D&D 5E (2014) Is Point Buy Balanced?

[Lanefan]

And I would insist on the rolls being done together, very publically, at Sesssion 0, with each player taking their turn while the others watch.

This should be SOP for rolling any player-side dice for any reason in the game (though not everyone else has to watch, just the DM). "Roll on the board where we can see it" has been a hard-line player rule here since forever.

 

[CleverNickName]

I do point buy while friends roll. My stats are usually lower. Hmmmm

I decided to model this in Excel, to see if it's a math thing or a perception thing (or a cheating thing).

Using the Standard Array (15, 14, 13, 12, 10, 8), you end up with a total of 72.

For the Point-Buy Method, I pulled up a complete list of all 65 valid ability scores that you can get using the Point Buy method (Source). I put them into a spreadsheet, and summed each result (15+15+15+8+8+8=69, 13+13+13+12+12+12=75, and everything in between.) Then I added them together and averaged the result.

The average for Point Buy is 72.3, but that's moot: the player chooses a result between 69 and 75.

And for the 4d6 Method, I had Excel generate six stats using the 4d6 method, and add up the total of all six stats.
And then I had it repeat that calculation a total of 2,560 times, and then average the result.
The average result from the 4d6 method was 73.5.

So it does look like the 4d6 method will---on average, across thousands of iterations--generate stats that are about 1.5 points better than the Standard Array, and about 1 point higher than the average value of Point Buy...but you can get results as high as 75 using Point-Buy, through careful selection of stats. (13, 13, 13, 13, 12, 11 gives you 75 points, which is much higher than any other method and it's 100% reliable.)

TL;DR: I think Point Buy is fairly well-balanced with the other methods. The advantage for point-buy is that the player can choose how over- or under-powered they want their stats to be, by as much as +/- 3 whole points from the standard array, while with the 4d6 method you are probably going to end up at +1.5.

 

[FarBeyondC]

I decided to model this in Excel, to see if it's a math thing or a perception thing (or a cheating thing).

Using the Standard Array (15, 14, 13, 12, 10, 8), you end up with a total of 72.

For the Point-Buy Method, I pulled up a complete list of all 65 valid ability scores that you can get using the Point Buy method (Source). I put them into a spreadsheet, and summed each result (15+15+15+8+8+8=69, 13+13+13+12+12+12=75, and everything in between.) Then I added them together and averaged the result.

The average for Point Buy is 72.3, but that's moot: the player chooses a result between 69 and 75.

And for the 4d6 Method, I had Excel generate six stats using the 4d6 method, and add up the total of all six stats.
And then I had it repeat that calculation a total of 2,560 times, and then average the result.
The average result from the 4d6 method was 73.5.

So it does look like the 4d6 method will---on average, across thousands of iterations--generate stats that are about 1.5 points better than the Standard Array, and about 1 point higher than the average value of Point Buy...but you can get results as high as 75 using Point-Buy, through careful selection of stats. (13, 13, 13, 13, 12, 11 gives you 75 points, which is much higher than any other method and it's 100% reliable.)

TL;DR: I think Point Buy is fairly well-balanced with the other methods. The advantage for point-buy is that the player can choose how over- or under-powered they want their stats to be, by as much as +/- 3 whole points from the standard array, while with the 4d6 method you are probably going to end up at +1.5.

The funny thing about point buy is that, for players who only need/want certain stats to be good, the lower total arrays are better than the higher total arrays. For instance, I'd almost never pick the 75-point (13, 13, 13, 12, 12, 12) over the 71-point (15, 15, 13, 10, 10, 8).

 

[CleverNickName]

That first histogram looked strange to me--I expected to see a smooth bell curve, and there were these really weird gaps in it. So to better see what was happening, I ran it again---and this time I pushed Excel even harder to generate over twenty thousand iterations. And this is what I got:

So there are now FOUR big gaps in the curve, where the statistical average drops to almost zero (about 1 result out of 20,480, or about a 0.0048% chance). Pretty weird, huh?

I was scratching my head until I realized what was happening: this 0.0048% is the chance of rolling four identical results on 4d6, for all six ability scores...not impossible, but highly improbable. There are the four gaps you can see, which occurred when Excel rolled all 2s, 3s, 4s, and 5s...and there are two gaps on either end of the graph where Excel rolled all 1s and all 6s. You just can't see them as well because they are at the far ends of the graph.

Stats are cool.

 

[M_Natas]

Your point about feats is well taken; in 5e feats allow a lot more customization and even a 1 point advantage to the initial set of scores allows you to take an extra feat while still letting you maximize your essential attributes.

Yeah, that is also my problem with the standard array. Basically you start with two 16s or one 17 and one 15 in your two important stats (after racial/origin boni) and then, yeah, you can take a half feat on level 4, but then level 8 is an ASI, so you can only really start taking feats at level 12, which is where most games end.
Because getting the main attribute to 20 first is the most efficient way to play.
Yes, you can take feats at levdl 4 and 8, but you will.soon be running behind the other players in ability, which is not fun.

 

[FarBeyondC]

Because getting the main attribute to 20 first is the most efficient way to play.
Yes, you can take feats at levdl 4 and 8, but you will.soon be running behind the other players in ability, which is not fun.

See, I know other people feel this way, but I've never really understood why.

Like, if there's any feat at all that's actually relevant for my character concept, getting that as soon as possible is always going to outweigh going from 16 to 18 or 18 to 20, because missing that aspect of my character hurts more than missing by 1 or 2.

 

[AlViking]

You literally did. This is you.

"You can get more variety by creating a number of arrays from point buy and roll for which array you use. Some will start with a 15 as a high, some with a 13 (with 5e you get a +2 and a +1 to add for a 17 or 15 high number). But the 15 will have some pretty low numbers, the 14 will gave all numbers above average. Everything from 13, 13, 13, 12, 12, 12 to 15, 15, 15, 8, 8, 8."

I didn't say you can get more variety than rolling, You can get more variety than the standard array, and you can get more variety than most people will generate from point buy.

So you want me not to be able to play the game, because that's what happens if rolling isn't at least equal as a default to point buy/array. I will never use point buy/array. Not having the ability to roll stats is a deal breaker for me.

Well, I will never use rolling.

The odds of rolling 18, 18, 17, 14, 14, 14 on 4d6 drop the lowest are incredibly small. Too small to even be a concern.

Yet I've seen it happen.

So that's wrong. There are other ways to get stats. A headband of intellect 19 is just an uncommon item. Tomes to get +2 to a stat exist and you can get them multiple times. And that's assuming that you need high stats in 5e. You don't. You're better off getting feats that influence your magic ability than raising your stats.

The stat items are attunement so very limited and dependent on DM and whether they're allowed or not. For example I use the 3e rules where stat adjustments add they don't replace. I've been in other campaigns where they simply didn't exist.

You aren't the only one who plays the game, so what concerns you isn't king. It doesn't trump what concerns me. My desire for rolling is every bit as meaningful as your desire for arrays. They should be equal options.

You do you, I'm just expressing my opinion. Rolling for ability scores adds nothing as far as I'm concerned. Other than practically guaranteeing some people will have inherently more powerful characters than others because of a one time roll of the dice, I see no reason to use rolling.

 

[CleverNickName]

The funny thing about point buy is that, for players who only need/want certain stats to be good, the lower total arrays are better than the higher total arrays. For instance, I'd almost never pick the 75-point (13, 13, 13, 12, 12, 12) over the 71-point (15, 15, 13, 10, 10, 8).

That might be why some people feel the 4d6 method generates higher stats than point-buy does. On average the 4d6 method will give you 72 points, which is about same as the standard array--but most people who use point-buy will choose an array with fewer points.

 

[EzekielRaiden]

Sure. I'm fine with this. Most of the bell curve's population is in what I call the mushy middle; and the question is how to game-mechanically deal with the extremes on either end of the curve.

The problem in game terms is that the WotC editions assume linear rather than bell-curve progression when applying bonuses, giving the game-mechanical difference between 10 and 12 in any stat the same mechanical heft as the difference between 18 and 20, or between 3 and 5. This is where things fall apart, as the difference between 10 and 12 (or even 8 and 12) shoud be mechanically zero as you point out, with bonuses and penalties only accruing at the outer ends of the bell curve.

No. It "should" not be any specific thing. It is a design choice. You are conflating "this is how 1e does it" with "this is how all games should do it". Both are valid. They're just different.

None of which actually addresses my core point. You are asserting that characters should have very high variability. I'm telling you they should not have high variability. They should in fact actually be mostly similar! Because, as it turns out, most people really are mostly similar. Comes with being a species and all. Your very foundational principle--the idea that characters should be highly variable--is inherently unrealistic. Anti-realistic, even. It is there solely to produce an actively un-realistic aesthetic that you personally prefer. Hence, an argument for it which claims to be based on realism is presumptively wrongheaded (since the alternative is to assume bad intent).

That's why in 1e there were minimum stat requirements in order to be a class. A Fighter had to have Strength at least (9? I forget and can't be bothered to look it up right now). A 5 or lower in a stat dictated your class; two stats of 5 or lower made the character unplayable as it couldn't qualify for any class.

Don't you think a system which can produce outrightly unplayable characters has a clear hole that should be fixed? That sounds like a pretty serious flaw. Folks love to crack jokes (or rarely, actually criticize) Traveller for being able to die in character creation. I should think "you can do character creation and be literally unplayable after the second step" is something needing fixing.

Doubly so when, y'know, we don't do assign-in-order.

We shouldn't expect someone with Str 3 to be a Fighter but that character could very well be playable as a mage type, or a talky type, where physical strength isn't a requirement.

I should think someone with Str 3 should never even think about delving into a murder-hole for fame and fortune. Mostly because, y'know, they're likely to die?

Wouldn't your hypothetical Str-3 Fighter (which even in my game can't exist in practice) put level-granted ASIs into Strength?

Why bother?

Seriously, why bother? That's such a blatantly unrealistic, rules-forward way of altering player behavior, when we could just...not. Like we could literally just say that characters already did their training....since that's literally what the books already say. A level 1 Fighter is not simply some random town guard or peasant with a pitchfork. They're an actual warrior, with fairly extensive combat training, whether or not they have actually seen their first battle (but the text strongly implies most Fighters already have real combat exposure before level 1.)

Not sure, though, how we got to a Str-3 Fighter in the first place.

Uh...you're the one who brought that up. I did not mention that number at any point in my post. I simply spoke of "low extremes", which in this context would be anything 7 or lower (at least as far as I'm concerned). Someone who is two standard deviations below the mean--meaning 97.5% of people are stronger than they are--absolutely should not be even attempting strength-based challenges, both because they will be objectively terrible at it and because, presuming at least a modicum of rationality, they won't want to do things they know they're really, really bad at.

I think we might be agreeing here only from very different directions.

Perhaps. You have advocated score requirements with no physical representation, character creation rules that can create truly unplayable characters as in the rules forbid you from playing them, and variability which does not actually capture what most people really are. Even under 3d6 strict, you get +2 and even +3 modifiers much too often.

I also think people tend to improve in what they practice at, and it makes sense that members of most adventuring classes are going to practice at those things which improve what they do as that class: Strength for a warrior type, Intelligence for a mage type, etc. The way to reflect this improvement-through-practice in 5e game mechanics would be to force level-based ASIs to go into those "primary stats".

Ooooor...we could just recognize that people already did get better at it...by recognizing that their ability scores at first level represent a mixture of both innate predisposition AND developed skill/intuition/ability. Which is much, much more realistic than the outrightly anti-realistic notion that the entirety of your strength or intelligence is innate and unchanging.

Before doing that I think the class-agnostic elements of the different stats need balancing, though I've no real idea how to do this. Dex is way too powerful (and has been in every edition), Con is too powerful (though less so than in the early days), while in the early editions Cha was fairly useless and often dumped.

I was told Cha was THE god-stat of early-edition D&D because it controlled reaction rolls and the ability to recruit followers, which could utterly transform the experience of fighting through a dungeon--because you could recruit your opponents, bolstering your own forces while weaking the dungeon's guardians.

But yes, Dex, Wis, and Con are varying degrees of too powerful, while Strength and Intelligence are weak in 5e, especially the latter. Charisma is more or less balanced as it has clear uses, but is not useful everywhere, and some things contextually make it more useful (like Blade Pact Warlock). 4e did quite a lot of work to address this:

• Constitution only increases your HP at first level in 4e. Technically, each point of Con mod also gives you another Healing Surge, but it doesn't change how much value you get out of them (unless you're a Dragonborn.) Hence, Con is MUCH less powerful in 4e--and this is partially compensated by giving it a skill, Endurance, which gives Con value in exploration and logistical challenges.
• Dexterity, while still very powerful, is less so because your Reflex defense (=Reflex save from 3e, just reformatted so attacker rolls), and light-armor AC, can come from either Dex or Int, whichever is higher--making Int more valuable. More could have been done, but it's not a bad start. Or if either of them also worked for Initiative, it'd work just fine.
• Wisdom is still a god stat and, of course, could be brought down a peg. (Personally, I think Perception should be divorced from Wis just as Initiative should be from Dex.) It doesn't need massive nerfs, though. Just pulling out Perception would probably be enough; being the "anti-mind-control" stat is a pretty good thing, and it is used for other skills (Healing and Dungeoneering--aka "dungeon survival skills"--are pretty solid.)
• I'm still unsure what to do with Strength. Athletics is a good skill, but other than that, it's entirely feast or famine. Encumberance isn't doing the trick--and not just because lots of folks ignore such things because they aren't engaging.

It really wouldn't take that much effort to find an initial solution. The actual effort would be doing the playtesting to see if the stats are in fact balanced against one another, and iterating on that until you reach the desired outcome (where players are reluctant to dump any stat, because all of them have important, desirable uses). I have no confidence that WotC as it exists today has the desire, let alone the werewithal, to do either the kind or the amount of testing required.

 

[EzekielRaiden]

I decided to model this in Excel, to see if it's a math thing or a perception thing (or a cheating thing).

Using the Standard Array (15, 14, 13, 12, 10, 8), you end up with a total of 72.

For the Point-Buy Method, I pulled up a complete list of all 65 valid ability scores that you can get using the Point Buy method (Source). I put them into a spreadsheet, and summed each result (15+15+15+8+8+8=69, 13+13+13+12+12+12=75, and everything in between.) Then I added them together and averaged the result.

The average for Point Buy is 72.3, but that's moot: the player chooses a result between 69 and 75.

And for the 4d6 Method, I had Excel generate six stats using the 4d6 method, and add up the total of all six stats.
And then I had it repeat that calculation a total of 2,560 times, and then average the result.
The average result from the 4d6 method was 73.5.

So it does look like the 4d6 method will---on average, across thousands of iterations--generate stats that are about 1.5 points better than the Standard Array, and about 1 point higher than the average value of Point Buy...but you can get results as high as 75 using Point-Buy, through careful selection of stats. (13, 13, 13, 13, 12, 11 gives you 75 points, which is much higher than any other method and it's 100% reliable.)

TL;DR: I think Point Buy is fairly well-balanced with the other methods. The advantage for point-buy is that the player can choose how over- or under-powered they want their stats to be, by as much as +/- 3 whole points from the standard array, while with the 4d6 method you are probably going to end up at +1.5.

View attachment 425876

View attachment 425877

A bigger issue is that I'd say about a quarter of possible rolled stat arrays, despite being legal, will get written off as being "too" bad, and thus the actual average shifts up. It's not cheating, but it is unfairly biasing rolled stats to have even further favor--the average result is slightly better to begin with, and then human interference makes it moreso, removing the worst parts of the curve.

I would not be at all surprised if rolled stats, on average, actually do objectively better in both raw scores and ability mods, even if not a single person cheats. The problem is, as I and others have said, that ability roulette means you MIGHT get stuck with something just barely not horrible enough to get re-rolled.....which is precisely why the "well those stats suck...ah just re-roll it and see what you get, but you have to take it" thing (or something like it) happens.