12 - Predict Imbalance Price

1. Context: Imbalance Penalties in Switzerland & the 2026 Single-Price Reform 
Switzerland is experiencing growing volatility in its electricity system—largely driven by a surge in photovoltaic capacity and less predictable renewable generation. This has led to a marked increase in the volume and cost of imbalance penalties (i.e., charges to be paid by Balance Responsible Parties (BRPs) or simply put electricity suppliers, for deviations between contracted schedules and real-time production/consumption). In addition, Swissgrid is introducing a single price mechanism for balancing energy starting 1 January 2026. Under this new system, only those participants whose deviations destabilize the grid will be penalized—while grid-supportive behaviour will be rewarded. This shift aims to incentivize proactive balancing across all BRPs, encouraging them to contribute to the stability of Switzerland’s single balance area rather than solely focus on balancing their own group.

2. Challenge: Forecasting When a BRP Should Accept Imbalance or Use Market Trades or other approaches
Your mission: develop a prototype (MVP) of a strategic decision-support tool that helps a BRP determine whether to accept the financial risk of an imbalance penalty—or instead adjust positions via market trades or other approaches to minimize exposure. Specifically: 

• Predicting when is it economically reasonable for a BRP to let an imbalance occur, anticipating that the penalty cost is low or manageable?
• Predicting when should a BRP intervene—by buying or selling energy in the market or other types of flexibilities—to avoid potentially high imbalance penalties?
• Factors to consider include historical imbalance penalty trends, price volatility, and the economics of trading vs. penalization. This provides a strategic, actionable output—without exposing any proprietary optimization or trading algorithms. 

3. Data Inputs, Evaluation, & Research Questions 
Participants should leverage publicly available historical and weather data to power their forecasts. Possible data sources include: 

• Historical imbalance volumes and penalties
• Weather and solar irradiance forecasts (to model PV variability)
• Market prices from exchanges (e.g., EPEX SPOT)

Teams will explore forecasting horizons: how far in advance can we predict imbalance penalty prices with practical accuracy? Is a 1-hour, 3-hour, or 24-hour ahead forecast more reliable—and how do these windows affect the decision threshold between trading vs. accepting imbalance? Challenge deliverables: 

• A back-testing framework that showcases how well the model would have worked historically
• A clear evaluation metric (e.g., cost savings, prediction accuracy)
• Insights on prediction lead time vs. reliability trade-off

Questions to guide you:

• How far in advance can you make a plausible, actionable imbalance price prediction?
• For which time windows (e.g., 1 h, 3 h, 24 h ahead) is the prediction “good enough” to inform a trade decision?

This challenge is about finding the best predictive model—one that helps a BRP strategically decide whether to accept imbalance consequences or mitigate them via market actions, using only open data and without exposing confidential modeling strategies.