As pointed out, the graphs in thedvdwars.com show a history of stock levels at Amazon. While tracking titles in the aggregate may lead to invalid results, tracking individual titles over a period when it looks like there was no stock replenishment might give some idea of the relationship between sales rank and sales numbers.
With this in mind I looked at the stock level history for the top 15 blu-ray and top 15 HD titles that are now in stock. The attached pdf has the data for the 30 titles. It shows the period I tracked (most recent period without a stock replenishment, i.e. no stock increase showing during the period), the average daily sell rate, and the current (as of a few hours ago) sales rank. On average there are 5-6 days of history for each title. I did not include "The Departed" even though it is now in stock on the HD side, this is both because less than a full day has elapsed since it went in stock (presumably fewer people buy a title at 2 AM that at noon) and because the inclusion of a single title probably selling more than 250 copies over the next 24 hours would significantly distort the results.
Conclusions drawn from data in the pdf
1 For in stock titles blue-ray outsold HD by 56% to 44%, based on total sales in tyhe data I looked at of about 2000.
2. Average blu-ray rank was 1323 versus 2217 for HD, a much bigger difference than the difference in sales.
3. Sales are about 25 a day for a rank of 600, and fall off to about five at rank 3000.
4. Some titles seem to have very high sales relative to their current rank, for example Rambo, and Superman returns (HD version).
5. For stocked titles Amazon keeps about 18 days of inventory on average. This number is distorted by Saw 3 which has a very high inventory level relative to sales.
When looking at the data I did notice two anomalies, both on the HD side. For Batman Begins, and for the Mummy, there were large increases in stock levels (200 copies) followed by a quick drop to almost the same starting level within a few hours or days. Unless someone is buying 200 copies of something the most likely cause would be input error followed by correction. One needs to be careful when looking at the raw data.
I will be looking for relationships between sales rank, sales, and stock levels. The amazon sales rank clearly depends on more than just a six day average, it is highly weighted to recent (last day or so) sales. Over the next week I hope to add a few more titles as they come in stock, especially those in the under 100 region. I might then be able to fit a curve relating sales to rank to get a quick idea of total sales without having to tediously check the in stock levels.
AmazonSales.pdf 33.587890625k . file