A lot of bad data has been circulated which resulted in closures of everything
Fauci mentioned two days ago this model operates by continual feeding of data. This means it will fluctuate and at this moment since we aren't doing enough testing it is lacking a lot of information. It could very much be an overestimation but we won't know without the proper data.
As we do more testing, it is very possible or actually very likely the death rate will go down. But again, this is why testing is important because you need the data.
But this constant push about how stay at home orders are the wrong way to go about this ignores the trend of other countries who we are trying not to emulate in death rates. Again for some reason, Americans think they are immune to the same issues because...'Merica.
The death rate of that graph actually underestimated the number of deaths from yesterday. The model predicted 900. We had 1049 which is higher than what they predicted today. So, depending on how you look at the data, either it overestimated in some areas and underestimated in others. I find it more concerning they underestimated deaths than hospitalization.
The model is here
IHME | COVID-19 Projections
editing: Also the shaded area can be viewed as margin of uncertainty . The line represents trend line.