Since the most recently available week of Nash Info Services data allowed me to really fine-tune my prediction formula, I figured I was about ready to put it to the test on amazon data and actually predict some US anime sales. So I set out to do so on 3 months’ worth of year-old releases, picking months with releases I have concrete year-one data for. Seemed pretty straightforward at the time.

…Turns out it’s never that simple. There were a couple of factors I had to consider when trying to make truly complete estimates. But with some fast and loose statistical gamesmanship, I was able to get some estimates. This time around, I have substantive reason to suspect they’re pretty good.

Continue reading →

On account of a Thanksgiving week where no anime is coming out, Novermber will end up being a pretty lean month, one currently set to have only 21 releases. I added two more to this dataset for the end of October; a pair of Lupin/Conan movies that got added to the end-of-October slate last minute. So I’ll be tracking 23 releases total this month in total. I’ll also be adding the DBZ/Marnie/Naruto movies to the special tracking list in a week when the October collection finishes up.

Continue reading →

Due to circumstances largely beyond my control I’ve had little time to work on the analysis of the amazon tracking data. The tracking is ongoing and easy but the writeups are decidedly not, and before I realized it I had 3 full weeks worth of data to expound on.

It’s not all bad – I had some time to do some stuff with the overall US charts and got a significantly better understanding of where the project as a whole stands.

Continue reading →

Part 3 of an ongoing project to investigate what correlation, if any, exists between the amazon rank for US anime releases and their actual sales.

When I first conceived of this little amazon tracking expedition, I envisioned it as a 3-4 week excursion at most. At the time I was pulling up numbers individually and inputting them into spreadsheets, and it seemed like it would be a stretch to do so for a prolonged period of time. All the more so because this would be happening in September of 2015, one of the largest months I’ve ever tracked containing a massive total of 44 releases. Add in the 22 releases on the special list, and that’s a few minutes a day more than I was used to spending on this little sub-hobby.

To make the task significantly less obtrusive, I wrote a script that does the retrievals for me at a leisurely pace of one page rank per second (so as not to appear as an attack on amazon’s servers). This has since paid exponential dividends, saving me hours over the course of a month and making it possible to continue long-term tracking in perpetuity. And as this week’s worth of data (from 9/7-9/13) shows, it will be very helpful to have those extra datapoints.

Continue reading →