We move to using forward power curves rather than forward gas curves in the forecast. We create a hourly forward curve using a daily shape on monthly futures contracts, based on time series analysis and a view of extreme pricing (both low and high) based on forecast volatility in the future.
Using the minimum and maximum daily values of price from this hourly forecast, we estimate the wholesale trading value for batteries of 1 and 2 hours in duration.
We also updated the build-out curve and update inputs to the wind, demand and dispatchable plant models, ensuring the machine learning model is robust to new data inputs.