The American-Made Net Load Forecasting Prize is designed to incentivize innovators to develop probabilistic models that predict amounts of net load a day in advance of the forecast and promote the adoption of probabilistic forecasts and evaluation tools for those forecasts. This prize offers up to $600,000 in cash prizes, with three anticipated winners and three anticipated runners-up.
Raising Forecasting Awareness and Standards
The U.S. Department of Energy’s Solar Energy Technologies Office, in co-sponsorship with EPRI, is launching the Net Load Forecasting Prize to:
Increase stakeholder awareness of the state of the art in net load forecasting;
Demonstrate the feasibility of fair and high-quality evaluations of probabilistic net load forecasts using a publicly available open-source platform; and
Promote the use of probabilistic forecast models and an industry-common evaluation platform with transparent metrics and specifications for probabilistic net load forecasts.
As co-sponsor of the prize, EPRI will provide historical net load data for competitors to train forecasting models and provide operational support for the forecast evaluation platform used during the evaluation phase of the prize.
Competition Structure and Prizes
Every day for four weeks (28 consecutive days), competitors submit forecast model results for at least four locations in the United States to the Solar Forecast Arbiter platform. This platform will be used to assess the models’ forecasting performance. The Solar Forecast Arbiter will compare how the forecasts perform to a benchmark forecast using a time-of-day persistence ensemble model.
The Net Load Forecasting Prize offers up to $600,000 in cash prizes. Administrators anticipate awarding up to 6 teams—3 winners and 3 runners-up—depending on performance of the competitors. See the official rules for full details.
Number of Prizes Awarded
Prizes
Winners
Up to 3 anticipated cash prizes
1st - $200,000
2nd - $150,000
3rd - $100,000
Runners-up
Up to 3 anticipated cash prizes
4th, 5th, and 6th each receive $50,000
Guidelines
What is “Net Load” Forecasting?
“Net load” refers to the difference between "true load"—the total electricity demand of the loads—and the electricity generation in the distribution system from resources such as solar and other distributed generators.
Learn more about net load forecasting in the official rules.
Who Can Compete?
Institutions, companies, and nonprofit organizations based in the United States are able to compete. In keeping with the goal of growing a community of innovators, competitors are encouraged to form diverse, multidisciplinary teams.
View the official rules for full eligibility requirements.
Solutions that Shape the Future of Forecasting
Are you an organization looking to advance the solar forecasting industry? You belong here. Join the American-Made Net Load Forecasting Prize to get started!
Follow the Challenge by clicking "Follow" in the upper right corner of the page to Get Started!
If you want to receive updates on the prize or have any questions, please subscribe by using the follow feature on the HeroX platform or message us directly at .
The American-Made Net Load Forecasting Prize is a multimillion-dollar prize competition designed to energize U.S. solar innovation through a series of contests that accelerate the entrepreneurial process from years to months. The American-Made Net Load Forecasting is directed and administered by the National Renewable Energy Laboratory and is funded by the U.S. Department of Energy Solar Energy Technologies Office. Learn more.
American-Made just launched a new $175,000 prize focused on distribution system state estimation (DSSE) tools. As a follow-up to the Net Load Forecasting Prize, the goal of the Data-Driven Distributed (3D) Solar Visibility Prize is to improve visibility of distribution systems and their operating conditions.
The prize, which is open now for applications, was developed by the U.S. Department of Energy’s Solar Energy Technologies Office to incentivize innovators to develop models and algorithms that can provide accurate and real-time information about distributed solar generation in electric power distribution networks.
This short, single-phase prize will require competitors to submit their DSSE solutions for two distribution system networks over the course of 14 days. The results will be compared against a set of bench metrics, and up to five winners will be awarded from a $175,000 prize pool.
The leaderboard results are in, and three teams—and three runners-up—have taken home top honors and a share of a $600,000 prize pool!
More than 70 teams competed over the summer, using their self-developed solar forecasting models to create net load predictions for four locations across the U.S. Once the competition kicked off, teams submitted their forecast model results every day for 28 consecutive days to the Electric Power Research Institute’s (EPRI’s) Solar Forecast Arbiter platform, which compared teams’ results to a benchmark forecast.
At the end of the test period, the following teams topped the leaderboard, and were announced as winners at the RE+ Conference on Sept. 12:
First place: Team Reliable Autonomy, Basking Ridge, NJ, awarded $200,000 (Represented by Garnet on the leaderboard)
Second place: Team ShiftedEnergy, Honolulu, HI, awarded $150,000 (Pearl)
Third place: Team Teraton Partners, San Francisco, CA, awarded $100,000 (Turquoise)
Additionally, these three runner-up teams received $50,000 each:
Team Leaptran, San Antonio, TX (Quartz)
Team Linear Intelligence, Apex, NC (Chrysocolla)
Team Innowatts, Houston, TX (Cuprite)
In partnership with EPRI, the U.S. Department of Energy’s Solar Energy Technologies Office developed the Net Load Forecasting Prize as a method to incentivize innovators to develop probabilistic models that predict amounts of net load a day in advance. Being able to accurately predict the net load of an electric grid allows operators to better plan for optimized procurement of reserves and utilization of other dispatchable energy sources.
Congratulations to the winners and everyone who participated!
The Net Load Forecasting Prize competitors have finally concluded their 28-day forecasting period. Thanks to the more than 70 teams who participated in this prize, improving their forecasting models every week.
At this time, the prize administrators at the U.S. Department of Energy and the Electric Power Research Institute are reviewing and validating final results. Winners will be announced live at RE+ in Las Vegas on Sept. 12. If you’re attending RE+, we invite you to join the announcement, happening during the American-Made Solar Prize Round 6 Go! Demo Day and Winner Announcement. If you can’t join the live announcement, we’ll share the results here, on HeroX, on Sept. 13.
As a reminder, the top three performing teams will win the following cash prizes:
1st - $200,000
2nd - $150,000
3rd - $100,000
Up to three runners-up will win $50,000 for their successful solar forecasting capabilities. Good luck to all competitors and stay tuned for the results in a few weeks!
We’re getting close to the end of the Net Load Forecasting Prize! More than 70 teams are completing their final days of the forecasting period and the top 10 teams are battling it out. At the end of week 4, we have Pearl moving back into first place, Garnet jumping up to second place, and Turquoise holding steady in third place.
Since June 18, teams have been submitting day-ahead forecasts every day for locations in Hawaii, Oregon, Texas, and Georgia. The forecasts are then compared against actual net load data (provided by four partner utilities) using the open-source Solar Forecast Arbiter platform. This makes the Net Load Forecasting Prize one of the largest and most realistic forecast competitions to date!
The leaderboard tab on HeroX shows the rankings for each of the four locations, with each team listed as an anonymous gemstone pseudonym. The rankings below show the cumulative results across all locations.
Competitors have just a few days left to upload forecasts and switch up the rankings. Check out the leaderboard tab for full standings in each of the locations and stay tuned for a final update next week and the winner announcement in September!
We’re halfway through the American-Made Net Load Forecasting Prize, with more than 70 teams battling it out to see whose probabilistic forecasts are best! At the halfway point of the competition, Quartz is in the lead, with Pearl in second place, and Turquoise rounding out the third-place spot.
Each team has been submitting day-ahead forecasts every day for locations in Hawaii, Oregon, Texas, and Georgia. The forecasts are then compared against actual net load data (provided by four partner utilities) using the open-source Solar Forecast Arbiter platform. This makes the Net Load Forecasting Prize one of the largest and most realistic forecast competitions to date!
The leaderboard tab on HeroX shows the rankings for each of the four locations, with each team listed as an anonymous gemstone pseudonym. The rankings below show the cumulative results across all locations.
Some notable highlights from the Net Load Forecasting Prize so far include:
Over 1 million forecast data points have been evaluated so far using the Solar Forecast Arbiter.
The teams include commercial forecast providers, startups, and 25 universities from across the U.S.
Locations with high solar penetration (e.g., Hawaii) have been the most challenging to forecast, due to the net load increasing significantly as behind-the-meter solar generation decreases from passing clouds. Unlike deterministic forecasts, the probabilistic forecasts can capture this uncertainty, as shown in this example:
Check out the leaderboard tab for full standings in each of the locations and stay tuned for weekly updates!
