(Edited) Using Amazon isn't entirely accurate to judge NPD sales. Here's why...

[binary solo]

I think it would be a brave fanboy to take a bet on a PS4 NPD win if PS4 and Xb one are within 5 places on the Amazon rankings. It has nothing to do with PS being more popular on Amazon than Xb (though that may or may not be the case because where is the objective proof of this either way?) But the fact that Amazon doesn't have the majority of market share in hardware. It's possible it may have the biggest market share, but that still may only be 15-20% among several players.

What we know, from January is that a gap as small as 9 places within the top 20, without any other significant sku's being elsewhere in the rankings, is highly likely to mean a NPD win for the top ranked console. Winning by "only" 40K in January is as much a reflection of the fact that January had low absolute sales as anything else. PS4 still beat Xb one by over 20%, and that is a significant gap. Indeed on a market share basis PS4 got a larger market share of combined sales in January than Xb one did in December Xb one was 17% ahead of PS4 in December.

When sales on Amazon are close it is likely to mean they are close at other retailers and there will be more retailers where Xb one is ahead simply because the market is not homogenous.

Saying Amazon isn't entirely accurate is statistically speaking so blindingly obvious that one wonders why there needs to be a thread pointing it out. When you take a sample of a population and use it to predict a result for the total population statistical axioms tell us that the sample might be giving you the wrong result. Therefore you work out the probability that your result will be wrong.

At present what we can reasonably say is that if the gap on Amazon is 9 or more and there aren't multiple skus of the second placed console that are charting we have 95-99% confidence that Amazon will correctly pick the winner. Though with only 1 month having a gap of 9 (the gap was 14 in December in favour of Xb one, and even bigger than that in November) we might have to lower the confidence until we get more months with a 9-10 gap correctly predicting the NPD winner. We also have Dec 2013 where the gap was only 4 places in favour of Xb one, and the NPD gap was only 45K, in a high selling month unlike January 2015, If anything that suggests Amazon could be very reliable in picking the winner even with only a small gap in the rankings. We need more months with small gaps in the rankings to be able to assess reliability. But a 9 gap month and a 4 gap month both getting it right does suggest a fairly remarkable reliability. One thing Amazon got wrong in December 2013 was that it was actually 3DS that won the month. So Amazon seems to be a bit unreliable at reading handheld hardware sales. So the limit of Amazon's reliability may be only with home console hardware.

For Feb we appear to be heading towards another month where the Amazon gap might be in the 8-12 range, and if it's under 10 it will give us another bit of information that can allow us to gain more, or less, confidence about months with sub-10 gaps.

From a gaming perspective if the Order flops badly with critics and also proves to be a one week wonder sales wise that might be sad for Sony and PS4, but it might keep things interesting from an Amazon vs NPD perspective as PS4 and Xb one might stay relatively close in the monthly rankings. On balance I prefer The Order to be successful, but an unsuccessful The Order will have a very thin silver-lining to it if you like analysing the sales.