The original idea of taking the ratio of cumulative deaths today to that three days earlier was to be able to detect a sharp turn in the direction of the virus crisis. But it is not taking sharp turns, and at this point it is dominated by day-of-week factors.

For the record, today it is at 1.08, continuing its very gradual decline. Outside of New York, it is also declining, standing at 1.09.

If we look at 7-day intervals to correct for day-of-week effects, then we are not going to detect any sharp turning point. But that is probably the best we can do. I think that the simplest number to report would probably be either the 7-day total, or that total divided by 7 to get a daily average.

What I would most like to report would be 7-day averages broken out by age and whether or not the person resided in a nursing home.

What I will report, given what I can find, is the 7-day average outside of New York, because I think that the trend outside of New York will determine whether or not I win my bet against the models. The highest value so far is for 1546 average deaths per day for April 25 through May 1. The second highest value so far is 1539, for April 29 through today, May 5.

This figure rose from 705 for April 1-7 to reach 1476 April 15-21. It has been a plateau since then. Perhaps the shutdown orders stopped what otherwise would have been a steady increase. I would prefer to believe that the lockdowns had no significant effect, but we should let the facts speak for themselves. Going forward I will follow this 7-day average outside of New York to see what happens with easing the lockdowns. I won’t report this indicator every day, because it moves too slowly. Probably report about once a week or so.