Balancing Beastieball

A look at what goes into tuning a PvP turn-based strategy game

Without a doubt, the most fun I have working on Beastieball these days is getting to work on the game balance. It sounds like thankless, grindy math… and sometimes it is! But it’s also an area of the game that’s still living, breathing, and changing every day… where math intersects with emotional experience, the very essence of game design. And seeing as we just released a large new balance patch, it seems like a good time to talk about what guides our process, and how we got to where we are.

What We’re Dealing With

Beastieball is a turn-based strategy game with no randomization whatsoever. It also has very little hidden information; players can see every option their opponent has access to, and exactly how much damage it will do. That means the game has to be pretty mathematically tight, because the numbers have an outsized role in the outcome of every game.

But the game also has a huge creative element in its teambuilding. Every player has their own favorite Beasties to use, and that influences their team composition. Beastie matchups are generally very fluid, without strict strengths or weaknesses—they’re more like “colors” in Magic the Gathering than “types” in Pokemon. In many ways, game designing an individual Beastie feels more akin to designing a fighting game character, with unique tools and a unique playstyle that caters to a certain kind of player. And our ultimate goal in balancing the game is to let players feel that they can be successful using ther favorite Beasties, no matter who they may be.

Laid out like this, it sounds pretty daunting to make math that feels tight enough to feel balanced, but loose enough to fit 100+ different character playstyles in it. Maybe even impossible! But somehow it’s been going well for us so far.

Measuring Success

We make use of analytics to track what happens in competitive PvP games. For every match where players have analytics enabled, we track what teams were brought to the game and which one won. And over thousands and thousands of matches, we can observe what trends emerge. Here is a chart!!!

This plots how popular a Beastie is (what % of teams it appears on) VS how successful it is (what % of its games does it win)

Each dot here represents a Beastie with a certain trait; so Daredillo with the “Self-assured” trait is a separate dot from Daredillo with the “Acrobat” trait, acknowledging their different playstyles are effectively different characters. Higher dots are used more often, and dots further to the right win more often. If we narrowed our data to only include top-level players, especially as the game is out for longer and the meta advances significantly, you’d expect this chart to show more of an upward slope where the Beasties that win more often also are used more often. But what we see isn’t like that, because players generally choose their Beasties by personal preference, and there aren’t any that are predominantly known to be better than others.

How we interpret this data

What we’re seeing is a “normal distribution.” Beasties that are used more often will trend towards a 50% win rating, partly because they’re statistically playing on both teams of more encounters, but also because they represent the aggregate skill of more players. Whereas Beasties used less often may have data from as few as 1-5 players, and so their statistics are basically just showing us the skill level of their inidivudal coach rather than the Beastie’s inherent strengths.

But those dots we’ve circled in orange are performing outside the curve—across many players and games, they’re clearly losing more than they win. And it’s similar with the dots circled in green: those ones are clearly and consistently winning more often. So this data gives us a strong starting point to figure out what Beasties we want to target with changes, either to buff or nerf. It isn’t strictly binary… dots close to the edge on either side likely need attention, and any change we make try to should account for precisely how far off expectations that Beastie is.

While hard data is comforting and helpful, we can’t rely on it to get the full picture. We’re always keeping tabs on player discussion and feedback as well. Sometimes, a particular strategy may just be unfun to use or play against, even if it isn’t statistically successful or dominant, and we’ll have to consider what we can do to adjust that area of the meta. There are other times where player speculation about a particular Beastie will gain traction even when it’s statistically flat wrong, though. At the end of the day, complaints about game balance are as inevitable as it is for every match to have a loser. So having data helps us keep feelings in check.

Making the Right Adjustments

Once we know what the problem is, the fix is easy. We just double the numbers that are too low and halve the ones that are too big.

Just kidding! This is the hardest part! Partly because there are a lot of constraints that we try to follow with every change we make. Knowing when to follow, push back or compromise on these isn’t a science—it’s an art.

Maintain unique strength

Masahiro Sakurai has a concise video on this topic—but to sum it up for our case, every Beastie and play and trait has some essential unique strength to it and we always try to respect that with changes. So when something is overly strong, we may try to shift its balance by exacerbating its weaknesses. And when something is underpowered, we try to strengthen its most unique elements, rather than giving it additional strengths that outshine its original intent. The goal is to maintain peaks and valleys across the game and make sure every element maintains its unique character, rather than all melting into flat blandness.

(But not too spiky)

Our predisosition to exacerbate strengths and weaknesses means we sometimes accidentally push elements too far, and they can dominate the game. Beastieball is also a game full of unique interactions, and for any drawback we add to any play, players always find some way to mitigate it. So anything that’s “very strong, but with a strong drawback” always winds up needing special attention from us. It can take a couple iterations to find the right maximum power level.

Maintain flavor

Game elements aren’t just code and numbers! We always want to make changes that are consistent with the characters and fiction we’ve established. For example, it may feel balanced to make a powerful support action give the user the “Angry” status, preventing them from using further support actions. But we’ll always resist doing this if it doesn’t make sense for that action to make the user feel angry. Finding adjustments that do make sense within the fiction can be challenging, but often results in more creative and evocative solutions.

Consider accessibility

When a Beastie is underperforming, we’ll look at their statistics with high skill vs low skill players. Sometimes a Beastie is perfectly strong, or even overperforming at high skill levels, but underperforming in general, which indicates that their best tools may be hard to access or that they are more challenging to pilot. We have to treat these cases with a delicate touch. As often as possible, we’ll try and create easier paths for low-skill players follow without actually strengthening them.

Consider the context

We can only measure Beasties’ effectiveness by their relationships to each other. Sometimes a Beastie is over or under performing not because they themselves are strong or weak. For example, one Beastie that overperforms will also raise the win statistics of the Beasties that they are commonly paired with. And the more often a particular Beastie is used, the better the win rates will be for Beasties that play well against them. So sometimes the perfect adjustment for a game element is actually to change something else entirely.

Consider the vibes

Even with all the data at our disposal, sometimes we’ll make a change just because it feels right to us. I have to wonder how often then happens in larger budget projects with more collaborators… but as an early access title that patches frequently, it’s freeing to be able to “try” things and see what happens, with the freedom to take it back if it makes the game less fun.

Putting it together…

One balance change I was quite pleased about was to a move called Set, an iconic passing move which made the ally feel Jazzed, buffing their next attack. It turned out that feeling Jazzed was a REALLY strong status effect, particularly because it could be combined with other strong things, and this play quickly took over the meta. Players wanted us to make Jazzed less effective, or add a debilitating cost to Set so that it was punishing to use.

While this conundrum was on my mind, I went out to play volleyball at a local casual league. I used to play with a group of friends who are, like me, all quite novice and unathletic. But on that week one of our members was out sick, so one of the coaches who runs the league offered to fill the spot on our team. And as an skills exercise, he encouraged us to all play different roles than we were used to, and so made me the setter on the team. I was pretty reluctant about it because I’m just not good at passing, and I was worried about holding the team back, but we went ahead with it. The game starts, and I’m waiting at the net and terrified about the pass I’m surely about to fumble. The first serve goes to the coach, he gets it perfectly and puts up a beautiful pass to me. And with such a perfect pass, better than any I was used to getting, it was like I became a different player. I had the extra time I needed to get to the right place, stop, square up, and put up a set with perfect form.

Now, I know I’m still pretty bad at passing. A great setter is great because they make great plays even under bad conditions. But this experience showed me how much easier it is to do a good job when you’re set up to do it. And when I got home that night I implemented my idea for an adjustment to the play called Set, which required that the ball be hittable for the play to be used, mirroring the way a setter in volleyball is always meant to get the 2nd touch on a ball, not the 1st. It wound up being just the right adjustment, keeping the play just as strong, but requiring players to support their setters a little bit to get the most out of them.

---

Balance adjustments are fun to do, but daunting, because the work never truly ends. With a meta that’s constantly shifting, and player creativity always discovering unexplored corners of the systems, the game can never be truly balanced forever. But it’s also the area of the game where the conversation between designer and players is most alive, and that’s what makes it so rewarding to come back to every day.