This page is very multimedia heavy. You might want to give it some time to load.


I cant recall the exact year when I first time watched eSports, but I remember spectating Halo 3 MLG Pro Circuit broadcasted from Dallas. All of the teams were from United States. Prize pool was $56K.

Few years after that I saw Dota 2 premiered at the first International. Teams around the world were competing for prize pool of $1M, which I like to think as a turning point in terms of prize pools of big events. I had heard of six figure prize pools in Dreamhack, but getting into the two comma club really showed that Valve was in for real with Dota 2.

Now four years have elapsed. The prize pool of The International 5 settled at around $18.4M, once again gathering the best teams in the world to play Dota 2 exclusively at Key Arena, Seattle Center, USA. Worth notifying is that from the $18.4M, circa $16.8M was crowd-funded by the community of Dota 2 by offering them a virtual event ticket for $9.99. From the sales of this "Compendium", 25% went to prize pool and the rest into Valve's pockets.

In return for their bucks the Compendium owners are not only able to watch the stream (which is broadcasted for free anyway), but they are also eligible for event exclusive in-game swag such as hats and visual effects, which they can wield on characters to show their support towards The International. Once the event has started, there is no way to receive these items anymore, which will make you look like a special snowflake in a year or two.

Fancy items or not, eSports is growing.

The International 5 in numbers

Roughly a year ago I wrote a script, which talked Twitch chat in loud. I was supposed to stream the audio track online during The International 4, but my 1MB ADSL upload bandwidth shut down my plans.

I ultimately wrote a blog post describing how I did everything, which received a good amount of hits and some feedback. The feedback taught me Twitch's chat is actually just an IRC server broadcasted on website.

I later updated my blog post to show you can leverage the IRC server avoid visiting Twitch's website to read the chat.

Inspired by the project and forced by this year's lack of time, I decided to take on a premise that Twitch's chat could highlight me the best parts of The International 5 while I was sleeping. It proved to work pretty well. I ended up collecting most of the chat data during The International 5, which this blog post is all about. So hang tight and buckle your belts, big data is coming for you with the fierce force of 400'000 Kappas.

Without any further talks, I'll get into graphs.

Wild card

The data is clipped because there were unexpected connection problems with Twitch IRC servers and the bot ultimately crashed close to midnight when the stream wen't down.

The Stage Just Before Doors Open on Day 1

Group stages

Group stages were a lot more stable in regards of data collection. Although, on third day the bot crashed because of memory leak. The memory leak was patched the next day.

MPV Phoenix vs. Vici Gaming

Main event

During the main event data logging worked as it should, except for two 30 minute blackouts during the finals. Being aware of the bot's bad manners, I had pre-emptively configured it to make persistent backup logs. In the end, we got a nice graph of all dates! Success!

Aegis of Champions
The Arena During the Grand Final

Some bullet points from stream graphs: