I've read through this Q&A:
My understanding from reading that Q&A is that in the early phase of an outbreak, there are two naive ways to estimate what the fatality rate will turn out to be when the epidemic is over. One tends to underestimate, Nd/Nc, the other tends to over estimate, Nd/(Nd+Nr).
Currently, Nd/Nc is 2%, and Nd/(Nd+Nr) is 33%
Supposedly there are advanced (stat/math/modeling) techniques that are much better estimators during the early phase of an outbreak.
However, the currently bantered about fatality rate (2-3%) matches the naive estimator, Nd/Nc. Since Nd/Nc is a known underestimator, the figure (2-3%) is not likely to have been derived by one of the advance, minimally biased estimators.
Since both of the naive estimators will yield the same result when the epidemic is over, I suppose one could track both, and extrapolate the curves to where they will meet. But I'm going to guess that that would be too naive as well.
Hence my questions, what is a sophisticated estimate of the 2019-nCoV fatality rate ? and how is it obtained?