Chatter Plots: Qlash Casters Invitational
Welcome to Chatter Plots, where we analyze tournaments for gameplay and army composition trends!
In this edition, we'll be looking at the Qlash Casters Invitational, a 16-man tournament featuring many of our favorite commentators and sponsored by Team Qlash and Matcherino. Competitors included 5 Protosses, 5 Terrans, 4 Zergs, and 2 Random players.
After an energetic and sometimes cheesy Ro16 and Ro8 were played online, Alex, Indy, Rotterdam, and Yogo (replacing JimRising who had to cancel last minute) converged on the Qlash house in Treviso, Italy for the semifinals. In the end, Indy walked away victorious with a 5-2 series against Yogo.
Over the course of the tournament, the casters played 79 games against each other.
Most matches were between six and twelve minutes long but they ranged from just 1 minute and 45 seconds to over 25 minutes (the two shortest matches were both failed cannon rushes).
Below: the distribution of match lengths in minutes for all games, games by race, and games by matchup/result. PvP and ZvZ games don't ever seem to run long!
For this segment of the analysis we used a player-games metric. This means that if both players made adepts, we count two uses of adepts for that game. Also, we counted all units involved in battles, which is why damaging buildings appear in the graphs.
In the visualization below, you can select units used for harassment, as part of a full-out push, or in defense against an attack of either type.
There were 44 Terran player-games in total. Most, but not quite all of them, used marines and many used reapers, tanks, and medivacs. After that, unit usage drops off a lot. Since few very long games were played, late game units were uncommon, and I can even tell you that even some of those were associated with specialized builds designed to push them out early (Wardi's ghosts or Nathanias' battlecruisers).
There were 62 Protoss player-games in total. However, only 2/3 of the games had the most common unit (the zealot) reported! We have two reasons for this. First, we probably missed some of the zealots that were made in lategame because Protoss deathballs make it really hard to see small ground units. Secondly, there were some cannon rushes and beelines for adept aggression that wouldn't involve early zealots.
This graph shows the process of Protoss Tech very well, with warp prisms, archons, phoenixes, and immortals in the 'middle' tier, and the much less common end game units like collosus, high templar, and tempests.
There were 52 Zerg player-games in total. Zerg players just about always used zerglings and queens for fights; it's even possible we missed some zerglings because they are so small, and of course all zergs make queens but we only counted the ones that engaged the enemy. Zerg midgame tech is very visible in the center of the graph, but as we head towards the bottom we see not only Tier 3 units but also rarer midgame strategies like mutalisks and swarm hosts.
Openings and Proxies
We recorded the first four nonsupply buildings made in each game (proxy pylons, however, were included). Not all Protosses made gateways or refineries at the start of a game since some were cannon rushing. All Terrans made at least a barracks and a refinery in their first four buildings, and all zergs made at least a hatchery and spawning pool so early on.
We counted 45 proxied buildings in the games. The majority of these (33) were placed by protosses; but a few terrans and even a couple of Zergs proxied buildings. Unsuprisingly, Pylons were the most proxied structure but there were also 5 games with proxy barracks, 5 games with cannon rushes and 3 where shield batteries supported attacks. The one 'proxied' command center was actually a hidden base.
We took note of a few notable events in the games as well:
Good Hold - we just thought this game had a really good hold
Group D, Set 3 - Rotterdam vs. Indy, Game 1
Baneling Bomb - banelings actually blow up some units while burrowed
Semifinals - Yogo vs. Rotterdam, Game 5
Trapped Unit - feed your schaudenfreude and tell yourself you would never rally a unit where it couldn't get out (lies)
Group D, Set 1 - Rotterdam vs. Maynarde, Game 1
Group D, Set 3 - Rotterdam vs. Indy, Game 1
Banter - Players telling jokes in chat
Group A, Set 4 - Nathanias vs. ElementFini, Game 1
Group A, Set 4 - Nathanias vs. ElementFini, Game 2
Group B, Set 1 - JimRising vs. Feardragon, Game 1
Group B, Set 4 - Wardi vs. Feardragon, Game 1
Group D, Set 1 - Rotterdam vs. Maynarde, Game 2
Group D, Set 4 - Lowko vs. Maynarde, Game 2 (just a little bit)
What is this? - in which a short but bizarre scrap occurs between mostly scvs and tanks
Ro8, Match 3 - Alex vs. Nathanias, Game 3
Investigate the data yourself in this interactive data visualization. Look at game length by player, units used by player, and more!
If you want to really dig deep, the raw data are available here. Please note that since this tournament was a test run for the data collection method, the data do have some gaps and errors; and the format changes slightly for data taken during the latter portion of the tournament.