4 Presentations

• 
  03:30 PM - 03:55 PM

  Limit Order Book simulation

  • Francis Zegbeh N'guessan, presenter, Université de Montréal

  Electronic markets have become the centerpiece of modern trading across all asset classes since the beginning of the 21st century. This has led to numerous research efforts aimed at understanding and developing algorithmic trading strategies. These algorithms require thorough testing before being deployed in a production environment to prevent catastrophic failures and the subsequent removal of the algorithm after inadvertently revealing a new strategy to the market. Several research avenues have attempted to address this issue using simulation environments.
  This document presents a novel methodology that bridges the gap between traditional parametric approaches and deep learning methodologies by applying a Conditional Variational Autoencoder (CVAE) with an LSTM architecture. This approach captures both autocorrelation and the limit order book’s level-by-level dynamics in a much more time-efficient manner than GAN-like approaches while being significantly more precise than non-machine-learning methodologies. This advancement paves the way for creating more efficient simulators and exploring the Canadian equity market, which has been less studied compared to its European and American counterparts.

• 
  03:55 PM - 04:20 PM

  A Distributionally Robust Optimization Strategy for Virtual Bidding in Two-Settlement Electricity Markets

  • Xavier Audet, presenter, Polytechnique Montréal
  • Antoine Lesage-Landry, Polytechnique Montréal
  • Kliti Qako, Hydro-Québec

  This work presents a new strategy to virtual bidding
  based on distributionally robust optimization (DRO) using a
  Wasserstein distance. Virtual bidding, a mechanism used in two-
  settlement electricity markets, allows participants to arbitrage
  price differences between day-ahead and real-time markets.
  Traditional optimization methods for virtual bidding often rely
  on precise probabilistic models of market behaviour, which are
  unavailable in practice due to the inherent complexity and
  volatility of electricity markets. To tackle these challenges, this
  work formulates the virtual bidding problem as a DRO problem,
  incorporating conditional value at risk (CVaR) into the objective
  to manage downside risk under volatile conditions. Tractable
  reformulations which can be efficiently solved to optimality
  are provided. The proposed strategy is developed and tuned
  using a 12-month training set to identify optimal parameters.
  The strategy is evaluated on historical pricing data from the
  New York Independent System Operator (NYISO) on an 8-
  month testing set. The results show improved performance over
  benchmarks, achieving higher Sharpe and Calmar ratios, as well
  as increased profit per MWh. Through this DRO framework, a
  more reliable virtual bidding strategy that enhances profitability
  while effectively managing risk in uncertain market environments
  is presented.

• 
  04:20 PM - 04:45 PM

  Cross-Border Renewable Integration: The Role of Quebec’s Hydro in Balancing Offshore Wind in New York

  • Soroosh Sabaei, presenter, PhD student
  • Michel Denault, HEC Montréal
  • Pierre-Olivier Pineau, HEC Montréal

  Decarbonizing large urban electricity systems requires not only the deployment of renewable generation, but also securing the flexibility needed to manage their variability. As regional grids become increasingly interconnected, new opportunities arise to share resources across borders and balance supply and demand more efficiently.
  In this context, New York State aims to install 9,000 MW of offshore wind capacity by 2035, potentially producing over 30 TWh annually—approximately half the electricity demand of New York City, which corresponds to Zone J within the New York Independent System Operator (NYISO). Integrating such a large volume of Variable Renewable Energy (VRE) will require substantial balancing capacity. This study explores the role of Quebec’s hydroelectric system, particularly the large reservoirs in the James Bay region, as a flexible and emissions-free balancing resource for New York’s Zones J and K.
  By modeling Quebec’s hydropower network using an hourly operational model, the study focuses on hydro operations in Quebec and long-distance transmission energy planning between Quebec and New York City. With the Champlain Hudson Power Express (CHPE) transmission line currently under construction, New York City is set to expand its access to clean electricity from Quebec. Additional transmission expansion could further enhance this cross-border integration.
  Moreover, we assess the potential for deploying wind power in the James Bay region. These high-quality wind resources, located nearly 2,000 km from the offshore wind sites planned along the U.S. East Coast, may exhibit complementary and potentially uncorrelated generation patterns—further supporting the overall balancing of the regional grid.
  This work presents an economic assessment of combining hydro flexibility, long-distance transmission, and expanded wind deployment in both regions. The results highlight the cost-effectiveness and reliability benefits of this approach, offering a viable alternative to natural gas peakers and battery storage.

• 
  04:45 PM - 05:10 PM

  On the Value of Saved Power in Net-Zero North-Eastern America

  • Michel Denault, presenter, HEC Montréal
  • Florian Mitjana, HEC Montréal
  • Pierre-Olivier Pineau, HEC Montréal

  We compute the Value of Saved Power: the value of replacing electricity by efficiency measures, delivering the same energy service.

  Our analysis is especially relevant in a transition era, where power prices are poised to increase under two factors: increased demand due to electrification, and decarbonization of the power system.
  
  In the various scenarios explored, we find positive value to saving power in all but extreme cases;
  and in what we consider to be the most probable scenarios, we find enormous value, of the order of 10 to 80 cents per kWh.

  For a power system undergoing major transformation (with load increasing due to electrification, and emissions curbed to limit climate changes), our results underscore the importance of computing the value of saved power by using the marginal cost of power including expansion costs; not the current power costs.