The CFR. Yes that overstates the risk because you don't know how many asymptomatic cases there are. But it does not overstate it by several orders of magnitude, which is what would be required for me to be wrong.