We don’t want to read too much into a two- or three-day trend, as Cuomo himself notes in the clip, but New York’s daily death toll may be leveling off at 600 or so per day. The IHME projections had the daily death toll increasing for another five days, reaching a peak of 855 on April 10, before the long journey down the other side of the bell curve begins.

That wasn’t the only good trend noted in this morning’s presser:

Yesterday’s number seems suspiciously small. Is it based on incomplete data? If so, Cuomo shouldn’t have run it out there. If not, it’d be the state’s lowest number of new hospitalizations since March 20, a nearly 50 percent drop from yesterday. Hard to explain why that might have happened.

The number of people who were positive in daily testing was also down yesterday but I think that’s a function of the fact that testing is down on the weekend. Per the NYS website, on Thursday 21,555 people were tested statewide and 10,482 were positive, a rate of 48.6 percent on the day. On Friday it was 10,841 positives out of 23,101 tests, a rate of 46.9 percent. Yesterday, if my math is correct, just 8,327 people tested positive — but the number of tests administered, 18,659 (per Cuomo’s presser), was also down noticeably. Even so, that’s a rate of just 44.6 percent positive. Make of it what you will. Maybe the spread really is slowing down. Or maybe the drop is an artifact of reduced testing capacity on weekends somehow.

How do these actual numbers match up with the IHME model of the epidemic that reportedly heavily influenced the White House’s own model? It depends on what you’re measuring. IHME’s forecast of the daily death toll has been quite accurate so far:

That number was within 13 of the actual number provided by Cuomo. If anything, the model has underestimated deaths so far: On April 3 IHME projected 538 deaths from COVID-19 in New York State whereas the actual death roll was nearly 100 higher, beyond even the upper bound of the IHME’s range of likely fatalities for that day. Higher-than-expected numbers recently would make sense if in fact New York is peaking now instead of five days from now, as IHME’s forecast imagined, but it’d be hard to square an early peak with the social distancing measures that NYS has put it in place. Flattening the curve is supposed to delay the peak of the epidemic, not accelerate it, and to reduce the number of deaths during the peak. Right now New York may be peaking early — while also holding the number of peak deaths down from IHME’s projection of 855. How to reconcile that?

If you’re looking at hospitalizations instead of deaths, then IHME is way, way, waaaaaaaay off:

They thought 65,000+ hospital beds and 12,000 ICU beds would be needed to house the sick as of yesterday. The reality?

The current number of hospitalizations is literally a quarter of the IHME forecast. That’s a big miss. The number of new hospitalizations noted in that graphic, 574, also popped for me. Look at the trend over the last five days and you’ll see that the state was adding around 2,500 new hospitalizations per day (give or take a few hundred) steadily. Suddenly the bottom has fallen out of that number, with three times as many people discharged from hospitals as added. I’m guessing there must be a weekend artifact in counting here too, with some counties in New York possibly not reporting their numbers until Monday.

No matter what the numbers are tomorrow, though, the IHME model is obviously way out of whack with reality in New York on hospitalizations. What explains it? It could be that the model is wildly overestimating the general hospitalization rate from COVID-19, obviously — although, as noted, it’s not overestimating the raw number of fatalities. (So far.) Alternatively, it could be that the aggressive social distancing measures taken in New York State have succeeded better than anyone might have hoped in limiting the spread, preventing a giant wave of infections that would have made IHME’s expectations more realistic. It’s hard to believe that that explains everything, though, given how wide of the mark IHME’s expectations were. Even the lower bound of their hospitalization range for April 4 was nearly three times the number of actual hospitalizations.

Maybe IHME expected that each hospitalization on average would last much longer than they’ve actually lasted in reality. If they assumed a one-week stay for all patients and the actual stay is more like three days, the much higher turnover in the latter scenario will mean many fewer beds are needed to cope with the surge. If in fact IHME wildly overestimated on duration of hospitalization, though, then why did they overestimate? Were they using bad data from China? Or have U.S. doctors hit on a course of treatment that’s helping patients bounce back faster than expected? That’s be surprising given that the death toll isn’t less than expected at this point, but I’m curious to hear more.

In lieu of an exit question, read this WaPo piece about how coronavirus deaths are being counted. Some skeptics of the Fauci/Birx shutdown strategy believe that COVID-19 victims are being overcounted, that some who are infected with the virus may be dying of other underlying/unrelated causes but being counted as coronavirus victims simply because they tested positive. In reality, says WaPo, it’s likely that coronavirus victims are being undercounted. That’s because there are obviously many people out there who’ve contracted the disease and haven’t been tested for whatever reason (probably because tests are too scarce) and then aren’t being tested postmortem because, well, tests are too scarce. We saw that in Italy and Spain too, remember. The official COVID-19 death toll in those countries doesn’t fully account for the huge difference in overall deaths during March 2020 versus deaths during March 2019. Authorities are missing some of the disease’s casualties.