The vS Data Reaper’s Radar: How Visual Analysis Can Track the Evolving Hearthstone Meta
With One Night in Karazhan’s release, the Metagame is sure to shake up over the next few weeks, as new archetypes form while other archetypes may falter. How is the Data Reaper able to track these changes and properly define archetypes by the cards that are being played? How can the Data Reaper determine whether an archetype has been fully refined with a consistent build, or is still in the process of experimentation? Do we rely on word of mouth, playing the game or internet decklists? To some extent, they help, but we do not rely on them.
The vS Data Reaper is actually capable of seeing the Metagame map, and by seeing it, we literally mean observing it with our very own eyes. Early on in the Data Reaper’s life, we came across Sentenza, who developed a tool to visualize card selections and interactions in a collection of decks. Together, we came up with an idea: applying this tool to the Data Reaper raw data in order to track the emergence, evolution, and downfall of archetypes in the Metagame.
How does it work?
We scan the database of games of a particular week, and proceed to run it through a code. The product is a chart; full of circles and links between them. Each circle on the chart is a card that an opponent has played. The circle size is an indication of the number of opponents that have played this card. Two cards are linked if they have been played by the same opponent. These links operate like springs: the larger the number of opponents that have played two cards together, the stronger the spring tension, and the closer the cards are on the chart.
Conversely, cards that have no link between them tend to repel each other. Applied to our data, these conflicting forces result in a visualization where core class cards shared by most decks (e.g. [Fiery War Axe], [Execute]) have a central location, while cards that characterize a specific archetype (e.g. [Alexstrasza’s Champion], [Blackwing Corruptor]) are clustered in a peripheral area.
In such a large number of games as our data contains, it looks almost as if every possible pair of tech cards have been played at least once, so that the visualization tends to be cluttered with a lot of irrelevant information. To reduce this noise, we exclude from the charts the cards and links that are less frequent by some threshold (namely 5% of games for cards and 1% for links).
So, how does it look?
We’ll present you with a great story of an evolving class during the WoToG timeline; one that has also boasted multiple viable archetypes during that timeline, the Warrior class of course!
This is the chart for the database of the first week of May, very early on after Standard Format’s launch.
Warrior Class Radar – May 1 – May 7, 2016
So what do we see here? In the very middle, we can spot the core Warrior cards that are used by the large majority of Warrior decks: Fiery War Axe, Slam, Execute etc. But on each side, we see two large clusters. The first one is the Control Warrior cluster, showing many cards that are primarily used by slow Warrior decks. What you can also spot is two layers within the cluster: The inner layer contains the core Control Warrior cards, while the outer layer contains the C’Thun Warrior shell. Since both Warrior archetypes utilize the inner layer’s cards (Brawl, Shield Slam etc.), it’s closer to the center of the map, while the other layer is unique to only the C’Thun Warrior archetype.
On the other side, Tempo Warrior is present. Back then, it was the most popular Warrior archetype in the game. We can also spot its cousin, Patron Warrior, within that cluster, since both decks have a very similar shell.
A bit further off, Pirate Warrior is spotted. Since Pirate Warrior is the most different archetype of the class, not playing core Warrior cards such as Execute and Ravaging Ghoul, it is the furthest cluster away from the center. However, it’s closer to Tempo Warrior than it is to Control Warrior, since their builds have a few cards in common (Berserker, Kor’kron Elite). Pretty cool, right?
Let’s fast forward a month, to the first week of June.
Warrior Class Radar – June 1 – June 7, 2016
The Control/C’Thun cluster is still visible, so are the Tempo and Pirate clusters. Tempo Warrior has begun to slightly drop off in popularity at this point, so it looks smaller. However, a new archetype has begun to form: Dragon Warrior, with Blackwing Corruptor, Twilight Guardian and Azure Drake surpassing the threshold of detection. We can see its proximity to Pirate Warrior and Tempo Warrior, and the cards shared by all three archetypes are spotted just between the three clusters. At this point, we had an idea of the incredible potential of Dragon Warrior based on its performance against the field, and were curious to see when/if the archetype explodes.
It happened a few weeks later. The Americas Spring Championship had three players bringing it to the event, and when Blizzard publicly posted the decklists just a few days before, the archetype erupted on ladder.
Let’s look at the chart for the final week of June, after the tournament concluded with Cydonia’s victory.
Warrior Class Radar – June 22 – June 30, 2016
We can see the Dragon Warrior cluster growing in size and rivaling Tempo Warrior, an archetype that is shrinking as more players switch over, realizing the advantage the Dragon archetype has over it in its performance against the field. Pirate Warrior is also beginning to fade away, while the Control/C’Thun Warrior cluster remains constant throughout.
Since then, Dragon Warrior continued to grow at an impressive pace, becoming the most popular archetype, not just within the Warrior class, but in the entire game. We also had another Warrior archetype, the Worgen OTK, emerge. If we look at the very last chart we produced for the final week before Karazhan’s release, we see a dramatically different picture.
Warrior Class Radar – August 3 – August 9, 2016
The Dragon Warrior cluster is now massive and rivals the Control/C’Thun cluster in size. Worgen Warrior’s emergence is also spotted, completing the current trifecta. Tempo Warrior has been swallowed nearly whole, showing some traces of presence near Dragon Warrior’s location, while Pirate Warrior has faded away under the threshold.
So we’ve shown you that the tool certainly looks neat, and provides us, the Data Reaper team, with a fantastic way to track the Metagame, spot emergence of archetypes and sub-archetypes, as well as constantly fine-tune our definition algorithms as the Metagame changes (how we determine an opponent’s deck choice by the cards that he plays).
However, this tool can potentially provide a lot of information to the community as well. We’re not limited to just mapping a class as whole. We’re also able to zoom in on a specific archetype and spot the mini-Meta within it. To give you an example of how the visual tool works on an archetype and what kind of information it provides, let’s look at Zoo Warlock data from last week, just before the adventure’s release:
Zoo Warlock Radar – August 3 – August 9, 2016
Just like core class cards will be located in the middle of a class’ map, core archetype cards will be located in the middle of an archetype’s map. We can clearly see it here.
So what do we see on the outskirts? Cards that are used by the archetype, but to a lesser degree, meaning: tech choices, or cards that are used by some specific builds. We also see Dark Peddler’s shenanigans: Corruption, Reliquary Seeker and Blood Imp are relatively popular Discover choices though they are never included in a Zoo decklist. We can also see how Doomguard and Leeroy Jenkins are on two different sides of the main cluster, indicating they are almost always mutually exclusive.
Can you spot another sub-archetype on this map? We can. The cluster of Lance Carrier, Wrathguard and Young Priestess is a fringe Zoo build we featured in the past on the Data Reaper Report. It was used by some notable players, such as Xixo, in high level tournament play. It cuts token synergy cards: Knife Juggler, Darkshire Councilman and Forbidden Ritual, which makes it quite different than a normal Zoo build, and as a result, it resides further away from the center.
We believe that these maps can provide a lot of information to the player that can give him a competitive edge on ladder, the most obvious one is the ability to spot sub-archetypes within a class, and predict potential cards the opponent may have by other cards that are being played, including tech choices. It also provides information on the different possible builds that are out there, and the level of variance in card selection each archetype has.
To conclude, we have plans for the not-too-distant future to incorporate the radar into future content on the website, which will go hand-in-hand with the Data Reaper’s Report.
We appreciate all of the community’s support for our content. It does not go unnoticed, and we hope that this additional content will continue to steer us to our goal: Become the #1 source for all things related to the Hearthstone Meta, and provide the community with data-driven information that it has never been able to access.
Our team is looking for people who are interested in getting involved in writing articles such as this one. We are happy to provide qualified individuals with data so that they can analyze it on their own, and then develop an article that we can publish together. Our goal is to expand the community’s knowledge about Hearthstone. We know that there are plenty of qualified individuals out there. If you are interested, please contact us here.
Our Data Reaper Project, including the Data Reaper Live (Beta) now has over 1,800 contributors. Without them, analysis such as this would not be possible, so we’d like to thank all of our contributors for helping us on this project. If you have not done so already, you can sign up with your Track-O-Bot information here:
This article was put together by: