Welcome!

Hi, welcome back to Design 101. Last time we talked about the Role of Randomness in Game design. In the process, I also mentioned that I’m currently working on a digital cardgame called Faeria. We just entered the major balancing phase of Faeria’s development, which makes this a great time to talk about the topic of how to balance games. We’ve got a lot to get into, so let’s get started.

Game Balance - What is it?

This may surprise you, but while the whole industry talks about “balancing games” there is surprisingly little consensus on what that actually means. Debates run wild in forums and article comment threads about whether a specific game is balanced or not.

As always, the best definitions are the most useful ones. I’ve found that the most useful definitions of balance are based on what we’re trying to avoid: Broken Gameplay. You know your game is broken the same way you know your printer is broken: it isn’t working the way you want it to. If a printer’s casing gets cracked but it still prints, it isn’t broken. It’s just damaged. It’s only broken when it stops printing at acceptable quality.

When you’re designing a game you naturally want to create a positive experience for your players. When your gameplay isn’t providing that experience, your game is broken. It’s that simple.

Broken Games

Let’s imagine adding the following card to my current project: Faeria. All you need to know about the game is that it’s a strategic card game, much like Magic: the Gathering, and that its core resource is called “Faeria”. Even weak effects cost at least 1 faeria to use.

Here’s the card.

All Too Easy - 0 faeria

Event

You win the game.

This card breaks the gameplay for obvious reasons. Like most strategy games, Faeria's fun comes from its interactive gameplay and meaningful choices. This card kills all of that. When someone draws All Too Easy, they have no interesting choice about which card to play. This card is always the right answer. Draw it and win.

All Too Easy would also completely ruin the interesting deckbuilding environment we want to create. Faeria allows you to build your own decks. Naturally we want that process to be a fun strategic puzzle. With this card in the game, the deckbuilding process would break down. First, you should always include three copies of it in all of your decks. This reduces the number of meaningful choices in deckbuilding, because you always know what the first three of your cards are going to be.

Furthermore, the only competitive strategies would likely end up being playing as much card draw as possible. This improves your chances of drawing “All Too Easy” before your opponent does. By making a single strategy the obvious choice, deckbuilding becomes far less interesting overall.

This is why having a card or strategy that’s too powerful can break games. If a single card or even an entire strategy is too powerful, the game breaks down because there’s only one correct decision.

Decision-making is the heart of most gameplay, particularly strategy titles. That’s why it’s also important to NOT have everything be of equal power level.

Wait… What?

The Problem with Equality

Take a look at this math problem, then choose the correct answer below.
 

Problem: 2x^2 + 4x − 4 = 0
 

A) x = −1 +/− √3

B) x = −1 +/− √3

C) x = −1 +/− √3

The problem might be interesting but the choices are all the same. It’s a multiple choice problem where your choice doesn’t matter.

This is basically what you get when all of a game’s cards and strategies are of exactly equal power. Each card you choose to include in your deck is a choice. Each strategy you choose to pursue is a choice. If all the choices are equal, the choice basically doesn’t matter.

Imagine how boring Chess would be if all your possible moves were equally good. There wouldn’t be any point in trying to figure out the best move. All the choices are equal.

Imagine how boring a Hearthstone draft would be if every card was exactly equal in power level. Suddenly your decisions become meaningless, automated only by the cost of other cards already in your deck. Even card synergies can’t save you here, because synergies already play a role in power evaluation. If all cards are equally good choices for your deck, why are you making the choice at all?

Perfect equality can make sense if a section of your game isn’t really focused on strategic choice. In Fighting Games, the character selection is usually supposed to be more about choosing the character you enjoy the most, not trying to figure out which is objectively strongest (though that happens of course). The real gameplay comes from reflexes and mind games within the fight itself. However, if all of each character's moves were equally powerful in every situations regardless of spacing or matchup: that core gameplay would suddenly be irrelevant. All your choices are equal.

The Balancing Point

All Too Easy was an example of what happens when certain cards or strategies are much stronger than everything else. The meaningless math problem is an example of what happens when all the cards and strategies are too close in power. If only we could find some middle ground that was… Well…

Balanced.

A powerful strategy game experience is one where there are multiple viable strategies (preferably supporting a variety of playstyles). Some strategies can be stronger than others overall, but they should still have weaknesses that can be exploited by other strategies. That way you can still play the way you enjoy, while the hardcore optimizers get a continually shifting puzzle to crack.

Here’s how it happens.

Balancing Tricks

It’s basically impossible to get an ideally balanced game through nothing but theory, but there are several tools you can use to make things easier. I’ll go through a few of the most powerful ones now.
 

The Power Curve

The first thing to do when balancing a game is to come up with a simple core formulae for how much power you should get in relation to how many resources you invest. What those resources may be depends heavily on the genre. Some genres rely on mana, others on minerals, others on time. When you're killed in League of Legends and waiting to respawn, you're losing time that could be spent gathering the game's other resources or pushing for a win.

If you have a mana system, figure out how much damage you should be able to deal for 1 mana in your JRPG. If your main focus is on time, take a look at damage per second instead. DPS is also a good way to evaluate the overall impact of a game with multiple resource systems (such as those combining mana and cooldowns).

Having a power curve gives you a firm foundation to base your balancing on. Figure out what your curve should be based on your intended play time. If you want a game to be over in 20 minutes, make sure that the resources a player can gain during that time allow them to push the game to a conclusion. Use this as the starting point for your numbers.

If you don't know your intended playtime, just pick numbers that are easy to work with for now. Deciding that 1 mana is worth 1 damage makes your math a lot simpler during balancing. This makes development faster and easier. You can fiddle with the numbers after your first playtest gives you more information.

The power curve is a clear guideline for how much you should cost your various effects at. However, you still need to figure out a lot of unknown information to predict how your items, powers, cards, abilities and so on will actually perform in game. Games are complex things, and it can be very difficult to estimate the relative value of various abilities.

Luckily, a certain scientist has paved the way.

The Fermi Solution

How many piano tuners are there in Chicago? No looking up any information, no asking for help. Sit down and figure out an answer with just your pen and paper. You can use a calculator too.

You might recognize this question as a famous Fermi Problem. Enrico Fermi was a brilliant physicist famed for his ability to reach precise answers from minimal information. His process, sometimes called the Fermi Solution, commonly gets you within 50 tuners of the right answer to this strangely specific question.

To explain why this works, let’s take a look at the wisdom of crowds. If you take 100 random people at a county fair and ask them to guess a pig’s weight - the average of their guesses will usually be more accurate than a single expert’s guess. Studies have shown this basic principle quite often. Lots of semi-accurate estimations average out to one very accurate estimation. Fermi’s trick was to make more estimations with only one person.

You can’t just estimate directly how many piano tuners are in Chicago. You’ll be off by a lot. What you can do is build an estimation chain. Start by estimating how many people are in Chicago. Then you can estimate how many people are in an average household. Then you can estimate how many of those people are wealthy enough to own a piano. Then you can estimate how many people actually do own one, how often the average piano needs tuning… Etc.

The trick here is that you’re probably as likely to over-estimate one of those numbers as you are to under-estimate one of them. Maybe your estimate for the number of households that can afford a piano is double the reality but your estimation for how often they need tuning is only half of reality. These two estimations would cancel each other out, and you’d still end up with the right answer.

Correctly applying the Fermi Solution is one of the most important aspects of game balancing. It’s also the backbone of why a Quick Pointing process works.

Quick Pointing

Quick Pointing is a process where a designer rapidly goes through a list of options, such as the items in a MOBA or the cards in a CCG, and gives each card a score ranging from 1 to 5. While a lot of these individual numbers may be wrong, we can then add up the Quick Point total of various categories and compare them. The combination of all these points is much less likely to be drastically wrong than an individual guess. By comparing the point totals of all content used by various factions or strategies, we can quickly analyze their rough power.

You can also quickly notice major disagreements about power this way. If any designer rates a spell 1.5 or more points higher than another designer, we know we have something to talk about.

Quick Pointing also allows you to plan out your Point Distribution. In order to avoid the problem with equality, you don’t want everything in your game to be scored as an average 3. Plan out ahead of time how many 1s, 2s, 3s, 4s and 5s you want. The ideal Point Distribution varies heavily based on your gameplay experience, but modeling it after a bell curve is usually a safe place to start. Playtesting will quickly inform you if something’s far out of balance here. Just be sure to make sure that your 4s and 5s showcase your most fun and flavorful options. You can bury the less enjoyable gameplay in the 1s and 2s.

Quick Pointing can’t catch everything though. Nothing can. But we’re not done yet.

Dark Magic

Most inexperienced designers assume that there is some magical balancing-formulae that experienced hands exhaustively calculate when determining ability costs. It’s this assumption that has led me to call what actually happens “dark magic”. As mentioned above, some abilities are much harder to evaluate than basic damage. Here’s an example from Faeria:

Fire Bolt - 6 faeria

Event

Deal 7 damage to a creature.

Lazy Flame - ? faeria

Event

Choose a creature. It takes 7 damage at the start of your next turn.

Lazy Flame is clearly the weaker card. It’s clearly better to be able to damage a creature now rather than having to wait a turn. The question is, “How MUCH better is it to get the damage now instead of one turn later?” What is the value of time-delayed benefits?

The answer lies in Dark Magic. We start by asking ourselves, “What’s a price for this card that is clearly and obviously too much?” In this case it’s 6 faeria, because that's what Fire Bolt costs. If 6 faeria is a fair price for doing 7 damage this turn, it has to be too expensive for a strictly weaker effect.

Then we ask ourselves, “What’s a price for this card that is clearly and obviously too low?” Obviously 2 faeria is too low, because there are lots of opportunities to kill a higher-cost creature with this effect in the average game. I’d gladly take a 1 turn delay to kill almost anything of my opponent’s for only 2 faeria. Trust me on this.

Then we ask ourselves if 5f is obviously too high and if 3f is obviously too low. We continue this process until we hit a number that we’re not entirely sure is too high, and the same for a number that we’re not entirely sure is too low. Once we have those numbers we pick something right in the middle of both of them for the effect. Then we usually give it a small boost.

Why the boost? Because it’s the first step of triple tapping.

Triple Tapping

You’re basically never going to get your game’s balance right on your first attempt. There are simply too many variables to consider. You’re going to playtest and find that some things are stronger or weaker than expected. This can easily warp the whole environment. Your goal isn’t to get it right the first time, but rather to minimize the number of iterations it takes to get there.

Enter triple tapping. Let’s imagine that one of your abilities costs 60 mana and turns out to be dramatically overpowered. You now know that this ability is way stronger than you wanted. Your gut instinct is going to be an attempt to nerf it down to its intended power level. For example, maybe you think the right cost for it would be 80 mana.

Do not price it at 80 mana for the next playtest.

If you do, the card will still be overpowered about 50% of the time. If it is, you don’t have much new data. You knew after your first test that the card’s right price was greater than 60 mana. Now you only know that it’s also greater than 80 mana. There are literally infinite possibilities remaining.

It’s better to intentionally overshoot the target. If you think the card’s correct at 80 mana, price it at 100 instead. If it’s now underpowered you have a lot more information. You now know that the card was overpowered at 60 mana and underpowered at 100. That gives you a narrow range of possibilities for the correct power level. And if it’s still somehow overpowered at 100, you just got that information a lot faster.

I like to call this technique triple tapping because you intend to take three shots at the right number. With this technique you can usually rapidly shrink the number of iterations it takes to get a card’s balance right.

So why the small boost from before? Because it’s better to start an ability overpowered in a playtest. If the ability isn’t fun to use when it’s overpowered, it’s definitely not going to be fun when it’s fair. The boost in power also ensures that the ability gets played more, giving you more data on it. If the ability is underpowered, people might never try it out at all. Not great if your goal is to gather information.

Safeguards

No matter how good you are at balancing, the collective intelligence of your players is going to outmatch you. If you want to be sure your game won’t break you should include safeguards that prevent things from spinning out of control.

In one card game I designed, none of the cards had any prices on them. Instead of drawing cards from a deck, players bid on new cards each turn. Therefore, it was the job of the players to set the price for each card. This meant that if a card ended up being more powerful than I expected, the players themselves were my safeguards. They would correct for it by simply bidding more the next time the card came up for auction.

Wizards of the Coast admits to rarely predicting the eventual competitive environment of their game correctly, but their safeguards make it very rare for even an entire global community of professional players to find a broken strategy.This is because they design their tournament environments to support strategies that prey upon each other. If one strategy becomes too popular due to its power level, players can switch to the strategy that defeats it.

Guilty Gear has managed a large cast of fighting game characters with varied mechanics by ensuring all characters had a wide range of defensive options. No matter what insanity the opponent pulls off, your character is rarely completely helpless. This means that if players discover combinations or tactics the designers didn’t foresee, players will still have a means to adapt.

Poker offers a different tactic. The hands at the table are never balanced. Someone always has the objectively best hand. However, the uncertainty in the game combined with the betting system allows players to steal wins and minimze losses. A big bluff can theoretically beat any hand. The bluffing and betting systems of poker ensure that players always have a way to respond to the perpetually imbalanced situations they encounter.

Installing safeguards into your game works wonders by giving players the tools to adapt to unforeseen problems before they happen. Always try to figure out how you can build some into your title ahead of time.

Wrapping Up

Balancing games is a huge topic and an ongoing challenge to game developers. One of my favorite GDC talks is called Design in Detail:Changing the Time Between Shots for the Sniper Rifle from 0.5 to 0.7 Seconds for Halo 3. It’s an hour long and they didn’t even get through all the slides. If you’re interested in diving deeper into this topic, that’s a great place to start.

See you next time,

Dan Felder

Design 101: Archives

1 - Design Goals

2 - Complexity vs. Depth
3 - The Structure of Fun

4 - Emotional Alignment

5 - The Role of Randomness

6 - Balancing Games

7 - Playtesting