Balancing an RPG system?

So, im working on a RPG system for my game, and im getting pretty happy with the basics and sonn its time to start the dreadful task of balancing it. How do you ppl tackle this task, what end to begin in? Obviously it requires a lot of both internal or external testing, but before that can start the system needs to be kind of semi-balanced. I have no idea where to start : D

 

You want to share some more details of your system? It makes a huge difference balancing for single player or multi player. An RTS has very different balance concerns to a FPS.

Normally initial balancing is a spreadsheet exercise. Extract all of the numbers you can use as leavers. Then work out a 'average damage per second' value. Or a whatever per whatever that makes sense. Then start tweaking numbers until you get things to look about right.

Then you want to start playing until things feel right.

 

Personally I make all the units in my game have a basic value of 10. So for balancing I just average the values of the stuff and tweak it till its about 10. Like:
damage 11
firerate 7
accuracy 12
AP 5
range 8
energy eff. 17
(11 + 7 + 12 + 5 + 8 + 17) / 6 = 10, so its balanced at 10.
This does require getting the other units pretty balanced however, so it does take a bit of trail and error.

 

BoredMormon: Its a multiplayer coop RPG game, pulling all numbers uinto a spreadsheet sounds like a good idea to get some overview.

RavenOfCode: Interesting technique : )

 

Wizards of the Coast have done so much research on balance. I would find a dungeon masters guide and take a look at how to adapt balance based on party size and power.

This might be terrible advice: When I was a DM for a few years, I had a basic way of balancing encounters for a party of heroes. first I assume that level/health/damage ratios are all consistent across the classes. then I just add up the total max Health of all the party members. then I make sure each wave/encounter I send at them have an equal amount of total health or even a slight more if I want to make sure one of them falls, thus requiring resurrection or healing skills. You always want healing skills to be needed. This helps the hero party self regulate and balance themselves across classes and skills.

when I play RPGs I like to be a healer class. this is where you need to be carful. If you make the need of the healer class mandatory then everyone will be a healer. The best balancing I've seen for this is to give every class a taste of each of the others abilities. ie. give the Healer an optional ability that can do a lot of damage at the cost of a healing ability. In the case that they are out trying to solo some objectives. also give your High damage class an optional ability for healing. other wise they may find themselves with out the ability to stay alive and it becomes a race with the enemy to who can kill the other the quickest. but please don't go Diablo3 with this. where it doesn't matter what class you choose. the balance is so flat and bland across the classes that you aren't playing to weaknesses and strengths of each individual character. (I haven't played D3 since Beta so maybe they fixed that.)

I have a crazy idea for balancing and RPG game. It's like my one golden idea but on account of work and family I will probably never get around to making it so here you go.

All stats character stats(class) are defined at character creation. Mass is a key variable. player adds strength/muscles to the way a character looks and it takes away from the dexterity stat. Use a triangle graph to plot these statistics between Dexterity, Strength, and Fortitude. at one point I did some number crunching and you have to add some curves to this graph. I'll see if I can send that to you latter. so they can make their character supper fat and it would give them a lot of health but less strength and dexterity. thus making it impossible to max all characteristics. it is a Role playing game after all.

Then you have them use another triangle graph to plot Knowledge, Charisma, and Wisdom.
Inteligence based abilities do more damage
Charisma based abilities do more Buff/debuff manipulation
Windom based abilities do more manipulation/controller abilities.
Then when players find magic items, different abilities are unlocked based on how strong each of these characteristics are. These magic items are embodied by gems or orbs that can be slotted into equipment.

so lets say you find an Ice eliment magic orb. then you slot it into your axe. you can then choose from a handful of options to what ability it gives you. you would have a series of passive and active abilities to choose from

On hit slows enemy movement speed
On hit slows enemy attack speed
On hit cold damage
Activate ability to Freeze enemy still for 1sec.
Activate ability to do cold damage in a cone.
Passive -3 damage from cold attacks

Then you might get a different set of options if you slot it into your boots. allow players to reselect from these options when ever they would like.

because of the graphs that can be used to create a character it would be very easy to randomly generate enemies. then the core of the game becomes, chose the best combination of abilities and weapons to defeat this randomly generated opponent. Player can create encounters to challenge each other. My favorite thing would be to spawn a randomly generated enemy into the open world. balance the enemy based on the number of players in a selected zone, and use my Max Health method. then see if enough players show up to defeat it in time.

... well that was longer than I intended. good luck.

 

To be honest, I think this is one of the biggest problems with multiplayer gaming currently. I noticed it first with WoW. Classes started to blend too much and everyone thought they were a healer. Group dynamics went to crap because not many people really had a clue what their role was any more. People also had no concept of damage mitigation and control. So for me personally, this paragraph is bad advice - depending on what the direction of your game is as far as longevity. Well, that and poorly designed cash shop implementations.. heheh

At any rate, balancing a game is a struggle. But I think you will find the player base a lot happier with having their roles balanced than just letting everyone be able to do everything.

 

Excel (or other spreadsheet software) is your best friend.

 

Seems like a lot of work. GMs job is already hard enough as it is. In my games I would simply cheat and change the numbers behind the screen. Players didn't care, they were having fun.

In single player you actually can cheat a lot like this. Spawn enemies just out of sight as needed. It's much harder in a multiplayer game, players will notice if you mess with their character in game.

 

Thank you very much for the input guys, interesting read for sure : )

 

Hope this helps :
http://gamedevelopment.tutsplus.com/series/balancing-turn-based-rpgs--gamedev-12702
I don't remember wether the author mentions anything about crowd control, which can be game breaker

 

A lot of games like to put the math on a "spine" such as the level.

MMO's and Diablo clones especially do this.

Basically you have all the math center back to the level of the player and the enemy.

So if you have something like:
chanceToHit = playerAgility / targetAgility;

instead do something like:
chanceToHit = (playerAgility / targetAgility) + ((playerLevel + targetLevel) * 0.1f);

I'm not a fan, but it's super common in a lot of RPG's.



Also, try to start off with a straight down the middle character, then create variations of them that polarizes their abilities.

 

This sort of thinking can just as easily be a trap. Unless these values are carefully calibrated from reverse engineered statistical analysis studies, these values will not be equal. The low hanging here is whether damage or fire rate impacts DPS more. Hell, just depending on the algorithm for fire rate, adding one point could produce wildly different results based on what the fire rate already was.

 

I like where this thread is going : )
Following with big interest.

 

What about AI balance? is there a good miss percentage or weapon spread rating? My mind goes strait to the days of Counter-Strike source. it was fun to mess with difficulty. AI would go from hesitating a few seconds before shooting at you. to knowing exactly where you will be and getting a head shot immediately.

 

I guess it depends on what type of game, for a Diably style RPG i dont think you need to put many lines of code into the AI, but for something more story heavy and slow paced you probably need to think a lot more about that.

 

So I have been working on a balancing system for a board game I am prototyping. this picture is a graph that lets you plot one point to determine Strength, Dexterity, Vitality scores.


Notice how I've made a large cap between Blue and Yellow. this is Dexterity and Vitality. I've done this to have a larger trade off between the two. for range char to have lower health. I've come up with my own base numbers that these will multiply.
Now I need help.
I want to make a similar grid to plot Wisdom, Intelligence, and Charisma. this will determine magic abilities.
-Intelligence will do more damage based abilities.
-Wisdom will be more summoning and environment manipulation/controller abilities.
-Charisma will be more like healing and buffing and debuffing abilities.
Do you think these three attributes should have an even trade off or should I offset them like I've done above?

 

I guess I'm not understanding the chart.

What's the context?

Is this of stat growth? Or is this a hard set value?
also,
Blue -> Dex
Yellow -> Vitality
Red -> Strength

Correct?

 

It's a triangle graph. It's been a long time since I saw one in a game (sim ant). But they are popular in engineering.

The user would choose their stats by picking a single point on the graph. The three stats can then be picked by tracing the lines.

 

I thought it was something like that. So it's just for a starting point or for a system where your stats don't change?

Also, what do the colored in parts mean then?

 

The way i do it, is to reverse design the system, start with the highest level / stats / gear and work your way down to level 1. Always consider Mob AI vs you, make "GM mode" where you can test things at all times.

 

Normally the coloured parts would simply represent a class. Red is strength dominant, so a fighter. And so forth.

Spore used a similar system in the space stage of the game. The players 'class' was based on their position on the triangle. Aggressive actions pulled the player towards militant. And so forth.

 

@Not_Sure in my implementation there are no levels. this is hard set number.
I decided that if I matched two average characters against each other I'd want it to take 4 hits to kill each other. and have an HP of 100.
Health comes from the Vitality number so 100/2= 50
then damage will come from either Dex or Str so. 100/4= 25 then 25/2.5=10
so my base Vit =50 Dex =10 Str=10.
then where ever you plot a point, you multiply against the base number.

 

I want to make a similar graph for my Wisdom, Intelligence, Charisma. do you think the 3 skill sets are even. and should have a 1to1 trade off? I'm leaning towards maybe having Intelligence being more dominant so that no matter where you go with your magic you can always do at least a little damage.

-Intelligence will do more damage based abilities.
-Wisdom will be more summoning and environment manipulation/controller abilities.
-Charisma will be more like healing and buffing and debuffing abilities.

 

ie
lets take an ability like cone of cold
High Intelligence, the ability would do damage to enemies in the cone.
High Charisma would slow or stun debuff on the enemies in the cone.
High Wisdom would create an Ice block in the way of the enemies to block their pathing.

 

@BingoBob :
Non damaging abilities impact is kinda hard to evaluate.
Just a thought that could help making balance ;
1- design your classes/characters/whatever you call it
2- create a elo rating system
3- code your combat system
4- create a round robin tournament system
5- code AIs fitting each class
6- make a huge round robin tournament, where each class fights the others (thousands times if your combat system integrates random elements, the more the better. You could instead conduct non-randomized tests ; add mean values instead of rolling dice (ie add 3.5 instead of 1d6))
7- get elo results, make adjustments, goto 6, or 5 (playing a class differently makes a big difference)

By making just one change at a time, I think you can see the impact of the edited ability in combat on every characters.
If after many tests one class prevails still, imho, it should be re-designed (cost more resources in a wargame, for instance)
Due to the nature of games we design, I doubt that we can use the elo rating in order to directly compare one class to an other. Here, I'd only use it for pure performance rating and statistics (you'll know what archetype beats an other one, what ability is op, etc...).

You probably notices the drawback : you need to code more stuff

 

Are you even balancing PvP or PvE? If PvE, single player or multiplayer? Because if it's not PvP, that ELO approach is literally as useful as a zeppelin filled with nitrogen. (For PvP, I doubt it's actually good for anything either, but at least it works in theory until you want to get anything done this century or so.) And all of these categories have vastly different goals for "balancing" (a word that somehow nobody ever clearly defines).

Balancing needs to consider a lot of things:
* different roles in a party (a healer doesn't necessarily need to hold up 1v1 against a fighter)
* skill ceiling (easy to play but somewhat average performance vs. hard to play with extremely high damage potential if you do it just right), handling (based on quick reactions vs. based on careful aim)
* synergies between individual abilities (of one class or of several characters within one party)
* linear-fighter-quadratic-mage problem (some classes might be weaker than others at low levels but immensely more powerful at high levels - e. g. D&D's wizards vs D&D fighters)
* versatility (extremely good in a very narrow category of situations, or decently competent in a bigger variety of situations?)
* usefulness against groups vs usefulness against strong individual enemies

... and a lot more possible problems. I have seen this whole process a couple of times in the tabletop RPG community, and it always keeps coming down to a lot of educated guesses, extensive playtesting and fast iterations.

Also, you can't just write some numbers for two different characters on a sheet of paper and decide if those two characters are balanced against each other without having a clear image of how the rules work. I'm fairly sure that has been mentioned before.

Note that CCGs seem to have very similar issues and usually don't manage to completely "solve" them. You might want to have a look at Hearthstone's update history for some examples of this.

 

My answer was directed to BingoBob, who's making a board game (he mentioned it few posts up there).
His posts are quite vague, so I came up with a generic solution.
Every posters here said, as well as you did, that balancing a combat system is a very difficult task.
My suggestion brings a way to get rough stastistics that give a glimpse on how one change could impact a system, on which one can start working before having his game into open beta test :
It is an automatic way to test the changes you do to your abilities. I think that's quite useful, when you're making a game all by yourself, to have something test stuff for you.

Once the code done, simulations should not take that long. Deciding what and how to change does.

 

But even then, your tests hardly are of any value, because who guarantees you the AI for all classes works equally (or at least comparably) well? How do you know you're not overlooking a giant loophole for one specific class that will throw off the entire game? In particular, when you don't just have to change numbers but mechanics as well for your balancing, you'll also have to change several parts of your tests, which will take a lot of additional effort.

Your solution is not remotely generic - it will only be usable for very small rulesets with few possible plays or relatively obvious "correct" strategies, and even then only in a free-for-all PvP environment. If players form teams, or play multiple characters, or whatever, the number of matchups and tactics you have to consider goes through the roof, and the information you get out of your tests gets a lot harder to evaluate. And you still get no information at all on how easy or hard the right playing style for any class is to grasp - something that will greatly impact which characters feel overpowered or underpowered.

I actually do write automated test suites at work right now - and it's not as easy as you think to design them in a way that gives you meaningful information.

 

Here is my old note that I have copied from the Extra Credits. I have copied and paste my old notes to here. It might be hard to understand because of my poor grammar, but I have linked the Extra Credits youtube video. Hopefully my notes will not be too long.



Basic Math Function Balancing

· Power of Randomness=

o Creates excited moments, give a weaker player to have a chance of wining a strong player, give a player to think and to play differently instead of only working in a one path.
o Delta Of Randomness that must have a lower curve gameplay than the average curve that will already exist to sometime get a higher game curve play. Each game will have an average power curve game that will balance gameplay. The randomness won the more higher curve, the more lower curve it must have at the start and have a lesser chances to win the randomness. But, do not create a too much a wild wide range of up and down power curve to the point it piss off a winner that loses in the end too much times.
o Can gain players not to win too much, or another player to survive in a gameplay for a longer time when they are losing, or gives another player to have a chance to catch up with their opponent.
o Giving randomness can give more excitement instead of only always giving a player with a honest bonus guarantee.

· Changing Average Power Curve=

o Do Not Always= some of the game developers will change the average power curve to be stronger to sell the newer expansion. Usually the buyer is those people who wants to stay above the most competitive players to buy the newer expansion with a power creep. But sometimes in the long run it will be a bad long term goal.
o Most of the gameplay should always follow around the average power curve.
o If some of your gameplay is falling too low power curve, than make a another new game expansion of another gameplay to boost some of the gameplay power curve to it average. It is okay unless you purposely put the power creep by putting in an over the power curve without any reason.
o If you by accidently put a one too high power creep gameplay, then you can balance the power creep by adding that power creep equals to the new expansion power or make a lot of the new expansion that have many power is stronger than the old power creep. But, do not let one sudden new power creep to happen, all else the whole chain rule of your power creep will need to change, and you will loose a lot of money to develop the power curve back to a normal position.

· Planning The Future Power Curve With The Right Chances=

o Always flag a point note to warn the game developer not to create a wrong over power curve design instead of always searching back and forward inside all of the document to make a late decision.
o Should design a game that creates a future proof from dying before a five years cycle. Most video games will only have a five years cycle only.
o Be aware of some game content that have a chain power curve. For example, some fighting monster can summon another random type of a fighting monster, which creates two-power curve at the same time. If that monster have a lot of chances to easily summon too much over power monster, than create more weaker monster so that monster will summon a more balance power curve monster.
o There are also such things as combination power curve. For example, if there is more monster fighting at one opponent, then the more monster will have more chance of winning a game in a long and short run even through each of the monster is weak and not as strong as the opponent.
o Be aware of any combination of two or more different types of gameplay will break the power curve too much when they combine each other. For example, if you play two cards at the same time it will be over power.

Statistic Math Balancing

· What are the lucky chances should the player win=

o The lottery should have more odds and lower chance of winning. So there will have a lesser winner and no one will win. When there is lesser winner the jackpot prices will gain more amounts to jump farther increase until only one player win a large amount within all at once. And then the entire jackpot will become very low then grow back up again if the odd of winning is too low. These is the game lottery cycle should be within one week.
o For example, the lottery game should be like if killing the three times harder monster will gain a lot more experience or rewards than only killing an easy monster to guarantee low rewards, unless you have take too long to kill many easy monster to get enough large amount of rewards.
o Some players will like to sometime see and do anything to risk getting a high amount of lottery instead of thinking of a real world solution.

 

I would like my board game made into a FPS one day. I don't know how to program so this is how I'm prototyping. I have a game shop near my home I plan on play testing at. because of the FPS nature of it, I'm not using any dice. I can't find a good way to simulate a skill shot, so every attack is successful. Damage can be reduced based on Attribute score. ie.. Defense for DEX is Reflex multiplied to BASE 5. so characters with DEX 4 score. can "dodge" 20 damage from ranged/DEX weapons. so if someone with a DEX score of 2 will technically always miss while shooting at someone with a DEX score of 4. (damage BASE is 10) so if you have a DEX score of 2 you probably shouldn't use a Bow as a main weapon and certainly not against someone who has DEX 4.

Also there are no levels. the concept of the game is that you are at max level. this game starts where most other MMORPGs end.

And I am trying to balance for PVP. in my opinion you shouldn't know the difference between PVE and PVP. the AI for PVE should be just as challenging to play against as any other player.
...
Also I have a cool board designed kinda like a giant BattleShip game. to give you the FPS feel and less meta gaming.
(if Scotland Yard was a RPG)

 

If you want to see an interesting board game fps, maybe look at Zombiecide or even Arcadia Quest. Its a great game and some of the mechanics could definitely be in your game.

 

I see your points now, thanks

 

thank you so much @RavenOfCode I picked up both those games yesterday. I've been reading through the rulebooks. there are some sweet concepts here. I can't thank you enough for pointing me in this direction.

The guy at the store was such a dip. He was like, maybe you should try this Ghostbusters one or this Star Wars one. What a terrible "salesmen". if someone walks into your store and say I want to spend $200 on specifically these two things. why would you dance around the store point at other game??.... sorry for the tangent.

 

http://www.gdcvault.com/play/1015037/Fundamental-Multiplayer-RPG

 

This has likely been suggested already, but using a spreadsheet and slap every number you have in there. I personally like to use a rating system for each point. Not everything is 1:1. One point of health is likely not worth 1 point of Strength, so to balance out the system have 1 point of health be worth 1 point, and 1 point of Strength be 5 points or whatever. This also allows you to add curves to the point system. As Strength gets higher it starts to cost more points. So you should be able to average out all the points and use that average to determine raw power. You can then play with the level and see how well it scales.

It's important to note that spreadsheets will never, ever, EVER balance your game. They're only the beginning. At the end of the day you just have to get in there and play test the crap out of it. When you think of balancing a game really what you're asking is, "where do I start".

 

The more complex the progression system the harder it is to balance and the more important it is that the design is balanced rather than just the numbers. When balancing PvE as example avg monster density and the availability of AoE is incredibly important things to consider. AoE for the player as an example increases effective DpS scaling with monster density. AoE for the Monsters however is primarily used for Area Denial("dont stand in the fire !!". Obviously skills that modify mob density(Pull, push, stun etc) can be incredibly powerful but also hard to balance, as it relies on so many different factors, mob density, AoE availability, Mob designs(Slow vs fast etc), CC resistance, Map layout etc

The first question you should ask of any Skill, Stat, Attribute etc is "How does this Scale? What other Skills, Stats, Attributes does this effect?. What is the mathematical relationship between these? General rule of thumb is that Linear Scaling is good, Each point invested is equal to each other point invested so as the player is not unduly rewarded or penalised for stacking that particular stat. In D3 as an example this is the reason that Armor and All Resist scale with "diminishing returns". Armor and All resist are %DR stats which therefore scale exponentially.(adding 25% DR to 50%DR is twice as effective as adding 25%DR to 25%DR for the same stat cost). Therefore Armor and All Resist have to scale with Logarithimic progression to even this out back to a linear progression, so as each point of Armor/All resist added is equal to each other point of Armor/All Resist,

Remember though that stats may scale with themselves linearly but combine with other stats to create an exponential relationship which can make it hard to balance. Think Critical Chance and Critical Dmg in D3. They are incredible powerful for two reasons: They scale exponentially when combined together and they exist in their own multiplier in a system that has multiple stats that scale multiplicatively. In other words they don't get "washed out" or diluted by other stats., think the "Dibs" or Class A buffs, also in D3.



TL/DR-Making sure your game rules are balanced to begin with and being careful not to "upset the apple cart"when adding new rules or skills goes a long way to making sure your game is balanced. It is very easy to maintain a balanced system, it is incredibly hard to balance an unbalanced system to begin with.

Note: I have recently theorycrafted a aRPG combat engine involving 3 offensive skills and 3 defensive skills using spreadsheet and something that really as helped me is a "Gain over Total Sum of its Parts" value. This allows me to compare how the different stats scale with each other and where the synergy between them lies. I am currently working (and learning ) in Unity to prototype this system for my current aRPG....so excited