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

[Lanefan]

But it is using the rules as written.

5e doesn't have a rules-based cutline below which you can re-roll, then?

I was just trying to stay on topic for once and discuss balance. Probably a lost cause.

The problem with balance in this case is that the supposedly-balanced methods eliminate a great many possible outcomes, which IMO is a cost rather than a feature.

One way you can achieve lower variance is to roll more dice. I pushed this as far as I could get it with an absurd 15d2-12 method. I call it the Coin Flip method.

That would certainly serve to put a lot of stats into the mushy middle; I'm no mathmologist but I'd guess the odds of getting anything outside the 8-13 range would be very very low.

 

[Lanefan]

Not necessarily...a lot can happen to a character in the 5-6 levels it would take to get those stats. They might find Gauntlets of Strength, or a really nice flaming greatsword, and decide to build their character around it instead of Yet Another Rapier, for example.

Good point - magic item acquisition can smooth out a lot of differences, depending which items the characters get and - much more importantly - how the party/players choose to distribute said items.

The character who rolls crap for stats probably won't be a combat-first type in any case; it'll either be a sneak (magic items can be a huge help with this) or a back-line caster (in which case its spell selection makes or breaks it anyway).

 

[Lanefan]

Now, some people build with more points like 32 or 36, maybe 40. Or even less points -- 15 points would be equivalent to the Classic (3d6) method.

Only if you allow one-for-one trading of scores so as to allow starting stats of lower than 8.

Otherwise, while the overall aggregated average would be equivalent the actual results on the character sheet would not.

 

[Lanefan]

You consider 65 options constrained?

When comparing 65 options to 54,264 options, that seems a self-answering question.

If you up the number of points available you're just forcing the DM to throw more difficult challenges at you and in my experience the characters area a bit OP.

The DM should ideally be setting the challenges without regard for anything anywhere near that degree of fine-tuning. Is this vaguely level-appropriate? Yes. Good enough. Roll initiative.

 

[Thomas Shey]

I'm not sure what balance is. Earlier we pointed out that the Standard Array is the most balanced. It is also the most constrained.

Well, that's not surprising. Its easiest to control a limited play-space, which is what an array is. I personally find the price there a bit high for it, but there's no question it serves a purpose.

We also pointed out that out of a possible 54,264 sets of scores their are only 65 that equal 27 points. Furthermore, after applying Optimization guidelines to those 65 arrays we ended up with 20 sets of scores.

I don't remember: were those just sets in terms of being six specific numbers or in arrangement? Because that's a big difference in my view.

Now, some people build with more points like 32 or 36, maybe 40. Or even less points -- 15 points would be equivalent to the Classic (3d6) method. So, we pointed out that the full range of scores that can be evaluated for the Point Buy method, from (8, 8, 8, 8, 8, 8) for 0 points to (15, 15, 15, 15, 15, 15) for 54 points. It's interesting that 27 points is smack dab in the middle of that range.

There have been other discussions about the impact of scores on various parts of game play, as well as Dump scores and the Cookie-Cutter issue.

It's been a pretty good thread and I think it still has some legs. I'm still learning things. Like, have you head about AnyDice.com?

Well, one of the issues I'm not sure has been touched--and to be clear, with 5e I'm unqualified to discuss it myself--is whether there are any of the six attributes that are only valuable to specialist characters and any that everyone genercally wants (in a lot of games including some F20 games, Constiution and/or Dexterity often land there). Those are going to at least slightly distort balance considerations.

 

[AlViking]

When comparing 65 options to 54,264 options, that seems a self-answering question.

The DM should ideally be setting the challenges without regard for anything anywhere near that degree of fine-tuning. Is this vaguely level-appropriate? Yes. Good enough. Roll initiative.

I'm not convinced there are 5k options unless you're counting order. I also don't think it really matters. I adjust encounter difficulty for each group all the time. It's harder to do if there's a wide spread.

But it doesn't matter. I prefer point buy, I don't see any benefits to random stats when nothing else about my character build is random. We all play for different reasons.

 

[Maxperson]

So it's useless to compare the difference in expected damage and HP? How? How else are you going to judge relative capability of a character in combat? Again. The percentage chance to hit and the damage actually done if you do hit is significant B does between 53-57% more damage per turn. If there's an error in my spreadsheet calculation please point it out because you're both oversimplifying and significantly narrowing the difference, it's a +4 to hit vs a +7 to hit. You're also ignoring the HP difference at 8th level - A has 60 HP, B has 100. How is that irrelevant?

So tell me? How do your numbers help me when fighting a monster with an AC of 10, 11, 12, 13, 14, 15, 17, 18, 19, 20, 21+? What you came up with is only really useful if everything has an AC of 16 AND someone rolls really badly AND the DM is one of the few who would make the player play the bad rolls.

Or in short, it's not very useful.

 

[Maxperson]

But isn't that inconsistent? On one hand, an extra +1 or +2 in battle - where you make a lot of attacks - doesn't matter, but an extra
+1 or +2 on checks you make very rarely matters a lot?

And it even only applies if you're the one called on to make these checks. The Rogue is still going to be the premier for choice for scouting, not the Fighter with the slightly better Wisdom than the Barbarian, because the Rogue also has the Stealth proficiency and Dex to back it up.

How is it inconsistent? I said...

"Sure. You dump social/informational stats and pump up combat. You'll do minimally better than I do with your +1 or +2 more than me. Meanwhile I'll be doing minimally better than you in the areas that have the most impact. Knowledge and information."

The main difference is that you can generally do much more to affect the game world out of combat than in it. You doing a few extra points to 3 ogres and hitting one extra time every 2.5 to 5 combats isn't going to have nearly the same impact as me periodically persuading the king to shift armies or believe us when we accuse the chancellor of treason.

 

[Lakesidefantasy]

5e doesn't have a rules-based cutline below which you can re-roll, then?

The problem with balance in this case is that the supposedly-balanced methods eliminate a great many possible outcomes, which IMO is a cost rather than a feature.
That would certainly serve to put a lot of stats into the mushy middle; I'm no mathmologist but I'd guess the odds of getting anything outside the 8-13 range would be very very low.

You're right. Here are 10 sets of 15d2-12, where the average is 10.5 and the standard deviation is 1.94. (That's tight!) This is comparable to the Classic (3d6) method where the average is 10.5 and the standard deviation is 2.96.

(10, 13, 11, 7, 10, 8) = Unevaluatable.
(10, 10, 10, 12, 9, 9) = 12 points
(9, 12, 14, 13, 8, 6) = Unevaluatable.
(8, 12, 9, 10, 10, 12) = 13 points.
(6, 11, 9, 8, 10, 13) = Unevaluatable.
(11, 10, 10, 14, 9, 13) = 20 points.
(9, 9, 11, 11, 11, 14) = 18 points.
(14, 8, 10, 12, 13, 11) = 21 points.
(12, 11, 9, 13, 9, 9) = 15 points.
(7, 12, 7, 11, 11, 10) = Unevaluatable.

Here are 10 sets of 18d2k15-12, where the average is 12 and the standard deviation is 2.12. (That's less tight, for whaterver math reason.) This is comparable to the Standard (4d6k3) method, where the average is 12.24 and the standard deviation is 2.85.

(8, 10, 8, 12, 12, 9) = 11 points.
(15, 8, 15, 13, 15, 12) = 36 points.
(13, 12, 11, 17, 7, 11) = Unevaluatable.
(12, 7, 15, 13, 13, 9) = Unevaluatable.
(13, 9, 11, 13, 15, 13) = 28 points.
(13, 9, 12, 13, 12, 14) = 26 points.
(15, 11, 12, 8, 10, 12) = 22 points.
(10, 9, 12, 14, 17, 10) = Unevaluatable.
(9, 11, 13, 14, 15, 10) = 27 points.
(15, 14, 15, 12, 12, 12) = 37 points.

 

[Lakesidefantasy]

Only if you allow one-for-one trading of scores so as to allow starting stats of lower than 8.

Otherwise, while the overall aggregated average would be equivalent the actual results on the character sheet would not.

I think I see. This is because the point buy value scale is not symmetrical around the Classic (3d6) method's average of 10.5.

We can make these 3 versions of a big, strong commoner.

(13, 10, 12, 9, 8, 11) = 15 points.
(11, 10, 11, 10, 10, 11) = 15 points.
(15, 8, 13, 8, 8, 9) = 15 points.

But we're only allowed to go down 2.5 points down from the average of 10.5, whereas 15 is 4.5 points away.

But, we can extrapolate the lower scale by reflecting the upper scale over it so that we can define 7 as worth a 1-point refund, and a 6 as worth a 2-point refund. We could extrapolate all the way down to 3, but we should stop at 6 because that would make the scale symmetrical around the Classic (3d6) method's average of 10.5.

So we could make these commoners to balance the ones above.

(14, 10, 13, 7, 6, 12) = 15 points.
(14, 6, 14, 6, 6, 14) = 15 points.
(15, 6, 15, 6, 6, 11) = 15 points.