Share this
For data scientists, data engineers, and financial traders, locational marginal pricing (LMP) data is critical to creating robust and accurate energy price forecasts for understanding and modeling future grid behavior.
What Is Locational Marginal Pricing?
Published by Independent System Operators (ISO) and Regional Transmission Organizations (RTO), LMPs are time-series data sets also known as nodal pricings because they provide the cost of energy at a specific time and location (or node) in the power grid.
A locational marginal price is calculated based on energy loss and grid congestion levels in the node at a specific point in time. Energy loss naturally occurs because of the physics of moving energy from one place to another via transmission and distribution power lines. It’s relatively consistent over time.
However, the congestion component of the LMP calculation can vary significantly by location across 24 hours. Like a busy interstate, congestion occurs during peak demand when the grid operates near its limits. A bottleneck happens when a node becomes congested, making it difficult for the grid operator to meet demand.
Real-time LMPs dictate the purchase price of the electricity required to balance supply with demand in the node. As congestion increases, so does the node’s LMP.
For the day-ahead market, LMP calculations are done hourly; in the real-time market, they’re recorded every five minutes.
LMPs Are Key Benchmarks for Energy Traders
LMPs are the most important data signal energy traders can use to understand and predict future energy prices and grid behavior.
Traders monitor LMPs across different nodes to identify price discrepancies for arbitrage opportunities. They generate profits by purchasing electricity where prices are low and selling where they are high.
The challenge with monitoring LMPs is that they’re among the most frequently revised data sets – and those data updates can complicate traders’ forecasting efforts.
When an ISO/RTO publishes revised LMP data, most power market data providers overwrite the existing data points in their system so that only the most recently published data is available. While some customer operations prefer this, others find that access to all historical LMP records is critical to creating hyper-accurate backcasting models.
Backcast models generate energy price predictions based on a historical point in time. Therefore, not having accurate time series data for a point in time can cause inaccuracies in backcast models, ultimately impacting traders’ assumptions about future grid behavior.
Yes Energy Now Preserves All Historical Records of LMP and Other Main Data
Yes Energy has always kept all historical records of forecast data. To provide customers with robust visibility into the past not found in normal real-time data collections, Yes Energy recently expanded our main data sets to include the ISO record history for all existing LMP, ancillary services, generation, load, flow, and forecast data tables. This process, which we’ve dubbed “vintaging,” is unique in the industry.
Below is an example of our new, expanded Data Signals® Cloud hourly LMP table.
As you can see from the DATETIME and OBJECTID columns, this table preserves the three data records for the same five-minute real-time LMP interval. The metadata_timestamp column illustrates how we now preserve all data records – the data originally published by PJM on November 29 (INSERT) and the two subsequent updates on December 2 and 5 (UPDATE) – making them available in Yes Energy DataSignals Cloud.
Previously, only the most recent update would have been visible, partially eroding the price signal. There were ways to still capture these signals, but they involved code development and management on the customer side.
Should PJM make additional changes to the data for this time and node, rows with the new data will be automatically added to the table.
These expanded tables, available for all ISOs/RTOs, provide visibility and insight into each grid operator’s price formation process. They are also good for tracking down billing discrepancies from the ISO.
Additionally, these tables exist alongside our existing database tables so users can programmatically join them using the OBJECTID and DATETIME fields, creating a many-to-one relationship with the finalized data in the base tables.
Vintaging LMP Data: The Ultimate Time Machine
Why is historical LMP data that is overwritten so important?
It allows energy traders and data scientists to travel back to a specific point in time and recreate that moment as accurately as possible. With access to more robust data, they’re better able to evaluate the decisions made given the information available at the time – ultimately enabling them to create better models that produce more accurate forecasts at the nodal level.
Specifically, if you’re backcasting or your model leverages machine learning, you want to constrain your model to the data you had at the time of the event you’re evaluating.
For example, if the LMP at the time of the event was $25, but the ISO later updated it to $26, $25 would be the more accurate data point to use in your model because you made your decisions on that day based on that price point. If your model uses the updated $26 price point, it will skew your backcasts.
Simply put, evaluating decisions you made based on data you didn’t have is a lot like comparing apples to oranges. However, the most accurate forecasts are always based on apples-to-apples comparisons.
Optimize Your Energy Trading with Vintage LMP Data
Yes Energy’s DataSignals Cloud and DataSignals Lake deliver powerful, insightful, and actionable data on nodal power markets. Engineered and model-ready in the format most suited to your business needs, Yes Energy has a data delivery solution to translate comprehensive energy market data into actionable insights to power your business.
The addition of vintage LMP and other main data sets is just one way Yes Energy demonstrates its commitment to evolving and expanding our state-of-the-art DataSignals Cloud product. At Yes Energy, we believe this new feature, which is unique to our solution, will improve your ability to predict energy prices at the nodal level more accurately.
Contact Yes Energy to schedule a demo and learn how vintage LMP data can improve your predictions about nodal energy prices, help you to make better business and trading decisions, and drive more revenue.
About the author: Sam Lockshin is the product manager of the data products at Yes Energy. He has a passion for programmatically delivering Yes Energy’s high-quality power market data catalog to customers so they can achieve their business goals. You can catch him at karaoke, playing piano, or checking out the latest horror flick.
