Agostino Capponi

2025

Read Agostino Capponi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-11979.txt.

Introduction

Agostino Capponi, Garud Iyengar, Bo Yang, and Daniel Bienstock model virtual trading in power markets. Their paper is from two thousand twenty-five, though the pack title names only the first author. In these markets, power is bought a day ahead and then balanced in real time. Virtual traders bet on the price gap between the two, with no power to deliver.

Without them, buyers bid low in the day-ahead market, which drags its price below the real-time price. Virtual trading shrinks that gap, and can close it when traders are many. Yet the amount bought a day ahead still falls short of true need. Data from two state grid firms back the model.

Catalog record

YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE126 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-11979

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Agostino Capponi Abstract Virtual Trading in Multi-Settlement Electricity Markets Garud Iyengar Department of Industrial Engineering and Operations Research, Columbia University, New York, NY USA [email protected] Bo Yang Department of Industrial Engineering and Decision Analytics, The Hong Kong University of Science and Technology Hong Kong SAR, China [email protected] Daniel Bienstock Department of Industrial Engineering and Operations Research, Columbia University, New York, NY USA Department of Applied Mathematics, Columbia University, New York, NY USA [email protected] Abstract In the Day-Ahead (DA) market, suppliers sell and load-serving entities (LSEs) purchase energy commitments, with both sides adjusting for imbalances between contracted and actual deliveries in the Real-Time (RT) market. We develop a supply function equilibrium model to study how virtual trading—speculating on DA–RT price spreads without physical delivery—affects market efficiency. Without virtual trading, LSEs underbid relative to actual demand in the DA market, pushing DA prices below expected RT prices. Virtual trading narrows, and in the limit of large number traders can eliminates, this price gap. However, it does not induce quantity alignment: DA-cleared demand remains below true expected demand, as price alignment makes the LSE indifferent between markets and prompts it to reduce DA bids to avoid over-purchasing. Renewable energy suppliers cannot offset these strategic distortions. We provide empirical support to our main model implications using data from the California and New York Independent System Operators. Keywords: Virtual trading, electricity market, renewable energy, supply function equilibrium, market design 1 Introduction The U.S. electricity markets utilize a dual settlement system consisting of a day-ahead (DA) and real-time (RT) markets for the procurement and transaction of electricity. Market participants, including suppliers and load-serving entities (LSEs), make electricity reservations in the DA market based on anticipated conditions in the RT market. Their planning takes into account expected demand and the projected availability of renewable energy sources. Any differences between the reserved amounts and the actual electricity deliveries are adjusted in the RT market. This structure is designed to accommodate the diverse startup and response times of generation resources, and offer participants the possibility to hedge against fluctuations in prices. However, despite its advantages, this two settlement structure can introduce inefficiencies into the market. One common problem is the strategic price manipulation by LSEs, particularly those holding significant market shares at specific grid nodes. These large LSEs might alter their bid quantities in the DA market to lower DA prices (see, e.g., New England ISO 2004 , and Jha and Wolak 2023 ). This tactic, while advantageous for the manipulating LSEs, is considered “extremely detrimental” to the overall market’s long-term vitality as it hinders other participants’ capacity to effectively hedge risks ( PJM 2015 ). Virtual trading was implemented in electricity markets to mitigate market inefficiencies, specifically the disparity between DA and RT prices (see, e.g., PJM 2015 and Hogan 2016 ). Virtual bids, purely financial in nature, allow participants to assume financial positions in the DA market, which are subsequently cash settled or reversed in the RT market. The term ‘‘virtual” indicates that no physical delivery or consumption of electricity is required. There are two major types of virtual bids: increment (INC) offers and decrement (DEC) bids. 1 1 1 Electricity markets also allow Up-to-Congestion transactions. This transaction is economically equivalent to INC and DEC bids submitted simultaneously at two different power grid nodes. INC bids involve a virtual sale of electricity in the DA market, followed by a virtual purchase of the same amount in the RT market. Hence, the INC bidder competes with the suppliers in the DA market. In contrast, a DEC bidder competes with LSEs by virtually purchasing electricity in the DA market and then selling this amount in the RT market. The importance of virtual bidding in mitigating market power, especially that of LSEs, has been highlighted by PJM, an independent system operator (ISO) that coordinates the movement of wholesale electricity for 13 states and the District of Columbia. In their report (see PJM 2015 ), they write: “ The opportunities for INCs to mitigate supply-side market power are few, whereas the market design features of the Day-Ahead Market that result in persistent underbidding of load in the Day-Ahead Market provide opportunities for DEC bids to mitigate demand side market power much more often. This is likely a primary driver in the reason cleared DEC volume exceeds cleared INC volume over the previous three planning years by a ratio of 1.5 to 1. ” Empirical evidence indicates that virtual trading has been effective in narrowing the price gap between DA and RT markets ( Jha and Wolak 2023 , Li et al. 2015 ) . Despite this, a detailed theoretical examination of the mechanisms leading to this price convergence is lacking. We introduce the first comprehensive framework aimed at explaining how virtual bidding contributes to narrowing the price differences between the DA and RT markets, as well as to aligning the trading quantities within these markets. Our model accounts for the growing penetration of wind and solar power—now supplying 30% of California’s electricity ( California Energy Commission 2022 ) —by explicitly incorporating renewable energy suppliers and their influence on market dynamics. With the flexibility to adjust output in response to weather conditions, renewable providers hold strategic advantages over traditional fossil fuel suppliers, influencing their market participation and behavior. We model the two-settlement market as a two stage game involving renewable and conventional suppliers, load serving entities, and virtual traders. 2 2 2 Our model is stylized, and in practice, markets are subject to numerous ad hoc adjustments and interventions. For example, bilateral agreements between generators, loads, and reserves are beyond the scope of this study and are therefore omitted. In the first stage, which we refer to as the DA market stage , both demand and the capacity of renewable suppliers are uncertain. However, the probability distributions governing these uncertainties are assumed to be common knowledge among all participants. 3 3 3 In practice, electricity markets have 24 hourly DA markets per day. Participants submit bids for each hour, and repeat this decision making process 24 times during the day. We consider one of these periods, consistently with standard modeling approaches in the literature (see, e.g., Sunar and Birge 2019 and references therein). The participants submit supply and demand bids, anticipating that they will be able to adjust their bids in the second stage, i.e., RT market stage , when the realized demand and capacity becomes common knowledge. 4 4 4 At the commencement of the RT market, the estimates for demand and capacities within the RT market period are highly accurate. In our model, we make the assumption that these quantities are perfectly known. In our model, both renewable and conventional suppliers submit supply function bids (referred to as cost curves in industry parlance) both in the DA and RT markets. These bids specify the quantities they seek to produce as a function of the market price. Renewable suppliers—whose capacities are subject to weather-driven uncertainty—have the opportunity to revise their supply functions in the RT market. In contrast, the DA supply functions submitted by conventional suppliers are fully observable to all market participants and are treated as binding. As a result, conventional suppliers are not permitted to revise their bids in

Read the complete text