Spike Defused
Decoding Valorant professional esports with numbers.
Valorant is one of the newest additions to the series of Tactical First Person shooters we have seen in the esports industry. Riot as a game producer has managed to create an industry around it and keep providing the audience with huge tournaments across the globe, including Challengers, Masters, and the biggest of all Champions. The main goal is to create the story of Valorant Esports in terms of the teams that have dominated and the meta changes that the game has gone through across two years of international events.
With Tactical FPS games comes a load of data points that can be tracked down for each round that is played. Kills, assists, deaths, and average combat score is usually just the surface level that we get to see in the games. But adding to that there are a lot of situations and data points that can be added to generate much more insight into the game and how the teams perform on the international stage. For this project, I am going to visualize the data for all matches from June 2020 to September 2022 that happened on the international stage where the best teams from various regions compete.
The Process
Valorant data is difficult to retrieve as the game is just 2 years old and the RIOT API is limited in terms of the data it provides for open source work. Run it Back (rib.gg) discord bot solves this problem as it makes the data more accessible for people to use for statistical work.
Literature Review
The literature review was aimed to understand the current state of data in not just the esports industry but also conventional sports which I can take inspiration from.
Esports and Sports analytics in general has come a long way from just helping out support staff to better understand the game and prepare better for upcoming events. It has been used in a meticulous way across media outlets and production for a better viewing experience for people who are indulged in the game. The space of creating simple on-the-fly visualizations for Valorant has just begun and the research shows how different researchers have used different tools to create visuals using different parameters and criteria that help them build a storyline. I focused on how different storylines were created around the visualizations and wanted to emulate the same with this project. Covering the short history of Valorant Esports became a priority from this point on.
Sketch Design
Showcasing the ever changing Valorant Agent Meta: Valorant by December 2022 has about 20 different agents or characters that you can play with. All characters have their own abilities and powers. By just looking at the agents you can usually understand what is going to be the style of Valorant they are going to play. The goal was to understand what agents are getting picked at the competitive level and also who are they getting paired with. I felt that Chord diagram was the best possible solution to this problem. It helps to showcase both the pick rate of each agent and their relationship with other agents.
Team Performance in a tournament: I focused on finding the winning formula for each team participating in these tournaments. I wanted to configure different paramers that would help a team win in the game. To visualize this I feel like a spider chart would be best to showcase all the parameters on the same scale.
Beta Release
As explained initial the chord diagram was going to be crucial for understanding the meta of each tournament and how agents are getting picked. I created a basic chord diagram that helped me visualize for the three tournaments for 2022. I wanted to see if it actually helps to visualize the different picks rates of each agent and if I am able to clearly see different partnerships between agents. Personally, I was able to see visually agents getting introduced into the game also see the meta shifting within just these three tournaments.
I am working on getting these heatmaps for champion teams and by the final release should be able to get a final heatmap for all the six teams who have won the different championships. (Source: rib.gg)
Adding to this I have finalized the parameters that are going to be used for analyzing each team participating in different tournaments. I will find the percentage value for each of these so that I can put them on the same spider chart. I have mentioned the parameters below:
- Gun Round Win%: These are the rounds where both teams have the same amount of money where they can buy heavy armory. ($$$$ v $$$$)
- Economy Buy round win%: These are the rounds where your team has a cheaper buy in comparison to the opponent’s heavy buy. ($ v $$$$)
- Anti-economy buy round win%: Where the team has a heavier buy in comparison to the opponents ($$$$ v $)
- Bonus round win%: These are round where you carry forward your guns after winnin 2 consecutive rounds at the start of a half.
- Anti-bonus round win%: Rounds where you have a losing streak of two rounds at the start of a half and you are facing a broken buy.
- Pistol round win%: Win rate on pistol rounds
- Trade%: These the deaths where your teammate in under 3 seconds kills the opponent which mean that this particular kill is traded.
- First Kill %: Getting the first kill in particular round
- 4v5 win%: Winning a round when your team has the first death.
- 5v4 win%: Winning a round when your team has the first kill.
- Plant/Defuse %: The spike plan or defuse leads to objective wins in the game and therefore this parameter addresses that.
- Post-plant win %: These are the wins where you win post the skike has been planted both on attack and defense.
User Evaluations
With user evaluations I wanted to go through varied users who had a basic understanding of the game rather than just general audience. So I went with three different types of users who would can closely related to the project: an analyst fro a professional team, a broadcast talent who has worked at different tournaments and an upcoming young player who is playing in weakly tournaments with a team to reach higher levels of competitive Valorant.
I observed how they are using it while asking them questions to find insights from the visualizations showcased on the website. I am noting down some of the key observations I made for each of the user.
Analyst: I feel like a lot of the feedback I received was out from prior experience with working and communicating with coaches and players about data in Valorant. Additionally, it felt like he had already seen some of this data before because he had a clear understanding of what was there on the website. It took him seconds to start giving feedback on some of the visualizations. At most points with the visualization, he wants to pair up stuff so that there are more inferences that can be brought out of them. In terms of this tool settling for preparing for a team at a tournament looks unlikely as he said there are a lot of different things they look into when they anti-strategize against a team.
Broadcast Talent: I think the platform better suits broadcast talent who want a quick summary of things happening in Pro Valorant. They took some time in playing around with the chord diagram and figuring out what it actually does. Once they had a clue of what a chord diagram is they were able to find quick and basic inferences from the visualization. The goal of the spider charts was always a good idea to give a simple tool for broadcasters to quickly have an insight into how a team plays. So I feel that goal was achieved during this testing phase.
Upcoming Young player: He got an understanding of the visualization and how to read them but didn’t actually call out any inferences from them. There was just a lot of playing around in terms of seeing the number of times each agent was paired with some other agent. I tried to find a pattern in that but couldn’t. Further visualizations were just skimmed through and not many inferences were taken from those.
Final Release
The final release consisted of the final website with all the changes made based on the feedback we got from our peers. Due to the problems I faced while building the charts with D3 and Charts, I tried out a ton of different options but finally created them using Flourish studio. You can now visit the website here.
Future Scope
- I would like to add a paragraph of text explaining each visualization so that there is more context to it.
- Adding a table view to the website for the spider chart is a pretty good idea that I can implement onto the website.
- Also, I would like to think about how to make the UI a little more fun so that non-technical users can enjoy the experience on the website as well as make some inferences out of them.
- The future scope can be to pair the chord diagram with different stats and have more details for a particular meta.
