Back in late July, I did an analysis of the FL outbreak spike and predicted that the peak of the seven day rolling average would likely be on or about 8/21 and that it would subside to the pre-spike levels around 9/27, just in time for the 50th. The peak wasn't captured in the case data because the positivity got so high during the plateau but the hospitalization graph shows the likely real shape of the case curve. I drew in red what it probably really looked like. If you take the endpoints of the plateau and find the middle you get 8/21.
The rate of decline slowed a bit this week likely due to some combination of labor day and school starting a few weeks ago. If the rate of decline doesn't increase again then it could be 4-6 weeks from now to get back down to that 1500 case per day level and sustained under 5% positivity. If the rate of decline increases again it will probably the first week of October.
My unsophisticated model didn't have the ability to predict the effect of labor day changing the variables up.
View attachment 587311