In the fast-paced and ever-evolving world of TFT, a group of heroes known as GAT (Game Analysis Team) play a crucial role behind the scenes. Their mission is to contribute to live balance and set design by ensuring the optimal gameplay experience for TFT from before a set even hits the ground up until the very last patch.
GAT is an eclectic mix of talented individuals, each bringing their unique background and expertise to the craft. But despite their varied background, each member has one common thread, their skill in TFT. Each GATter (can I call them that?) is between Masters and Challenger tier rank—that’s the top .5% to 2% of players depending on when we are in a set. Even with their high rank, they bring much more than just being some of the best TFT players in the world. So let’s meet em’!
Iniko is the newest member of the team. He’s a former TFT professional player who has achieved the coveted rank 1 multiple times. Iniko's invaluable in-game knowledge and insights from the pro player's perspective bring a great addition to the team.
Victor came from VALORANT GAT to join the brand new TFT team. Victor has reached Radiant and Challenger in the respective games and has also reached Legend Rank 1 in Hearthstone before. Combining experiences from different perspectives across gaming, Victor aims to help make TFT fun and accessible for all.
Brian joined from Wild Rift GAT, having reached Rank 1 in WR and looking for a new challenge in the form of TFT. Brian's bringing along all of his expertise and learnings in quality assurance to the game now closest to his heart (that’s the game with the Penguin wielding a wooden sword btw).
Finally, there’s David Lim, who will be walking us through the rest of the article—thanks David. David initially became involved in gaming through coaching organizations such as Team Liquid, Clutch Gaming, XL Esports, and FlyQuest, across both the LEC (LOL European Championship) and LCS (League Championship Series). He eventually joined the GAT team when it was exclusively focused on League of Legends. However, his passion for TFT grew after the game’s release, so when the opportunity arose to spearhead the GAT team for TFT alongside Dave Park, he enthusiastically embraced it. Okay, now to pass it off to David to go through a day in the life of GAT.
Thanks Rodger, GAT’s day-to-day tasks really depend on the specific phase of set development.
Before the set is even ready to playtest, GAT is responsible for design validation—the stage where the team compares the new content to our evergreen set standards for units, traits, Augments and more. There’s a ton that goes into this, but here are the general components that the team must check off before engineers bring the paper design into a playtestable environment:
Beyond general combat and trait structures, the focus of the team becomes much narrower. The team moves to focusing on individual units and traits in early stages by meticulously examining their satisfaction and coherence within the game's framework. A crucial aspect of this process is the "path-to-carry" sweep, guaranteeing that each carry unit has a viable pathway to build towards in-game. For example we would take a 4-cost carry such as Aphelios and analyze what comps or units are good to build into him. Carving out a path to carry for Aphelios enabled the now-popular 3 Freljord, 4 Deadeye comp.
But the “path-to-carry” is just one component of our unit sweep. Another piece of early set collaboration with GAT is aligning design intentions with the characteristics of each unit. Here we ask designers what they want the fantasy or role of each unit to be, and then we work with them through playtests, sims (more on this later), and paper designs to accommodate that fantasy/role into a functional piece of the set that is intuitive and fun. We take into account how each unit fulfills its intended role here, judging units like Sejuani on her durability and utility more than the damage she gets from her passive (which is an addition that makes her a 4-cost tank and not a 3-cost tank).
Augments get a similar treatment from GAT during the design phase. We spend a lot of time trying to break Augments by theorycrafting and simming optimized combos that could create unbalanced situations (there will always be good combos of Augments that synergize—we just want to make sure that synergy doesn’t deny the success of other combinations). We also try to make sure that there is Augment diversity with both the flexibility of Augments (some being niche to specific comps, and others being more general), and the skill requirements of Augments (Cluttered Mind and Think Fast are good examples of high skill Augments that must be balanced by easier to use ones like Tons of Stats! or Ascension).
Augments are complex sources of power in TFT. They can provide access to powerful trait breakpoints (Like Void 8 or Piltover 6), rework how you build a comp (Built Different, Double Trouble), or break rules that we’ve learned about TFT set after set (Infernal Contract, Cruel Pact, Hedge Fund). With all of these complex sources of power, GAT also helps to ensure that each of these, especially the more wild examples, has appropriate balance levers. Take Ancient Archives, which grants a Tome of Traits and a bit of gold. If at any time the benefit from the Emblem becomes too weak on average, we are able to grant players who take it more gold—and we can always do the reverse if the Augment gets too strong as well. Finally, GAT is responsible for blacklisting specific Augment combinations. For example if you take Hedge Fund on 2-1, you shouldn’t be offered Rich get Richer+ on 3-2. For Runeterra Reforged, we also had to blacklist some Augments for specific portals—you wouldn’t want your opponents to get Cruel Pact on the Targon Portal that offers loot once you drop to 40 Tactician Health?
As alluded to earlier, regardless of where we are in a set’s development, we do a lot of simming (using simulations). Simming is where we have ideal (or close to ideal) team compositions fight against each other. To facilitate comprehensive simulations, we use board strings, or pre-designed late-game boards that the team has crafted to be representative of distinct mid to late game comps in a set. For Runeterra Reforged, the team made 82 possible late game boards as strings to sim against each other more efficiently. With these 82 distinct armies, we can make more informed balance and design decisions before a set even goes to PBE. From these strings we can make adjustments too in order to isolate certain factors (Augment vs. Augment, Unit vs. Unit, Item vs. Item) of a comp to compare their power. Simulations play a pivotal role in our work, by allowing us to test different compositions and gauge their power levels more accurately than just data or just playtests. Unfortunately that also means that a lot of our gaming time is just us playing against ourselves in sim environments, but it’s been working well, so while it may be lonely, it’s definitely worth it!
But board strings would be meaningless without the context provided by theoretical strength frameworks. This ensures that compositions of similar cost and accessibility levels are equally powerful, promoting fair and balanced gameplay. Here’s a bit of that theoretical framework:
Board strings, power frameworks, and simming make for just a few of our tools mostly used during pre-PBE stages. Once a set goes to PBE, we’re able to make use of an even greater trove of knowledge…big game data.
PBE gives us a trove of data to make informed decisions with. We sift through data that we can then test/provide story to through our own gameplay (we play a lot) and simming, in order to recommend changes needed during PBE and then again during live, patch after patch.
We look at things such as LP Delta, which measures the difference in LP gained or lost when activating specific units or traits, but also things like round win-rates percentage, and daily meta snapshots. This data-driven approach combined with our deep understanding of the game's intricacies as players, pros, and analysts enables us to make informed and impactful adjustments to every patch.
Reflecting upon GAT's achievements, it is fascinating to observe our growth and transformation across sets. In the early stages, during Monsters Attack!, the team faced challenges related to providing effective feedback and establishing a seamless feedback cycle. When we first started on TFT, we had three members jungling, which is a Riot term for temporarily moving teams to see how other teams operate and what they do—think about it as a seasonal apprenticeship. We did very surface level sims and design feedback sweeps without knowing how to best provide that feedback, the cadence of feedback, or even the content itself. Despite our expertise as individuals, we weren’t as confident as a team on what feedback was useful to designers.
We did eventually learn to acclimate to the pace at which the TFT dev cycle works—plot spoiler, it’s incredibly fast. I heard one designer say that if video game development was a train leaving the station, slowly gaining speed as it’s picked up passengers (employees), TFT is like a train that’s at full speed, ramming through the station, and splitting into three separate realities as it has broken the speed of light and our own ability to perceive it visually. Try hopping on board that train—there’s bound to be some bumps in the tracks. But, with a strong focus on process improvements, GAT quickly accelerated pace in providing game design teams with feedback and analysis.
Runeterra Reforged was the first time GAT has been involved in the fine tuning portion of the set’s design. We wanted to be a lot more involved and have a much more balanced set release where there were many different comps and paths to victory. We are proud of the work we put in with sweeping all the new Augments and our fine tuning simming process we touched upon earlier. We simmed every day and sent biweekly reports to the designers to help tune numbers to the last day before PBE release. And for the most part it did work. Runeterra Reforged was released without a mid-week update (sometimes referred to as B-patch) prior to patch 13.13, and from when I am writing now (during the end of patch 13.12) we’re seeing a boatload of comps find success. This isn’t to say there aren’t issues with the 13.12 meta—there are (Zeke’s Herald), but with how many moving parts there are during a set release we consider Runeterra Reforged’s release a success when we compare it to releases of sets past (Dragonmancer Nunu of Dragonlands or reroll Vayne of Reckoning).
It’s important here to call out where we’ve had success, but also, what we can improve upon. As a certain balance designer you may know once said, “For TFT balance is a process, not an endpoint,” and I think this modified Mort quote is especially insightful when thinking about how TFT balances traditionally, and now with GAT. Before GAT, 100% of our balance adjustments came from a combination of data, community sentiment (yes this informs decisions), and our own play experiences—that’s why each TFT patch is a sort of balance collaboration and conversation with players who will continue to uncover new combos and define the meta. With GAT we are able to get a head start on these conversations before release—they’ll still happen when the set goes live, but the hope is the meta will start off in a much better place. With Runeterra Reforged the meta did start in a better place, but in addition to getting ever-better on our ability to predict balance issues, there’s one other thing we want to improve upon: pacing.
Runeterra Reforged had incredibly fast pacing upon its initial release in PBE. In TFT we refer to pacing as the rate at which your board grows stronger over time. We had an inkling that there was inflation because the new set mechanics, Region Portals and Legends were both added at the same time, and both had ways to increase gold and xp. This along with having over 100 new Augments, led to a lot of ways to earn gold or level faster which greatly accelerated the pacing of the game. While sims do a great job of capturing board and unit strength at a fixed board state, they don’t capture the pacing of the games as much as playtests. We will be looking to incorporate more playtests near the times of PBE release to account for facets of TFT that don’t show up in sims. In fact, we’re already doing that.
The team is hard at work with our tenth set and has been playtesting the set for nearly three months now. This is the first time we’ve been able to playtest a set so far ahead of release, and we’ve already been able to contribute dramatically to the set as a whole—which is a good thing because TFT has had considerably more variables added set after set, and this set is no exception. In addition to getting familiar with the new traits, units, and mechanics of TFT’s tenth set, we’re also working with the Set Design Lead, Matthew Wittrock, on ensuring each unit, trait, and mechanic not only meets design goals, but also checking our design validation requirements discussed earlier.
Our work on our tenth set is the first time GAT has been able to focus completely on one new set (and live). For much of Runeterra Reforged we were still working on the mid-set for Monsters Attack!, Glitched Out!!, in addition to assisting the live team where we could. But now that we’ve switched to a three set cycle, we’re able to dedicate more time to each set, as each set has a longer development timeline. You can check out Mort and Peter’s update on the new timeline here. With more time, we can do more than generalized sweeps and gameplay validation; we can add and run even more scenario sims, new Augment combinations, and test out how the new set mechanic interacts with it all. It’s a brand new world for TFT, and we hope that’s been apparent for the release of Runeterra Reforged, but we’re even more excited to show off just what we can add to TFT with even more time to work on the set!