Y'all have seen my rants against the Murray #COVID19 model.

It predicts OVERWHELMED hospitals in the next 2 weeks!

Now we have actual stats on our hospitals and the models have MASSIVELY OVERSTATED the impact.

Are we wrecking the economy for no good reason?!

Here. We. Go.

Explainer:

# of hospital beds needed on the left (Y-axis)
the dates are at the bottom (X-axis)

What you're seeing is a wave (or as they say... the curve)

The big red mountain is what Murray is predicting for the ENTIRE country.

The blue line is the ACUTAL # of hospitalized

So based on the 40+ states which provide burden stats on hospitals Murray predicted that yesterday (April 1) we would been 103K beds for #COVID19 patients.

All of those states combined published that they ONLY had 30K beds with patients. and more than HALF of those were in NY

Here are 3 of the biggest targets for #COVID19. Washington was first out of gates. Murray predicts that YESTERDAY WA should have needed nearly 2000 beds for these sick patients.

They only have 254 in the hospital.

NY
Murray: 50K beds
Actuals 18K

CA: 4100 vs. 1100

Florida
Alaska
Arizona
Arkansas
Colorado

Miss, miss, miss, miss.

Even if you factor in data "lag" (which is a real problem)... it's hard to imagine that MURRAY is getting any of this right overstating things by 3x 4x 5x the number of beds needed.

CT
DE
FL
GA

If I'm misinterpreting these numbers... someone stop me..
But I don't think I am... because others have called it out

Caution and "buffer" is good... but to shut down the country when the model missed this badly?!

More here.
ID
IA
KS
LA

Notable: Louisiana has been hard hit and they give us a great test case for OTHER data. Murray predicts they would need 800 ventilators (they are using 500) - BUT Murray keeps dropping his number (does he read our tweets!)

He's updated the model 4 times

ME
MD
MA
MN

So the issue is... if Murray's "reverse engineering" of deaths into hospital burdens is true... then low ACTUAL hospital burdens implies the death projections are wrong.

MS
MT
NC
ND

What's that? But he's getting deaths correct?
The whole idea of the model was to take ACTUAL deaths and project hospital needs. So he updates with actual deaths.
1) he's almost ALWAYS shallow
2) predicting next day deaths is like hitting the side of a barn.

NH
OH
OK
OR

If your state isn't listed here it's likely because they don't publish hospitalizations yet.

PA
RI
SC
TN

Look - stay safe. COVID19 is nasty. The jury is still out on if this peaks in the next few days or goes on the awful run that Murray says it will (34K dead by April 15th)

But I worry that his burden estimates are so bad

Texas
Utah
VA
VT

Over half of the deaths we are seeing are in the New York City area. By one account 10% of deaths nationwide were in one borough!?

Something happened there. I have theories... but later.

WI / WY

Maybe by April 15th the modeling will show that (yay!) all of our social distancing worked!

I hope that is the case... I have other theories. But let's get this done and hope something is left of the economy when we get to the other side.