By using this site, you agree to our Privacy Policy and our Terms of Use. Close
abronn627 on 21 November 2018
Edit Thread
globalisateur said:
Symbios63 said:

The idea of this thread is to use the online best selling list of main US retailers to try and build a formula allowing us to predict the outcome of incoming NPD.

 

After a few month, here is what we do :

- The retailers tracked are :  - Amazon, - Best Buy, - Gamestop

- The basic idea is to allocate points to hardware depending on their ranking on the 3 retailers online best selling list : Rank 1 = 210 points, Rank 2 = 205 points, ... , Rank 21 = 110 points, Rank 22 = 108 points, ... , Rank 51 = 50 points, Rank 52 = 49 points, ... , Rank 100 = 1 point. The amount of points for each rank has been determined empirically over a few month.

- Retailers are weighted in the formula as follow : Overall points = Gamestop x 1 + Best Buy x 0.2 + Amazon x 0.8. Again, this was determined empirically.

- Concerning Gamestop, we look at Top 200 instead of Top 100 because the amount of figurines and T-shirt can make it difficult to find hardwares in the top 100. However, if an item is ranked 124, we consider it to be ranked 62. If it is ranked 10, we consider it to be ranked 5 etc... Then the point system applies normally.

 

After the last NPD I think we need to change the methodology, like :

- Put way more weight in the first rankings. Rank 1 should get more way more points. All those Xbox bundles were a problem this month.

- Finally reduce the Amazon weight. It's time (I already talked about that previously). We should weight the retailers using their actual known sales.

If you have time you should try to modify those variables and retrospectively apply them for october until you get the actual results... or at least the correct ranking !

I would say that we don’t necessarily need to adjust the value of each of ones because Gamestop’s way of tracking is not based on actual sales but viewing trend on their website, having the 500M PS4 on their rankings even if it was limited and sold out is pointing to that outcome.

What I would like to suggest is to track preorder separately and do an average based on the numbers of days they where available for preorder, then add the points to the release day totals. This will cause the spike you should expect on the chart without affecting the monthly average.

I think we should also make based on the numbers of skus available, this should prevent false sales spikes like we saw with the Xbox One.