Enertel’s Newest Product: Bidding Benchmarks

Backtested strategies and how traders can use them

How will an incoming windy, cloudy cold front raise net load? How will a line outage affect congestion at your gas plant? Predicting how grid conditions and weather events will affect power prices is challenging, yet the often‑stressful decisions that power traders make nationwide are precisely what keep the grid running every second of every day.

We wrote last week about how most decisions that a trader makes are based on predictions, which themselves are based on historical data inputs from a variety of sensors and ledgers.  It can be difficult and time consuming for traders to quantify how the predictions they’ve used all along – load, weather, or price – impacted their decisions. How did their outlook for real-time prices on July 23rd in CAISO or May 8th in ERCOT impact the quantity of power they purchased in the day-ahead market?

Today, we are formalising that bridge with Bidding Benchmarks—a new module that transforms Enertel’s nodal probabilistic forecasts into transparent and backtested bidding strategies to work alongside traders’ judgement.

How We Got Here

Enertel's forecasting services support our clients in the operation of a diverse energy portfolio exceeding 10 GW. Probabilistic forecasts help clients measure market risk and find new opportunities. We’ve made technical advances in how large neural networks can mimic the power grid. When we release new models, they use newer inputs with more data, allowing us to scale our neural networks even further.

Probabilistic forecasts, from many architectures of models, can yield an unruly amount of data for data scientists

We quickly evolved from a forecasting company that predicted hub prices to a data company that releases new probabilistic models monthly. More models and more probability levels caused the volume of this data to grow by orders of magnitude. The message from clients was the same – it’s helpful, but how should we use it all?

For some clients, forecasting a spike that never arrives merely forfeits upside, but missing an actual spike can wipe out a month of gains. Some clients measure the type 1 and type 2 errors of a new forecasting model, but many more prefer to see how the impacts show up on a cumulative profit plot. Some clients still just want to see our best guess at one number, which we can provide to them through many channels.

For all the different use cases, the end goal was always to use machine learning models and statistics to support power traders’ decision making. Predictions were a good start, but it only solved half the problem.

Enter Bidding Benchmarks

Bidding Benchmarks are backtested and fully transparent bidding strategies built on Enertel’s price forecasting engine. They are used by power traders to see how a given trading strategy (Risky, Conservative or Custom) has performed at their asset over time, and what those same strategies are suggesting for tomorrow.  

Hourly bids for the next day from a Risky strategy, shaped to a client's solar profile.

Bidding Benchmarks improve confidence for traders by providing a full suite of forecast models and their suggested bids based on your risk profile and asset type.

• Multiple Strategies– Conservative to aggressive, each strategy uses the same nodal probability paths

• Generation and Load Profiles – The module weights errors (and rewards) by the asset’s generation or load shape.

• Historical Backtests and Live Leaderboard – A full year of vintaged history plus a daily scoreboard showing which strategies have performed well

• Click-Through Audit – For any hour in the past year, you can open the original forecast distributions and see exactly how it shaped the bid.

• Human Override – Traders can nudge, replace, or merge strategies together

• Simple Baselines – They’re benchmarks for a reason: compare all trading strategies to DA-only, RT-only and Perfect Foresight

  
  

Day-ahead traders can reference these benchmarks after forming an opinion for the next operating day. Developers can see how their asset could be traded in the day-ahead market. What sort of day-ahead premium is being left on the table, and what’s the associated risk with it?

Probabilistic forecasts are tricky to use; Bidding Benchmarks converts them into a custom bidding strategy at your asset

Positive Feedback Loop

Enertel remains a forecasting company – Bidding Benchmarks is the optimization built on top of those forecasts that helps make a large volume of data more actionable for traders. The Forecasting and Bidding products work in sync: forecasting models are improved based on actual P&L, and the Bidding product is more transparent when tightly coupled with forecasting outputs.

Better inputs enable better predictions; better predictions enable better decisions; and better decisions, measured in dollars, spotlight where the model must improve next. That feedback cycle is the shortest path we know from probability to profit.

Enertel AI provides short-term energy price forecasts and bidding strategies for utilities, independent power producers (IPPs), and asset developers. You can view our data catalog, book a demo, or reach out to learn more.