As the author of Software Estimation: Demystifying the Black Art, it was natural for me to apply my forecasting skills to the pandemic. I began forecasting in May, and I saw that my 2 week forecasts tracked within 5-10%. This seemed like a normal level of accuracy to me, but I was seeing forecasts reported in the media that missed by 50-100% or more. That level of inaccuracy did not seem normal, useful, or necessary.
For the forecast shown at right, my error of 7.2% for the period would have made it the third most accurate out of the 27 forecasts submitted to the CDC. So in August I began officially submitting forecasts to the CDC for use in their Ensemble model. Since then, my forecast model, CovidComplete, has ranked among the most accurate models for every forecast period. You can see the details on my forecast evaluation page.
This page shows my current forecasts, most recent forecasts submitted to the CDC, and how my most recent forecasts are tracking.