Unlocking the Power of Esports Statistics
Which team is smashing critical objectives in their matches? Which player tops the charts in damage during certain phases of play? Which team scores the highest or lowest average kills on which maps? — statistics are the cornerstone of context in esports and gaming, with the ability to tell deep and exciting stories.
Whether you’re a dedicated fan looking to uncover the next meta build, a professional team seeking actionable insights, or a game developer aiming to optimize your title — the answer lies in understanding and leveraging detailed statistics.
In this article, I will delve into one of our products, the GRID Stats Feed, discussing the team’s design philosophy, challenges we faced in development, and product features to help you unlock the power of statistics using in-game data.
Problem Statement
In the current landscape, users of esports data (and, in this specific case, esports statistics) face the following challenges:
- Quality of data powering up the statistics: Many data sources are unofficially obtained (i.e., scraped) and fragmented, leading to inconsistent and inaccurate stats. The digital nature of games offers unparalleled data granularity with the one real source of truth — the game server. Due to this, however, the potential of official data can only be realized through specialized technology.
- Market fragmentation: The existing stats feeds are built mainly by fans for fans and focused on a specific game title. These are often passion projects built in a format preferred by the creator behind it. Extensive title-by-title integrations are required to use these stats for any commercial project, increasing the cost-to-value ratio for data consumers. Users need a solution that allows for the seamless integration of various titles into their existing products.
These issues hinder users’ ability to fully leverage esports statistics, leading to missed opportunities for fan engagement, additional insights, and performance optimization.
The GRID Title-Agnostic Infrastructure
At GRID, we source, process, and externalize thousands of real-time data points directly from the game server every second. While an esports series (match) is in progress, we use these data points to continuously build and maintain the series state, an accurate representation (snapshot) of everything happening in the series. Once the series is over, the end state (final outcome) is stored in our database and becomes available as historical data (as opposed to live data).
The stats feed is an additional layer built on these stored end states. It empowers users to aggregate data to extract valuable insights easily, and it is exposed via a GraphQL API to allow for an efficient and flexible way to interact.
GRID’s series/end state — — and consequently the stats feed — have a few key characteristics that effectively address the key challenges described above in the problem statement section.
- Series are exceptionally granular. GRID’s series state supports hundreds of data points, from hero and map information to damage and kills to objectives and economy values. You can see the full list on our documentation page (you’ll need to log in to the GRID Portal to access it).
- The framework is Game-Title Agnostic. Every title’s data is represented through a common set of objects and mechanisms. This allows us to retain each game’s depth and uniqueness while empowering users to develop and operate products that can seamlessly integrate across titles. Read more examples of mechanisms on the GRID Knowledge Hub.
Each data point from the end state can be processed to produce statistics. The available operations and associated statistics depend on the type of data point.
For instance, int fields can produce stats about the total, sum, min, max, avg, and ratePerMinute. Examples of int fields include kills, deaths, damage, etc.
Boolean fields can produce stats on the total count, percentage, and streak (min/max/current). An example of a boolean field would be the outcome (win or loss) of a series, game, or round.
Another interesting stat example is based on string fields and the completion of objectives.
This approach allows GRID to immediately make new titles available to the stats feed without additional development work. As a result, all titles using the GRID Data Platform are automatically supported, including Counter-Strike, Dota2, League of Legends, VALORANT, Rainbow 6 Siege, and PUBG.
The other key component of our stats feed is the available filters, which allow users to define and operate on custom data subsets. Users can select data for a specific team, player, tournament, or time period. Certain filters can also be combined for even more granular control.
Get Domi’s stories in your inbox
Join Medium for free to get updates from this writer.
Our documentation page contains a complete list of all the available statistics and filters and a “playground” section. A GRID Data Portal Account is required to access the documentation; to request free access, apply for GRID Open Access.
A Practical Example
Let’s examine a recent League of Legends tournament, the 2024 Mid-Season Invitational, and see some statistics we can generate using the stats feed.
Gen.G ended up winning against Bilibili Gaming in the finals. Please note that Bilibili Gaming played more series and games overall, but the average values can still provide a clear picture.
This is what the win, loss, kills, and deaths look like for Gen.G and Bilibili Gaming during the playoff stage:
Diving into something more detailed, we can see Gen.G having a superior early game resulting in a higher average number of turret plates destroyed per game,
Gen.G then transitioned into a more decisive late game, which is depicted in the average number of baron kills.
A lot of Gen.G’s dominance was down to their mid (Chovy) and top (Peyz) lanes, who both ended up with the highest K/D/A in the tournament, with Peyz scoring an impressive 28 kills in a single game setting a new all-time MSI record.
Canyon was ahead of the curve in the Jungle, posting excellent results in average net worth and levels per game.
Conclusion
Regarding stats and esports data, the possibilities are endless. As shown above, one-dimensional statistics can be easily combined to produce much deeper insights; team participation in kills and objectives, time-based stats such as Gold Per Minute, and performance ratings are just some examples. The stats feed, and GRID’s entire suite of products have been specifically designed to help unlock the real potential of official esports data. They deliver high-quality, granular data directly from the game server, ensuring accuracy and consistency while providing flexible and seamless integration across multiple game titles so we can unlock the value of esports data for everyone.
Are you ready to Play with Data?
If you are an exciting GRID user, you can start in the GRID Data Portal, where all the linked documentation is available. If you don’t have access to the GRID Data Porta but would like to use GRID Data Feeds to power your project or product, apply for free access via the GRID Open Access; GRID Open Access offers a portion of official esports data from Counter-Strike and Dota 2 tournaments free of charge. While this article uses LoL Esports data as an example, similar stats feeds can be created with other datasets thanks to GRID’s Title-Agnostic Infrastructure.
About the Author
John Neofotistos is a Head of Product at GRID with over a decade of experience working in product roles across betting and sports data. At GRID, he’s responsible for developing the GRID Data Platform. You can get in touch with John on Linkedin.
