OK so the one thing that forecasters can not do is predict, or forecast when it arrives, what we called an exogenous event or shock. Mundane fairly accurate forecasts are based on all things equal. That assumes the past will be an accurate prologue to the future. But even those forecasts are a result of a forecast of the variables within the equation used. For example we see forecasts of future virus deaths, they must be based on previous history where the number of future patents is a forecast. Given no historical experience with this virus in the current situation we have to use as comarable history our statistics for SARS, Influenza and MERS, the most recent examples of virus deaths. Now using those curves does not reflect the mitigation, aka suppression, efforts underway. So there's a possible source of error which results in the forecast of deaths being changed almost from day to day. That is because as more data comes in they can model the virus based on its own actual data. That is why they are using other nation's experience, the problem being how accurate is that information.
Bottom line, look at the trend of the forecasts to understand where we are gong. Eventually as the virus data becomes super abundant we will know how this virus is affecting us as all the forecasts will converge.
P.S. Also we need to keep in mind that many persons who are being reported as dying from the virus have underlying problems which in and of itself could cause death. I do not know of any multivariate analysis, one scientist mentioned this issue, which has been done to take out of the forecast deaths from other causes from those attributed to the virus alone
.