discharge and charge prices are

$\min(q_i^b,q_j^s)$0

where $\min(q_i^b,q_j^s)$1 is the energy price and $\min(q_i^b,q_j^s)$2 is the individual state-of-charge price. In the summarized results, rolling-window TLMP eliminates lost opportunity costs and provides truthful-bidding incentives for price-taking firms under arbitrary forecasting errors (Chen et al., 2021).

Product differentiation also appears when energy is co-optimized with flexibility products. In the flexible ramping framework, the real-time market procures both energy and upward or downward ramping capacity under single-interval or multi-interval rolling-window stochastic optimization. The paper compares LMP with uniform pricing rules called maximum dispatch cost pricing and maximum temporal locational marginal pricing, and reports that with out-of-market bid cost recovery, LMP yields discriminatory energy prices, whereas MDCP eliminates BCR and MTLMP does so in most cases; empirical results on CAISO and ERCOT data show that MDCP and MTLMP increase producer profits with negligible BCR, albeit at the expense of higher consumer payments relative to LMP (Chen et al., 17 Jun 2026).

A different settlement design appears in Flexibility Options, which are dual-trigger financial products co-traded with day-ahead energy. Upward options are exercised when actual real-time output is below a quantity trigger and the real-time price is above a strike price, while downward options are exercised under the opposite directional conditions. The summarized formulation states that the product is co-optimized with day-ahead energy, is physically backed, has multi-tier differentiation, and ensures revenue-neutrality for the system operator because day-ahead premiums collected from buyers are paid to sellers while real-time settlements reverse the flow (Spyrou et al., 2024).

Across these designs, pricing is not merely a transfer rule. It is the main instrument through which differentiation by node, time, state of charge, ramping scarcity, or imbalance risk becomes economically operative.

4. Physical constraints, convexification, and decentralized computation

A product-differentiated real-time market is tightly coupled to physical feasibility. In the transactive retail market with DER retailers, the DSO clears a bi-level optimization in which flexible interval pricing is modeled through binary revenue constraints and lower-level dispatch uses an undirected second-order cone-based AC radial power flow model. The summary reports that approximation and relaxation techniques transform the original bi-level mixed-integer quadratic framework into a solvable mixed-integer semidefinite programming problem, and that the average absolute error between relaxed and traditional solutions is below 1% in the reported case study (Huang et al., 2020).

The DSO-centric retail market addresses the same coupling through distributed optimal power flow rather than a centralized bi-level formulation. The PAC algorithm decomposes the OPF into local subproblems with atomic constraints, while preserving nodal balances, voltage limits, line limits, and privacy of local information. Because d-LMPs are dual variables of the power-balance constraints, price differentiation is inseparable from network-feasible dispatch (Haider et al., 2021).

The peer-to-peer MBED framework tackles decentralization through a different route. The Relaxed Consensus+Innovation algorithm updates bilateral prices, reciprocity constraints, and local power bounds iteratively, while each agent shares only trade volumes and prices with its counterparties. The paper summary describes convergence with a negligible optimality gap and limited information sharing, which is central when differentiation is embodied in bilateral, peer-specific products rather than centrally standardized products (Sorin et al., 2018).

In multi-product local energy markets, the computational burden shifts from OPF convexification to preference elicitation and price discovery. The ML-aided combinatorial clock exchange uses all past package queries to learn surrogate value functions by inverse optimization, including a monotone-valued neural network example, and then selects adaptive step sizes through a learned approximation of the dual function. The stated purpose is to speed up convergence and lessen the communication burden while preserving a simple package-query interface (Singhal et al., 31 Oct 2025).

The transactive-energy limit order book occupies a comparatively lightweight end of the design spectrum. The studied case focuses on a single distribution feeder where network constraints are minimal, but the summary suggests that the mechanism can scale hierarchically to reflect physical distribution network limitations and that network usage costs can be handled through a tariff layered on settlement (Sreekumar et al., 2023).

5. Welfare properties, efficiency, and long-run investment effects

The strongest welfare result in the supplied literature is the analytical comparison among short-term market mechanisms for distributed solar investment. In the product-differentiated real-time market, solar electricity is traded as a distinct product, and when realized supply is scarce the equilibrium price is

$\min(q_i^b,q_j^s)$3

The same summary gives the Nash equilibrium aggregate investment condition

$\min(q_i^b,q_j^s)$4

and states that this condition is identical to the first-order condition for the social welfare optimum. The theoretical conclusion is that the product-differentiated market always supports socially optimal investment, the single-product real-time market consistently results in under-investment, and the contract-based market leads to over-investment when the extra valuations of users for solar energy are small (Davoudi et al., 8 Sep 2025).

At the operational level, the transactive retail mechanism with flexible interval pricing reports that, compared to fixed pricing, flexible interval pricing supports higher clean energy penetration and lower market-clearing costs. The summary also states that the fairness index in the objective function rewards DERs that offer aggressive lower prices, and that the mechanism can eliminate market power and the resulting market failures in the reported case study (Huang et al., 2020).

In peer-to-peer markets, product differentiation can change physical flows as well as welfare allocation. The 12-agent case study summarized for the MBED+RCI framework reports an almost identical objective value to centralized optimization, with only about 0.03% cumulative optimality gap over a year, about 0.1 s computation time on standard hardware, and average 298 iterations per time-step. Under a distance-based penalty, the same summary reports a more than 95% reduction in energy flow and more than 40% peak power reduction on the inter-bus line with less than 2% direct cost increase (Sorin et al., 2018).

The combinatorial clock exchange adds a separate efficiency result: multi-product markets significantly outperform sequential product-by-product markets, particularly under forecast uncertainty, by internalizing interdependencies and increasing social welfare. The same summary reports rapid convergence of the imbalance index, with basic CCE requiring about 15 iterations on average to reach market clearing and MLCCE converging even faster (Singhal et al., 31 Oct 2025).

Taken together, these results indicate that product differentiation affects not only short-term allocative outcomes but also long-term investment incentives, congestion patterns, fairness measures, and the relationship between decentralized preference revelation and welfare maximization.

6. Design tensions, misconceptions, and research directions

The literature does not confine product differentiation to source labeling alone. It includes node-level pricing, time-varying bids, real and reactive power, ramping services, bilateral preference criteria, imbalance-hedging instruments, and combinatorial bundles. A common misconception is therefore to equate differentiation only with “green energy” tagging; the supplied formulations are substantially broader (Haider et al., 2021, Chen et al., 17 Jun 2026, Spyrou et al., 2024).

A second tension concerns uniform versus differentiated pricing. For storage and other intertemporal resources, the summarized results state that uniform pricing schemes such as LMP often require out-of-market uplift or bid cost recovery, whereas TLMP embeds the state-of-charge value in market prices and MDCP or MTLMP can eliminate or largely eliminate BCR in flexible-ramping settings (Chen et al., 2021, Chen et al., 17 Jun 2026). At the same time, the ramping study reports that higher producer profits under MDCP and MTLMP come with higher consumer payments relative to LMP, so the choice of pricing rule remains distributive as well as operational (Chen et al., 17 Jun 2026).

A third tension concerns expressiveness versus usability. The combinatorial local market paper begins from the observation that prosumers may have complex, product-interdependent preferences but limited cognitive and computational resources, and answers this by replacing detailed bid curves with package queries and linear prices. The summary argues that the simple package-query format and transparent linear pricing reduce cognitive and computational burden, while theory based on the Shapley-Folkman-Starr theorem suggests that, as the number of prosumers increases, the duality gap and residual imbalances become small (Singhal et al., 31 Oct 2025). This suggests a trade-off between expressive product spaces and practical bid interfaces rather than a strict opposition between them.

A fourth issue is institutional. The DSO-centric market summary reports substantially differentiated prices—up to 2× in real power and 3.8× in reactive power across space and time—and an average retail price of $\min(q_i^b,q_j^s)$50.114/kWh in the reported simulation. The same summary states that the resulting lower revenue stream for the DSO highlights a move toward performance-based ratemaking and that advanced metering infrastructure and peer-to-peer communication are essential (Haider et al., 2021). Product-differentiated real-time markets therefore alter not only dispatch and settlement but also utility revenue models, metering requirements, and regulatory design.

The research trajectory represented by these papers points toward broader product sets, tighter coupling between physical constraints and settlement, and market interfaces that can express heterogeneity without overwhelming participants. A plausible implication is that future real-time market design will be judged less by whether it preserves a single commodity price and more by how effectively it exposes the relevant dimensions of flexibility, location, time, and preference within an implementable clearing mechanism.