Zettawatts REC Price Forecasting
- Dave Galinski
- Nov 26, 2024
- 4 min read
Updated: Nov 30, 2024
November 2024
Download a formatted PDF version of the forecast below.
Overview
The purpose of this document is to describe the method and outcomes of Zettawatts spot REC price forecasting for prospective customers. The forecasts presented are the culmination of Zettawatts’ analysis, modelling, and consultation of industry experts.
The forecast models presented in this report utilize a dataset of historical REC pricing catalogued by ARGUS Media. The timeframe for the training data for this report captures data from January 2019 through August 2024. RECs which do not have pricing data for the full timeframe will omit values for years where there is no pricing data (e.g. Virginia which begins in 2021) and will be noted in the comments of that REC’s forecast.
Method
Zettawatts models future pricing using a technique that decomposes pricing history into multiple components and independently forecasts over each using Monte-Carlo Markov Chains (MCMC). This results in a forecast which not only provides average price expectations for a wide time horizon using a Bayesian approach, but confidence constraints from the probability distributions created by the MCMC chain to provide context for inherent risk.
The price data informing the model is midpoint (average) price captured on either daily or weekly time frames and then converted into a monthly average before being ingested by the model. The model then forecasts on a monthly frequency and is aggregated to a yearly value by taking the average of each future year’s month level projections.
The model generally uses multiplicative seasonalized growth to forecast RECs prices with some exceptions (noted in each REC’s comments). For example, RECs which have high downward price pressure, such as the Green-e. In these cases, growth is modelled logistically around a floor at 0 to prevent negative price forecasting.
Logistic growth is also used for some compliance RECs which are constrained by their state ACP to incur a soft price ceiling.
These forecasts are then further transformed according to an index driven by the consensus of Zettawatts market research which approximates the expected future REC supply and market demand for each individual REC.
Graphs
The graphs on the following pages display the forecast values and the confidence bounds of the forecast. The forecast value for each point is defined where the dark-yellow and light-yellow areas meet. These values are represented above the graphs in a table format. The upper bound of a 95% confidence interval is where the light-yellow area ends, and the lower bound of the confidence interval is the bottom of the dark-yellow area. The red line shows the actual average price values for comparison against the test forecast.
REC - California - 3 Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled additively.
REC - Connecticut Class 1 Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $55 based on current ACP expectations.
REC - Connecticut Class 3 Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $30 based on current ACP expectations.
REC - Maryland Tier 1 Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled additively.
REC - Massachusetts Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $45 based on current ACP expectations.
REC - National Green-e
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled logistically to maintain a price floor above $0.
REC - NEPOOL Dual Qualified Class 1
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $45 based on tether to MA Compliance and NH Compliance ACP.
REC - New Hampshire Class 1 Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $45 based on current ACP expectations.
REC - New Jersey Class 1 Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $80 based on current ACP expectations.
REC - New Jersey Solar Compliance Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled on a logistic curve.
REC - New Jersey Class 2 Compliance Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $40 based on current ACP expectations.
REC - New York Tier 1
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled additively. Reported data for model begins October 2021.
REC - PA Solar Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $50 based on current ACP expectations.
REC - PA Tier 1
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled additively.
REC - PJM Tri-Qualified Class 1 Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled additively. Highly correlated to PA tier 1.
REC - Texas Compliance Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled logistically to maintain a price floor above $0.
REC - Texas Green-e Wind Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled logistically to maintain a price floor above $0.
REC - Texas Solar Year
Comment
Seasonality modeled on multiplicative growth, using a yearly basis. Growth trend modeled logistically to maintain a price floor above $0. Reported data for model begins October 2021.
REC - Virginia Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $45 based on current ACP expectations. Reported data for model begins October 2021.
SREC - DC Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $430 based on current ACP expectations.
SREC - Maryland Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $60 based on current ACP expectations.
SREC-II - Massachusetts Compliance Year
Comment
Seasonality modeled on additive growth, using a yearly basis. Growth trend modeled on a logistic curve constrained at $60 based on current ACP expectations.
Comments