pith. sign in

arxiv: 2601.05085 · v3 · pith:G6GXLOWDnew · submitted 2026-01-08 · 💱 q-fin.TR

Trading Electrons: Predicting DART Spread Spikes in ISO Electricity Markets

Emma Hubert , Dimitrios Lolas , Ronnie Sircar This is my paper

Pith reviewed 2026-05-21 16:50 UTC · model grok-4.3

classification 💱 q-fin.TR
keywords DART spreadselectricity marketsprice impactspike predictionNYISOINC/DEC tradingoptimal tradingbid stacks
0
0 comments X

The pith

A structural price impact model from day-ahead bid stacks turns DART spike forecasts into optimal multi-zone INC/DEC trades that improve risk-return in NYISO.

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

The paper extends spike prediction for day-ahead to real-time price spreads from single zones to a multi-zone setting while treating positive and negative spikes in one statistical model. It introduces a price impact model built directly from day-ahead bid stacks to determine the best vector of zonal trade quantities. This model accounts for how buy and sell orders affect prices differently and how congestion links zones. When tested on NYISO data the resulting strategy shows a stronger risk-return profile than trading fixed unit sizes, with clear differences across seasons and markets. The work matters because electricity markets have real frictions that simple forecasts ignore, so translating signals into impact-adjusted positions can change net outcomes for traders.

Core claim

The central claim is that a structural price impact model derived from day-ahead bid stacks produces closed-form optimal INC/DEC quantities that capture asymmetric buy and sell impacts plus cross-zone congestion, and that applying this impact-aware strategy to NYISO data yields a significantly better risk-return profile than unit-size trading while revealing substantial heterogeneity across markets and seasons.

What carries the argument

The structural and market-consistent price impact model based on day-ahead bid stacks, which supplies closed-form expressions for the optimal vector of zonal INC/DEC quantities.

If this is right

  • The impact-aware strategy improves the risk-return profile relative to unit-size trading in NYISO.
  • Performance shows substantial heterogeneity across markets and seasons.
  • The unified model handles both positive and negative DART spikes without separate procedures.
  • Closed-form solutions incorporate asymmetric buy/sell impacts and cross-zone congestion directly into position sizing.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar bid-stack impact models could be calibrated and tested in other ISO markets such as PJM or MISO to check whether the performance gains generalize.
  • Improved trading accuracy might reduce the frequency or size of extreme real-time price swings if many participants adopted the approach.
  • The framework could be extended to include additional constraints such as transmission limits or ramping costs for even more realistic position sizing.

Load-bearing premise

The day-ahead bid stacks accurately reflect how chosen INC/DEC quantities create asymmetric price impacts and cross-zone congestion effects in real time.

What would settle it

A direct comparison of model-predicted price changes after specific INC or DEC trades against the actual observed real-time price movements in the same zones and hours.

Figures

Figures reproduced from arXiv: 2601.05085 by Dimitrios Lolas, Emma Hubert, Ronnie Sircar.

Figure 1
Figure 1. Figure 1: Zonal maps for NYISO, ISO–NE, and ERCOT. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: NYISO: cumulative P&L for NYC and Long Island under the INC/DEC bench [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: ISO–NE MAINE zone — P&L curves for overall, INC-only, and DEC-only strate [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: ERCOT WEST zone — P&L curves for overall, INC-only, and DEC-only strategies [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Supply stack and linear approximation near the DA price-setting intersection point [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Demand stack and linear approximation near the DA price-setting intersection [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Cumulative P&L: total (top) and by side and attribution view (bottom). All [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: NYISO: cumulative P&L by zone for remaining regions under the INC/DEC [PITH_FULL_IMAGE:figures/full_fig_p024_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Scatter of (Forecasted Load,Loss+Congestion) for 100 random Summer–Peak hours [PITH_FULL_IMAGE:figures/full_fig_p028_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Predicted vs. realized spike PDFs across [PITH_FULL_IMAGE:figures/full_fig_p031_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Predicted vs. realized spike PDFs across month of year (test period 2022–2025). [PITH_FULL_IMAGE:figures/full_fig_p031_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Cumulative P&L for the restricted (statistically significant) strategy, test period [PITH_FULL_IMAGE:figures/full_fig_p032_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Cumulative P&L with side-frozen (clipped) strategy. Top: total portfolio; bottom: [PITH_FULL_IMAGE:figures/full_fig_p033_13.png] view at source ↗
read the original abstract

We study the problem of forecasting and optimally trading day-ahead versus real-time (DART) price spreads in U.S. wholesale electricity markets. Building on the framework of Galarneau-Vincent et al., we extend spike prediction from a single zone to a multi-zone setting and treat both positive and negative DART spikes within a unified statistical model. To translate directional signals into economically meaningful positions, we develop a structural and market-consistent price impact model based on day-ahead bid stacks. This yields closed-form expressions for the optimal vector of zonal INC/DEC quantities, capturing asymmetric buy/sell impacts and cross-zone congestion effects. When applied to NYISO, the resulting impact-aware strategy significantly improves the risk-return profile relative to unit-size trading and highlights substantial heterogeneity across markets and seasons.

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 manuscript extends spike prediction for day-ahead versus real-time (DART) price spreads from single-zone to multi-zone settings in U.S. ISO electricity markets. It introduces a structural price impact model derived from day-ahead bid stacks that yields closed-form expressions for the optimal vector of zonal INC/DEC quantities, incorporating asymmetric buy/sell impacts and cross-zone congestion. Empirical results on NYISO data indicate that the resulting impact-aware trading strategy improves the risk-return profile relative to unit-size trading and exhibits substantial heterogeneity across markets and seasons.

Significance. If the closed-form optimal quantities prove robust and the reported improvements hold under realized slippage and out-of-sample conditions, the work would provide a practical, market-consistent framework for position sizing in electricity trading that integrates directional forecasting with execution modeling. The explicit treatment of multi-zone effects and asymmetry is a constructive extension of prior single-zone approaches.

major comments (2)
  1. [§3.2] §3.2, price-impact derivation: the closed-form optimal INC/DEC quantities are obtained by inverting a static day-ahead bid-stack model; the manuscript must demonstrate that this inversion remains accurate when real-time offers, ramp constraints, and intra-hour congestion deviate from the day-ahead stack, as these deviations directly affect whether the claimed risk-return improvement is realizable rather than an artifact of the modeling assumptions.
  2. [Table 4] Table 4 (NYISO results): the reported improvement in risk-return metrics versus unit-size trading lacks accompanying statistical significance tests, realized slippage measurements, and explicit out-of-sample validation periods; without these, it is difficult to assess whether the heterogeneity across seasons and zones supports the central claim or reflects in-sample fitting.
minor comments (2)
  1. [§2] Notation for the multi-zone impact matrix and congestion terms should be defined more explicitly at first use to aid readability for readers unfamiliar with ISO market mechanics.
  2. [Abstract] The abstract and introduction would benefit from a concise statement of the precise performance metric (e.g., Sharpe ratio or certainty-equivalent return) used to quantify the risk-return improvement.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We are grateful to the referee for the thoughtful and constructive feedback on our manuscript. We address the major comments below and outline the revisions we intend to incorporate.

read point-by-point responses

  1. Referee: [§3.2] §3.2, price-impact derivation: the closed-form optimal INC/DEC quantities are obtained by inverting a static day-ahead bid-stack model; the manuscript must demonstrate that this inversion remains accurate when real-time offers, ramp constraints, and intra-hour congestion deviate from the day-ahead stack, as these deviations directly affect whether the claimed risk-return improvement is realizable rather than an artifact of the modeling assumptions.

    Authors: We recognize that the closed-form solution is derived under the assumption of a static day-ahead bid stack. Real-time conditions can introduce deviations due to updated offers, ramping limits, and congestion. In the revised manuscript, we will expand §3.2 to include a discussion of these modeling assumptions and their potential impact on the optimal quantities. We will also perform a sensitivity analysis by introducing controlled perturbations to the bid stack to assess robustness of the risk-return improvements. Full empirical validation against realized real-time data would require additional data sources beyond the scope of the current study. revision: partial

  2. Referee: [Table 4] Table 4 (NYISO results): the reported improvement in risk-return metrics versus unit-size trading lacks accompanying statistical significance tests, realized slippage measurements, and explicit out-of-sample validation periods; without these, it is difficult to assess whether the heterogeneity across seasons and zones supports the central claim or reflects in-sample fitting.

    Authors: We agree that additional statistical rigor would strengthen the empirical claims. In the revision, we will augment Table 4 with statistical significance tests (e.g., using bootstrap methods or paired t-tests) for the differences in Sharpe ratios and other metrics. We will also explicitly delineate the out-of-sample periods used in the rolling-window evaluation and clarify that the backtests incorporate the modeled price impact but do not include additional realized slippage beyond the structural model. We will add a note on this limitation and discuss how the heterogeneity across seasons and zones is consistent with known market characteristics. revision: yes

standing simulated objections not resolved
  • Full demonstration that the inversion remains accurate under real-time deviations would require access to granular real-time offer and constraint data not available in public datasets used for the study.

Circularity Check

0 steps flagged

No significant circularity detected; derivation remains self-contained

full rationale

The paper extends the Galarneau-Vincent et al. framework for single-zone spike prediction to a multi-zone unified statistical model and separately develops a structural price impact model from day-ahead bid stacks to obtain closed-form optimal zonal INC/DEC quantities. No load-bearing step in the abstract or described chain reduces a prediction or optimal quantity to a fitted parameter by construction, nor does it rely on self-citation chains or imported uniqueness theorems. The impact model is presented as market-consistent and structural rather than ansatz-smuggled or renamed empirical pattern, and performance claims are evaluated against unit-size trading benchmarks on NYISO data. The derivation chain therefore retains independent content from its modeling assumptions and external data application.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities can be extracted or verified from the provided text.

pith-pipeline@v0.9.0 · 5661 in / 1058 out tokens · 28982 ms · 2026-05-21T16:50:02.618523+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

34 extracted references · 34 canonical work pages

  1. [1]

    An optimal trading problem in intraday electricity markets.Mathematics and Financial Economics, 10(1):49–85, 2016

    Ren´ e A¨ ıd, Pierre Gruet, and Huyˆ en Pham. An optimal trading problem in intraday electricity markets.Mathematics and Financial Economics, 10(1):49–85, 2016

    work page 2016

  2. [2]

    Optimal execution of portfolio transactions.Journal of Risk, 3(2):5–39, 2001

    Robert Almgren and Neil Chriss. Optimal execution of portfolio transactions.Journal of Risk, 3(2):5–39, 2001

    work page 2001

  3. [3]

    Bushnell, and Frank A

    Severin Borenstein, James B. Bushnell, and Frank A. Wolak. Measuring market inef- ficiencies in California’s restructured wholesale electricity market.American Economic Review, 92(5):1376–1405, 2002. 21

    work page 2002

  4. [4]

    Christensen, A

    Timothy M. Christensen, A. Stan Hurn, and Kenneth A. Lindsay. Forecasting spikes in electricity prices.International Journal of Forecasting, 28(2):400–411, 2012

    work page 2012

  5. [5]

    Christensen, Stan A

    Timothy M. Christensen, Stan A. Hurn, and Kenneth A. Lindsay. It never rains but it pours: modeling the persistence of spikes in electricity prices.The Energy Journal, 30(1):25–48, 2009

    work page 2009

  6. [6]

    Clements, Rodrigo Herrera, and A

    Adam E. Clements, Rodrigo Herrera, and A. Stan Hurn. Modelling interregional links in electricity price spikes.Energy Economics, 51:383–393, 2015

    work page 2015

  7. [7]

    Stochastic behaviour of the electricity bid stack: From fundamental drivers to power prices.The Journal of Energy Markets, 2(1):29–69, 2009

    Michael Coulon and Sam Howison. Stochastic behaviour of the electricity bid stack: From fundamental drivers to power prices.The Journal of Energy Markets, 2(1):29–69, 2009

    work page 2009

  8. [8]

    Powell, and Ronnie Sircar

    Michael Coulon, Warren B. Powell, and Ronnie Sircar. A model for hedging load and price risk in the Texas electricity market.Energy Economics, 40:976–988, 2013

    work page 2013

  9. [9]

    Estimating fast mean-reverting jumps in electricity market models.ESAIM: Probability and Statistics, 24:963–1002, 2020

    Thomas Deschatre, Olivier F´ eron, and Marc Hoffmann. Estimating fast mean-reverting jumps in electricity market models.ESAIM: Probability and Statistics, 24:963–1002, 2020

    work page 2020

  10. [10]

    Electricity intraday price modelling with marked Hawkes processes.Applied Mathematical Finance, 29(4):227–260, 2022

    Thomas Deschatre and Pierre Gruet. Electricity intraday price modelling with marked Hawkes processes.Applied Mathematical Finance, 29(4):227–260, 2022

    work page 2022

  11. [11]

    ERCOT market data.https://www.ercot.com,

    Electric Reliability Council of Texas. ERCOT market data.https://www.ercot.com,

    work page

  12. [12]

    Accessed for Day-Ahead and Real-Time prices, load forecasts, and market data

    work page

  13. [13]

    Hedging local volume risk using forward markets: Nordic case.Energy Economics, 68:490–514, 2017

    Rune Ramsdal Ernstsen, Trine Krogh Boomsma, Martin Tegn´ er, and Anders Skajaa. Hedging local volume risk using forward markets: Nordic case.Energy Economics, 68:490–514, 2017

    work page 2017

  14. [14]

    Self-exciting jump processes with applications to energy markets.Annals of the Institute of Statistical Mathematics, 70(2):373–393, 2018

    Heidar Eyjolfsson and Dag Tjøstheim. Self-exciting jump processes with applications to energy markets.Annals of the Institute of Statistical Mathematics, 70(2):373–393, 2018

    work page 2018

  15. [15]

    Distributional forecast- ing of electricity DART spreads with a covariate-dependent mixture model.Energy Economics, 144:108332, 2025

    Anthony Forgetta, Fr´ ed´ eric Godin, and Maciej Augustyniak. Distributional forecast- ing of electricity DART spreads with a covariate-dependent mixture model.Energy Economics, 144:108332, 2025

    work page 2025

  16. [16]

    Foreseeing the worst: Forecasting electricity DART spikes.Energy Economics, 119:106521, 2023

    R´ emi Galarneau-Vincent, Genevi` eve Gauthier, and Fr´ ed´ eric Godin. Foreseeing the worst: Forecasting electricity DART spikes.Energy Economics, 119:106521, 2023

    work page 2023

  17. [17]

    No-dynamic-arbitrage and market impact.Quantitative Finance, 10(7):749–759, 2010

    Jim Gatheral. No-dynamic-arbitrage and market impact.Quantitative Finance, 10(7):749–759, 2010

    work page 2010

  18. [18]

    Springer, 2 edition, 2009

    Trevor Hastie, Robert Tibshirani, and Jerome Friedman.The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Springer, 2 edition, 2009

    work page 2009

  19. [19]

    The modeling and forecasting of extreme events in electricity spot markets.International Journal of Forecasting, 30(3):477–490, 2014

    Rodrigo Herrera and Nicol´ as Gonz´ alez. The modeling and forecasting of extreme events in electricity spot markets.International Journal of Forecasting, 30(3):477–490, 2014. 22

    work page 2014

  20. [20]

    William W. Hogan. Virtual bidding and electricity market design.The Electricity Journal, 29(5):33–47, 2016

    work page 2016

  21. [21]

    FAQs: Locational marginal pricing, 2025

    ISO New England. FAQs: Locational marginal pricing, 2025. Accessed: 2025-03

    work page 2025

  22. [22]

    ISO New England markets and operations data.https://www

    ISO New England. ISO New England markets and operations data.https://www. iso-ne.com/markets-operations/markets, 2025. Accessed for Day-Ahead and Real- Time prices, load forecasts, and market data

    work page 2025

  23. [23]

    Kelly Stegmann, Mathangi Srinivasan Kumar, and Gina E. Craan. Congestion price component. Technical report, New York Independent System Operator (NYISO), September 2025. LBMP In-Depth Course, Remote Learning

    work page 2025

  24. [24]

    New York Indepen- dent System Operator (NYISO), October 2025

    Mathangi Srinivasan Kumar.Locational Based Marginal Pricing. New York Indepen- dent System Operator (NYISO), October 2025. New York Market Orientation Course (NYMOC), Rensselaer, NY

    work page 2025

  25. [25]

    Forecasting day- ahead electricity prices: A review of state-of-the-art algorithms, best practices and an open-access benchmark.Applied Energy, 293:116983, 2021

    Jesus Lago, Grzegorz Marcjasz, Bart De Schutter, and Rafa l Weron. Forecasting day- ahead electricity prices: A review of state-of-the-art algorithms, best practices and an open-access benchmark.Applied Energy, 293:116983, 2021

    work page 2021

  26. [26]

    Modeling and forecasting electricity spot prices: A functional data perspective.The Annals of Applied Statistics, 7(3):1562–1592, 2013

    Dominik Liebl. Modeling and forecasting electricity spot prices: A functional data perspective.The Annals of Applied Statistics, 7(3):1562–1592, 2013

    work page 2013

  27. [27]

    Divergence measures based on the Shannon entropy.IEEE Transactions on Information Theory, 37(1):145–151, 1991

    Jianhua Lin. Divergence measures based on the Shannon entropy.IEEE Transactions on Information Theory, 37(1):145–151, 1991

    work page 1991

  28. [28]

    Longstaff and Ashley W

    Francis A. Longstaff and Ashley W. Wang. Electricity forward prices: A high-frequency empirical analysis.Journal of Finance, 59(4):1877–1900, 2004

    work page 1900

  29. [29]

    Learning in Stackelberg games with application to strategic bidding in the electricity market

    Francesco Morri, H´ el` ene Le Cadre, Luce Brotcorne, and Pierre Gruet. Learning in Stackelberg games with application to strategic bidding in the electricity market. In 2024 20th International Conference on the European Energy Market (EEM), pages 1–7. IEEE, 2024

    work page 2024

  30. [30]

    NYISO market data.https://www.nyiso

    New York Independent System Operator. NYISO market data.https://www.nyiso. com, 2025. Accessed for Day-Ahead and Real-Time prices, load forecasts, and market data

    work page 2025

  31. [31]

    An empirical com- parison of alternative schemes for combining electricity spot price forecasts.Energy Economics, 46:395–412, 2014

    Jakub Nowotarski, Eran Raviv, Stefan Tr¨ uck, and Rafa l Weron. An empirical com- parison of alternative schemes for combining electricity spot price forecasts.Energy Economics, 46:395–412, 2014

    work page 2014

  32. [32]

    Locational minimum installed capacity requirements study for the 2024–2025 capability year

    NYISO. Locational minimum installed capacity requirements study for the 2024–2025 capability year. Technical report, New York Independent System Operator (NYISO), December 2024

    work page 2024

  33. [33]

    Forecasting day-ahead price spikes for the ontario electricity market.Electric Power Systems Research, 141:450–459, 2016

    Harmanjot Singh Sandhu, Liping Fang, and Ling Guan. Forecasting day-ahead price spikes for the ontario electricity market.Electric Power Systems Research, 141:450–459, 2016. 23

    work page 2016

  34. [34]

    Magableh, Oraib Dawaghreh, Caisheng Wang, Jiaxuan Gong, Zhongyang Zhao, and Michael H

    Xuesong Wang, Sharaf K. Magableh, Oraib Dawaghreh, Caisheng Wang, Jiaxuan Gong, Zhongyang Zhao, and Michael H. Liao. Deep learning-based electricity price forecast for virtual bidding in wholesale electricity market.Preprint arXiv:2412.00062, 2024. A Appendix A.1 Tables and Figures for NYISO (a) CAPITL — INC (b) CAPITL — DEC (c) CENTRL — INC (d) CENTRL — ...

    work page arXiv 2024