Rather than guessing, he should go to the source. It's a statistical model, using empirical data collected in the US and overseas. They've documented it well, including the source code. Their main article on it is at:
http://www.healthdata.org/sites/default/files/files/Projects/COVID/RA_COVID-forecasting-USA-EEA_042120.pdf
They intended the model to be used for predicting hospital resources needed. It's getting misused and they seem to be allowing for that to happen. Now they are predicting dates for relaxing containment, but are basing that on one reported case per million of population!
Joke post?
Please don’t make me laugh too much ok?
It’s a crap model. They pick a number in their eange. The range runs across stupid values. eg 5-430 deaths in a day