Quantifying Portfolio Risk for Better Decisions at Scale: How ACES Manages 1,000+ Models with Positron & Posit Team

ACES operates in highly dynamic wholesale energy markets, helping cooperatives, municipal utilities and other participants manage complex power portfolios across all major U.S. regions. Each portfolio can span multiple fuels, locations, and regulatory environments, each with its own risk profile, time horizon, and market rules.

Frank Hull’s data science and analytics team is responsible for more than 1,000 models that determine a portfolio’s capacity, modeling everything from next-hour demand to 25-year price scenarios. Their core challenge is to deliver quantified risk scenarios that determine precisely when and why a portfolio, city, or county may be short on power, and the associated cost. 

For example, they calculate the short position and corresponding prices during high-stakes events—such as a 105℉ summer day where a shortage is unacceptable—versus manageable events like a spring nighttime shortage.

As renewables (wind and solar) have driven unprecedented volatility and complexity, ACES legacy analytical stack began creating unacceptable operational risk. They began facing the operational burden of:

• Unscalable Model Maintenance: Maintaining and documenting 1,000+ models across 40 to 50 portfolios was becoming an impossible administrative task, creating massive technical debt.
• Increasing Edge Cases: New solar farms, rooftop solar, EVs, data centers, & battery storage; completely change the hourly demand shape, price shape, and supply stack. This necessitated the shift from traditional, less adaptive Monte Carlo approaches to modern machine-learning-based stochastics with tidymodels.
• Fragmented Workflows: Managing this complexity with a patchwork of individual scripts, spreadsheets, and ad-hoc open-source tooling made it difficult to scale, standardize, and govern their analytics stack, threatening the timely, trustworthy forecasts needed by traders and strategists.

To tame the complexity and support sustainable growth, ACES adopted Posit Team (Posit Workbench, Posit Connect, and Posit Package Manager) as a unified analytics platform and embraced Positron as the primary development IDE for Hull’s data science & analytics team.

• Posit Workbench:  ACES data scientists benefit from a secure, robust, and polyglot environment that offers immediate access to scalable, server-side compute, allowing them to work seamlessly in both R and Python. As Frank Hull notes, “Positron (on Posit Workbench) stands out as a more robust developer IDE with strong polyglot support, seamlessly handling both R & Python to create a more versatile and productive experience.”
• Posit Connect: Enables ACES’s data team to deploy APIs and interactive applications that transform intricate models and forecasts into accessible tools for portfolio managers, strategists, and other business stakeholders . 
• Posit Package Manager: Centralizes internal packages the team uses across hundreds of models and ensures that packages and dependencies are reproducible and governed, minimizing breakage as open-source libraries evolve.

On top of Posit Team, the group standardized on tidymodels and Git-based workflows so that every new model is version-controlled, testable, and easier to document and reuse. Over time, they have moved from scripts maintained by individuals to packaged, documented, and shareable modeling components that the whole organization can build on.

These models built and deployed through Posit Team allow ACES to assist their members and customers in understanding supply, demand, price, and portfolio risk in the following ways:

• 1-5 Year Portfolio Modeling: Analyzing which hours in each season it’s necessary to hedge based on length and price extremity. In 1-5 years, they can’t build a new power plant but could use this analysis to make the decision to lock-in a Power Purchase Agreement (PPA) with another counterparty.
• 5-25 Year Integrated Resource Planning: Models directly inform investment decisions across strategic time horizons. Hull notes, “In 5-25 years, we could originate a long-term deal or build a farm/plant. We also update our long-term load forecast for the members each year to understand what they may need to plan for.”

With Posit Team and Positron, ACES can now support a rapidly growing portfolio of models and services without sacrificing reliability or speed.

• Scalable model portfolio: Hull’s team can more easily manage over 1,000 models—spanning hourly time-series, engineering models for wind and solar, and ML-based stochastic simulations—while keeping documentation, ownership, and governance under control.

• Faster, more productive development: Positron’s developer-first IDE experience (minimap, navigation, Visual Studio extensions) and strong polyglot support help the team move faster, debug more effectively, and comfortably mix R and Python where appropriate.

• Standardized, reusable modeling framework: By packaging functions, building internal R libraries, and using Posit Package Manager as a single source of truth, the team reduces technical debt and avoids “script sprawl.” New models are easier to build, review, and maintain.

• Better alignment with the business: Robust APIs and applications on Posit Connect allow portfolio strategists, traders, and risk teams to interact directly with model outputs—exploring scenarios, asking “what if?” questions, and incorporating model insights into operational and long-term planning.

• Readiness for emerging risks and opportunities: As renewables grow, prices become more volatile, and data centers drive new demand, ACES is equipped with a modern, flexible analytics stack that can evolve with the market—without having to re-platform or re-tool every time a new modeling challenge appears.