For those looking toward a more reasonable measurement and have the patience to wait daily (asking a lot I know) consider this. If we use Seece's same method and apply a degrading sales amount considering we have 9 months of accurate data and apply it to the 2014 amazon bestsellers list we can get this:
Degrading rank starting at 2 million units sold for #1 to 0 units sold for 100 this applies 20,000 units per spot and would give us the following sales through 10 months of 2014 on amazon for console sales with a good barometer being 9 months of NPD sales
PS4 #4 and #28 = 1.6 and 1.46 for a total of 3.06 million
One #51 and #52 and #98 = 1 and .98 and .02 or 2 million
very similar to our known data acquired via NPD
Granted one must assume with these figures that the Titanfall bundle outsold the base One SKU by double and that the destiny bundle sold nearly as much the entire year as teh base PS4 model. Using this method we see roughly the same sales gap as tracked by NPD, but somewhat lopsided toward the top numbers on the rankings.
This means it makes more sense to assume that each spot on the yearly seller list is closer to 20,000 units sold for each movement. Again, it is imperfect but it at least is reasonable, actually, comparible to known numbers, and decreases volitility and error for arbitrary hourly rankings with no concept of quantity.
