If you are bewildered by the ever-changing and often conflicting Covid-19 projections coming out of Washington DC and state departments of health, you are not alone. Because of this, I decided to have a try at my own Covid model. A large part of my day job is data analytics so this is familiar terrain for me.
This project has two goals: predict short term deaths and long term deaths. I use the accuracy on the micro (daily) to guide the confidence interval on the macro (months out).
The “Ted Model” is evolving with the situation. It started out too conservative but now takes into account a combination of historical rates of change from both European countries and from the US. It also tries to anticipate the effects of the US’s distancing policies. For example, as distancing eases, the case numbers will accelerate, the death rate will decline initially, and then deaths and the death rate will accelerate again.
If you know of a model that is better predicting the future please let me know so I can incorporate additional ideas. So far, the projections that I’ve seen the government are laughably optimistic. We keep blowing past their projections and they never seem to comment on why they were so wrong.
In the US, my projections show 1-million deaths, plus or minus 500,000. And if I had to bet, I’d bet to the high side, not the low side.