On July 02 2012 06:10 Bromazepam wrote: While the drop is not as significant as last month's, if you sum all the viewer differences (excluding new entries) the result is still about -1900.
Also no one peaked 10k this month.
I do believe there is a minor mistake with the gathering system that the OP is using.
On June 26th Stephano peaked over 10k while he was streaming the TSL qualifiers.
OP, is there a reason this might not have been recorded?
Hm, that's interesting. I'd guess the values are taken on intervals, and there could have been a peak that didn't last long that appeared right between intervals.
Unless there have been multiple sightings of higher viewer numbers. In that case I'd be more inclined to think there's a discrepancy between the system used here and the number reported by Twitch.
On July 02 2012 06:10 Bromazepam wrote: While the drop is not as significant as last month's, if you sum all the viewer differences (excluding new entries) the result is still about -1900.
Also no one peaked 10k this month.
I do believe there is a minor mistake with the gathering system that the OP is using.
On June 26th Stephano peaked over 10k while he was streaming the TSL qualifiers.
OP, is there a reason this might not have been recorded?
Yes, definitely. My script runs every 15 minutes, so if there is a peak within those 15 minutes it won't reach my database. On top of that, I get my viewer numbers from teamliquid and not directly from the various streaming websites, making the data slightly less accurate.
I just looked at what my database has to say about this:
select date, viewers from log inner join players on players.p_id=log.p_id inner join dates on dates.d_id=log.d_id where name="Stephano" and date>"2012-06-26 00:00:00" and date<"2012-06-27 00:00:00";
On July 02 2012 06:10 Bromazepam wrote: While the drop is not as significant as last month's, if you sum all the viewer differences (excluding new entries) the result is still about -1900.
Also no one peaked 10k this month.
I do believe there is a minor mistake with the gathering system that the OP is using.
On June 26th Stephano peaked over 10k while he was streaming the TSL qualifiers.
OP, is there a reason this might not have been recorded?
Yes, definitely. My script runs every 15 minutes, so if there is a peak within those 15 minutes it won't reach my database. On top of that, I get my viewer numbers from teamliquid and not directly from the various streaming websites, making the data slightly less accurate.
I just looked at what my database has to say about this:
select date, viewers from log inner join players on players.p_id=log.p_id inner join dates on dates.d_id=log.d_id where name="Stephano" and date>"2012-06-26 00:00:00" and date<"2012-06-27 00:00:00";
The numbers are pretty close, so my guess is that the peak happened within those 15 minutes, probably right before Stephano signed off.
This is true, he turned his stream off 10 minutes later. But a peak is not an average (not saying it's the data you provided is an average but you get the idea) but a maximum and I think it should be recorded as one.
Edit: Using a script to find the maximum might work better, although I do not know how hard it might be to write.
On July 02 2012 06:10 Bromazepam wrote: While the drop is not as significant as last month's, if you sum all the viewer differences (excluding new entries) the result is still about -1900.
Also no one peaked 10k this month.
I do believe there is a minor mistake with the gathering system that the OP is using.
On June 26th Stephano peaked over 10k while he was streaming the TSL qualifiers.
OP, is there a reason this might not have been recorded?
Yes, definitely. My script runs every 15 minutes, so if there is a peak within those 15 minutes it won't reach my database. On top of that, I get my viewer numbers from teamliquid and not directly from the various streaming websites, making the data slightly less accurate.
I just looked at what my database has to say about this:
select date, viewers from log inner join players on players.p_id=log.p_id inner join dates on dates.d_id=log.d_id where name="Stephano" and date>"2012-06-26 00:00:00" and date<"2012-06-27 00:00:00";
The numbers are pretty close, so my guess is that the peak happened within those 15 minutes, probably right before Stephano signed off.
This is true, he turned his stream off 10 minutes later. But a peak is not an average but a maximum and I think it should be recorded as one.
I'd love to have a more accurate way of recording peaks, of course, but I don't know what else could be done here, other than potentially lowering the rate at which I take my data. I'm pretty sure the API's for the streaming sites don't have a "peak" value that I could grab.
A nice month for Squirtle even though he streamed so little. I know a few other Startale players stream (Virus, Sound) but they never get any decent numbers.
On July 02 2012 06:10 Bromazepam wrote: While the drop is not as significant as last month's, if you sum all the viewer differences (excluding new entries) the result is still about -1900.
Also no one peaked 10k this month.
I do believe there is a minor mistake with the gathering system that the OP is using.
On June 26th Stephano peaked over 10k while he was streaming the TSL qualifiers.
OP, is there a reason this might not have been recorded?
Yes, definitely. My script runs every 15 minutes, so if there is a peak within those 15 minutes it won't reach my database. On top of that, I get my viewer numbers from teamliquid and not directly from the various streaming websites, making the data slightly less accurate.
I just looked at what my database has to say about this:
select date, viewers from log inner join players on players.p_id=log.p_id inner join dates on dates.d_id=log.d_id where name="Stephano" and date>"2012-06-26 00:00:00" and date<"2012-06-27 00:00:00";
The numbers are pretty close, so my guess is that the peak happened within those 15 minutes, probably right before Stephano signed off.
This is true, he turned his stream off 10 minutes later. But a peak is not an average but a maximum and I think it should be recorded as one.
I'd love to have a more accurate way of recording peaks, of course, but I don't know what else could be done here, other than potentially lowering the rate at which I take my data. I'm pretty sure the API's for the streaming sites don't have a "peak" value that I could grab.
I understand. Keep up the good work I enjoy these lists.
On July 02 2012 08:21 -Kira wrote: I don't know who the hell is "Hui" but it bothers me that his name means literaly "Dick" in Polish...
iirc he is a Taiwanese player. Basically the "Destiny" for Asia if that's a good comparison. A lot of viewers, but not really too successful in the pro scene. Unless there's a better analogy lol.
On July 02 2012 13:22 j3i wrote: ForGG has been my go-to stream for the summer. I hope he gets more viewers!
Definitely mines as well! Ever since HerO stopped streaming that is.
Absolutely love this information. Thanks so much for putting it together.
Things like this give me hope for increased income for these streamers. I know many make a good living and there's sponsorship dollars, donations, and all that but I mean in terms of direct "here's money for the entertainment" cash. Bands, comedians, and the like can make a great deal in a short amount of time; these guys are putting in hundreds of hours a month (I know there's a difference between the Dave Matthews Band or Louis C.K. and Whitera, but I think in terms of hard work and enjoyment experienced by the viewer there may not truly be). The amount of time people spend doing this kind of thing professionally can be hazy and hard to quantify so placing the factual numbers on your lap is nice.
On a related note, if you're one of those views and can afford to do so, maybe give the Subscribe button a little poke!
On July 02 2012 04:38 graNite wrote: Great! Is there a way to make extra statistics for viewers of a tournament?
It depends. I do gather all the events data as well (everything that's under the "on air:" column to the right of teamliquid.net), and it's possible to get various data from this. Unfortunately, all the big, premier events (GSL, IPL, MLG) do not provide viewer data from the streams themselves. So, I cannot reasonably create a "most popular tournament" list, since the tournaments that are obviously most popular won't be included. For smaller tournaments, however, I can provide such data.
There's other caveats too, though. At times tournaments restream their events under the same event name, for instance, and there's no way for me to differentiate between the original airing and the restream.
Ok, but then make the stats you can create I would like: average viewers per NASL Day, viewer peak etc...
On July 02 2012 23:04 Koshi wrote: I would keep both tables if it isn't too much extra work for you. I love the effort you put into this, I really like these stats.
Thx Conti.
Completely agree.
The first table is more or less a popularity contest and shows who the community want to watch. It is not perfect, but will give a good idea about popularity. Popularity from viewers could easily be used as a tool for finding invitees to larger tournaments and thereby assure some advertising in that way!
The second is more of a useful indication of the best economic prospects. The higher, the V*H, the better economy for the stream itself. If you can get information about prices of 1000 viewer impressions, some indication of how ofthen the different prices occur or if it is flat-rate, an indication of how often commercials occur on the particular stream and how many actually see the commercials, the data can be used to directly indicate economic feasibility of the streaming itself. Getting all of these numbers with any kind of certainty will probably be extremely difficult, but the possibility is there.
Cutting one of them would be unfortunate since they tell such different stories!
EPT![[image loading]](http://i.imgur.com/JTBOq.png)
