Not really. Averaging DOES work for the yearly ratio of sales, it just doesn't work for how many sales there are.
On Steam the game has 4.8 million sales. On PS4 it should have about 6.54. But The Witcher 3 is heavily advertised on GOG, a store owned by CD Projekt Red. So I wouldn't be surprised if the difference between sales was only a million, maybe even less.
But except it doesnt work like that.
Lets take a look at this made up scenario. Witcher 3 had 90% of its lifetime sales in 2015, with the remaining 10% in 2016.
Lets say the split was 75% ps4 and 25% pc in 2015, while in 2016 was 10% ps4 and 90% pc. The way you would calculate it, PC would have 57.5% of sales, and PS4 would have 42.5%.
But that would be wrong. In reality, PS4 would have 68.5% of sales, and PC would have 31.5%. Since the gap in 2015 in terms of units sold was way in favor of PS4, and the 2016 gap between PC and PS4 is much smaller in terms of units sold, because it had very little sales in 2016 when compared to 2015. When the amounts for each year are different, just adding up the percentages and dividing doesnt work because the amounts each year arent equal.
In this scenario, percentage doesnt tell the whole story, since the gap in units sold between ps4 and pc is much smaller in 2016, even though the percentages are heavily in favor of PC, because 2016 only counted for a small fraction.
Bet with Intrinsic:
The Switch will outsell 3DS (based on VGchartz numbers), according to me, while Intrinsic thinks the opposite will hold true. One month avatar control for the loser's avatar.