Being able to forecast the generation mix (the percentage of each electricity producer- for example, coal, gas, wind) is extremely beneficial for the day traders in order to make informative price point decisions. The problem is very complex, given the length of time to predict ahead by, a large number of players, reliance on a grid infrastructure and variable out-of-grid/in-the-grid flow. We were able to build a successful deep learning ensemble model that predicted a day-ahead generationmix of electricity producers for a large US market.

‍