Estimations work when dealing with symmetrical distributions. You might be wrong either way. But these distributions are fundamentally not symmetrical! Also, you should compare deaths with number of cases a couple of days ago (like 7), since that’s the time between diagnosis and death.