[109 / 10 / ?]
Quoted By: >>6027857 >>6028703 >>6028800 >>6028821 >>6029066 >>6029242 >>6029997 >>6030645 >>6030944 >>6033001 >>6033048 >>6033337 >>6033571 >>6033958 >>6066428
See the picture for the numbers. Just look at the big table on top for the quickest results; or the one on the bottom for individual ranks..
If you want a quick summary, most categories of vtubers are growing in June, when compared to may, which I assume is because of the return of restrictions due to covid in many countries in the world, including of course Japan - so people had more time to spend on watching streamers than usual. I`d expect them to fall back in the following months, as those restrictions are likely to end once again.
The top JP vtubers in general have received a great boost (yes, Coco and Lulu grads do impact it, but not by so much: if I remove them, Hololive goes from 26% to 18%; Nijisanji from 10% to 7%); while other jp categories stagnated; NijiEN`s fallen after the debut buffs; HoloEN slightly growing due to playing good games and collabs; and VShojo still growing as usual, although they seem to be decelerating a bit if compared to previous months.
I’m using the metric (vod) views/time because it seems to be the best metric we can use for comparisons. There’s no way to calculate direct hours watched without having access to data analytics from inside each channel owner or bullshitting it; simply counting total views doesn’t take into account how much did they stream to get there (and they usually count small skits and music videos as well); and live viewers (on youtube) are very buggy and cut off a significant portion of that real number (for example, comparison between Miko’s twitch and youtube minecraft streams seem to indicate that 35% of her yt live viewers numbers are hidden [read the OP or the whole thread >>2253454 for anons discussion on the possible reasons for it) (not to mention that twitch max live viewers are constantly influenced by twitch raids coming from other channels).
Reading and trying to compare all those ratios for each hour length would have been a pain for any of you, so that’s why I’ve already gone ahead and made the ranking tables, with ranks for inside each time length. The longer the stream time, the fewer vtubers that had any videos available to calculate. Never compare general view/time ratios of streamers who have very different stream duration averages - that’s why I painted those general rankings on right in grey. Only compare data (or ranks) from the same time length.
For combining those ranges from each time length, i inverted each position into a fraction (position/total), calculated the average of all available time-length fractions for each streamer, and then compared those resulting numbers: the smaller your fraction is, it means the higher yours positions were, so you rank higher. You can see a practical example (from may) here: >>4423534.
Also, I had a cap of 43 videos per streamer, so it might miss a few videos for very few chuubas, but it’s good enough to get their ratios already. Video types like short skits, short amvs/covers/original songs, watchalongs, members only, or private were not included for the calculations. Videos analyzed for 'this month' are from the time period of 28-05-21 to 30-06-21 (~11pm jst). (with exception for Coco, who I also took her last stream from july 1st to have her complete).
You can see the biggest individual growth fluctuations with those vtubers who had some huge special stream (like 3ds, costumes, returns, tournaments). They can be big rises if the special vids happened on this month; or big falls if the special vids occurred on last month. On those cases you can see % change of one specific time length way superior than the rest for that vtuber.
Collecting all this data was a whole lot of work, so no promises that I’ll be doing this again, or as big of a list. I don’t think I missed anyone too big, I left out the smaller nijisanjis, or holostars or whoever else because I don’t have the time to include them all - but I did include a few smaller ones just out of curiosity. And no "now im suddenly a vtuber" trendhoppers were included either. If you think I left out someone relevant you can try convincing me to do their numbers as well, although if they are too small I’ll just tell you to fuck off.
If you want a quick summary, most categories of vtubers are growing in June, when compared to may, which I assume is because of the return of restrictions due to covid in many countries in the world, including of course Japan - so people had more time to spend on watching streamers than usual. I`d expect them to fall back in the following months, as those restrictions are likely to end once again.
The top JP vtubers in general have received a great boost (yes, Coco and Lulu grads do impact it, but not by so much: if I remove them, Hololive goes from 26% to 18%; Nijisanji from 10% to 7%); while other jp categories stagnated; NijiEN`s fallen after the debut buffs; HoloEN slightly growing due to playing good games and collabs; and VShojo still growing as usual, although they seem to be decelerating a bit if compared to previous months.
I’m using the metric (vod) views/time because it seems to be the best metric we can use for comparisons. There’s no way to calculate direct hours watched without having access to data analytics from inside each channel owner or bullshitting it; simply counting total views doesn’t take into account how much did they stream to get there (and they usually count small skits and music videos as well); and live viewers (on youtube) are very buggy and cut off a significant portion of that real number (for example, comparison between Miko’s twitch and youtube minecraft streams seem to indicate that 35% of her yt live viewers numbers are hidden [read the OP or the whole thread >>2253454 for anons discussion on the possible reasons for it) (not to mention that twitch max live viewers are constantly influenced by twitch raids coming from other channels).
Reading and trying to compare all those ratios for each hour length would have been a pain for any of you, so that’s why I’ve already gone ahead and made the ranking tables, with ranks for inside each time length. The longer the stream time, the fewer vtubers that had any videos available to calculate. Never compare general view/time ratios of streamers who have very different stream duration averages - that’s why I painted those general rankings on right in grey. Only compare data (or ranks) from the same time length.
For combining those ranges from each time length, i inverted each position into a fraction (position/total), calculated the average of all available time-length fractions for each streamer, and then compared those resulting numbers: the smaller your fraction is, it means the higher yours positions were, so you rank higher. You can see a practical example (from may) here: >>4423534.
Also, I had a cap of 43 videos per streamer, so it might miss a few videos for very few chuubas, but it’s good enough to get their ratios already. Video types like short skits, short amvs/covers/original songs, watchalongs, members only, or private were not included for the calculations. Videos analyzed for 'this month' are from the time period of 28-05-21 to 30-06-21 (~11pm jst). (with exception for Coco, who I also took her last stream from july 1st to have her complete).
You can see the biggest individual growth fluctuations with those vtubers who had some huge special stream (like 3ds, costumes, returns, tournaments). They can be big rises if the special vids happened on this month; or big falls if the special vids occurred on last month. On those cases you can see % change of one specific time length way superior than the rest for that vtuber.
Collecting all this data was a whole lot of work, so no promises that I’ll be doing this again, or as big of a list. I don’t think I missed anyone too big, I left out the smaller nijisanjis, or holostars or whoever else because I don’t have the time to include them all - but I did include a few smaller ones just out of curiosity. And no "now im suddenly a vtuber" trendhoppers were included either. If you think I left out someone relevant you can try convincing me to do their numbers as well, although if they are too small I’ll just tell you to fuck off.
