D&D 5E Point Buy vs Rolling for Stats

[Oofta]

Now this, I have to say, is cool stuff.

By any chance do you happen to remember what the overall average point-buy cost turned out to be?

Also, how did your program (if it did at all) assign point-buy values to stats below 8?

And one thing that needs to be noted: point-buy values really skew high once an 18 gets rolled. Particularly if numbers less than 8 are treated like an 8 you could hypothetically have these two characters:

12-12-12-12-12-12 - average 12.0 - net overall bonus +6 - point buy cost 24
18-14-7-7-7-7 - average 10.0 - net overall bonus -2 - point buy cost 23

Roughly the same point cost...and about there all similarities end. Which means, while in large-data analysis pont-buy values is an OK comparison it might not always stand up for comparing individual characters.

My analysis was based on numbers below 8 just counting as the number minus 8 because I didn't want to skew the results too much.

Things do get screwy though when you try to do the comparison, and I wish there was a better way. The standard point buy seems to be based on the 3.5 elite array (15, 14, 13, 12, 10, 8), which also seems to be pretty close to the average according to this.

It seems to me that a few high stats mixed in with a few low stats is going to average out to a character that's much more competent in combat than someone that has all low or average stats, although saves and out-of-combat options may suffer of course.

According to my app, the average minimum for any given character is 8, the high is 15. Limiting numbers to that range, the result is a little below 27 if you roll. If you follow the 3.5 point buy and allow 18s it's 31, if you ignore numbers below 8 it's a 32. That to me indicates that the high numbers are probably a bit too costly ... or that at a certain point you're just comparing apples to oranges.

 

[Maxperson]

Again, stating your opinion as fact. We have different definitions of "Mary Sue", "Superhero", etc.

I'll say this one last time. Get. Over. Yourself.

You don't get to tell me what my opinion is.

Sure, but if your opinion over inflates what a "Mary Sue" or "Superhero" is, don't expect people to agree with you.

It's unlikely. It happened. Two eighteens for a character are going to happen about .38% of the time according to this independent article.

So about 1 in 300 just for the two 18's. Add in the 17, the 16 and 2 14's and it's no wonder that in about a thousand characters I've never seen it and probably never will. It's not just unlikely, it's extremely unlikely.

But significant differences in the results of 4d6 drop lowest are pretty common as anyone with some d6's and a little time can tell you (and have). I've tried to do some analysis on overall trends and given you samples that you choose to ignore.

I know there are differences. I've seen them. They just don't matter much once you realize that stat numbers are really the least influential of the major game influences. Roleplaying, character abilities and player ingenuity all far surpass stat numbers for game influence.

And again, you opinion stated as fact. We have different definitions of "fair". I don't think random results are fair unless averaged out over time.

I'm using the real definition, though, not a made up one. Fairness doesn't mean that the results have to be fair, and that really is a fact. If you and I are equal at darts in skill, the rules don't favor either one of us, and we use the same quality darts, that's fair, even if you lose. Results don't have to be equal to be fair.

 

[Oofta]

Sure, but if your opinion over inflates what a "Mary Sue" or "Superhero" is, don't expect people to agree with you.



So about 1 in 300 just for the two 18's. Add in the 17, the 16 and 2 14's and it's no wonder that in about a thousand characters I've never seen it and probably never will. It's not just unlikely, it's extremely unlikely.

I know there are differences. I've seen them. They just don't matter much once you realize that stat numbers are really the least influential of the major game influences. Roleplaying, character abilities and player ingenuity all far surpass stat numbers for game influence.

I'm using the real definition, though, not a made up one. Fairness doesn't mean that the results have to be fair, and that really is a fact. If you and I are equal at darts in skill, the rules don't favor either one of us, and we use the same quality darts, that's fair, even if you lose. Results don't have to be equal to be fair.

Dude, I'm done. I've violated a New Year's resolution to not get into pissing contest. We have different definitions of "Mary Sue", "Superhero", and "fair". You have different ideas of whats fun.

There's no point to continuing.

 

[Lanefan]

It's unlikely. It happened. Two eighteens for a character are going to happen about .38% of the time ...

So, slightly more often than 1 in 300. Yeah, going by memory that sounds about right from what I've seen here...only we use 5d6k3; and out of maybe 1300 characters over 35 years of various games and campaigns I can only recall three or four* double-18s (and they do tend to stand out in the memory!)...and no triples.

* - ignoring one or two that I dead-shot know weren't entirely honest...

Now, to save me going back and doing lots of reading, am I right in remembering that the example you used involving you and your wife rolling stats included one of you rolling a double-18?

Reason I ask is if this was the case then it sounds like you hit the jackpot for stat disparity...no wonder you have a dim view of rolling. It's not always that bad...

Lanefan

 

[Lanefan]

My analysis was based on numbers below 8 just counting as the number minus 8 because I didn't want to skew the results too much.

So, 7 = -1, 6 = -2 and so on? OK, makes sense.

Things do get screwy though when you try to do the comparison, and I wish there was a better way.

Yeah.

Whenever I try to analyze these things I usually end up looking at the average of the 6 rolls - e.g. the 15-14-13-12-10-8 array would give 12.0 - and comparing those.

It seems to me that a few high stats mixed in with a few low stats is going to average out to a character that's much more competent in combat than someone that has all low or average stats, although saves and out-of-combat options may suffer of course.

This is what messes up any analysis unless one is far better at math than I am: accounting not only for the stat average (or point-buy cost, whatever) but both the high-low range within the 6 stats and the internal variance. For example 18-18-18-6-6-6 and 18-12-12-12-12-6 both have a 6-18 range and a 12.0 average but they're sure not the same.

That said, a few high stats mixed in with a few low ones is far more fun to play than all-medium...at least the forced array system mandates a range.

According to my app, the average minimum for any given character is 8, the high is 15. Limiting numbers to that range, the result is a little below 27 if you roll. If you follow the 3.5 point buy and allow 18s it's 31, if you ignore numbers below 8 it's a 32. That to me indicates that the high numbers are probably a bit too costly ... or that at a certain point you're just comparing apples to oranges.

My first reaction to this is that in order to better mimic the rolled results the point-buy total maybe needs to be jumped up to 31 or 32, and higher (and lower) numbers be allowed.

Maybe to avoid shenanigans there could be some riders attached: you may (and must) only have one single-digit stat, and no two stats may add to more than 31 (thus if you buy an 18 no other stat can be higher than 13; or 17-14, 16-15, etc.). With this, you could almost go to a choose-your-stats model.

Lan-"though part of the fun for me is the creativity of trying to come up with a character to suit what the dice give me to work with"-efan

 

[Tony Vargas]

But, ultimately, what's an un-verifiable, anonymous, on-line anecdote about events 30 or 40 years ago worth?

Quite a bit. I have enough D&D hours to qualify as an expert on D&D, and so have you and many others here.

So about 1 in 300 just for the two 18's. Add in the 17, the 16 and 2 14's and it's no wonder that in about a thousand characters I've never seen it and probably never will. It's not just unlikely, it's extremely unlikely.

But, since you put high value on unverifiable, anonymous, internet anecdotes, you're stuck with the fact that it has happened, it just happened to him, not to you.

And, really, with the random method, that's inevitable. The crazy results are going to happen to someone, and it's not even a little bit likely to happen to the someone it 'should' happen to...

 

[Maxperson]

But, since you put high value on unverifiable, anonymous, internet anecdotes, you're stuck with the fact that it has happened, it just happened to him, not to you.

Sure, but there's no use stressing over it. With around a 1 in 300 chance just to get two 18's, the odds probably jump into the thousands when you add a 17, a 16 and 2 14s.

And, really, with the random method, that's inevitable. The crazy results are going to happen to someone, and it's not even a little bit likely to happen to the someone it 'should' happen to...

Sure, then you shrug and move on. No use stressing over something that is incredibly unlikely to happen a second time.

 

[Captain Panda]

Wow, 49 pages of hostility. It's amazing how much pointless bickering over minutia people can engage in in this community.

If you want roughly equal PCs, use point buy. It's more balanced that way. If you want the world to have a bit more verisimilitude, rolls make for more organic players. Though the OP's problem is not rolling. It's that he didn't insist on witnessing and recording rolls himself.

 

[Oofta]

So, slightly more often than 1 in 300. Yeah, going by memory that sounds about right from what I've seen here...only we use 5d6k3; and out of maybe 1300 characters over 35 years of various games and campaigns I can only recall three or four* double-18s (and they do tend to stand out in the memory!)...and no triples.

* - ignoring one or two that I dead-shot know weren't entirely honest...

Now, to save me going back and doing lots of reading, am I right in remembering that the example you used involving you and your wife rolling stats included one of you rolling a double-18?

Reason I ask is if this was the case then it sounds like you hit the jackpot for stat disparity...no wonder you have a dim view of rolling. It's not always that bad...

Lanefan

The last time I rolled for stats my wife rolled poorly while another gal at the table got incredibly high results. We all rolled in person, roll 4d6 drop lowest. And yes, I agree it was highly unusual.

However, because my game was cancelled and it was an interesting challenge I did another variation of my program. I repeated my hell hound fight (1 lvl 4 PC vs a hell hound).


I then took a few examples of my "average different characters for a group of 6". Where I took the numbers from the lowest in the party, the highest in the party and then made the following assumptions

• Dwarven fighters because it's one of the better options for a low(er) stat character
• Dex is a dump stat
• AC 20 (plate + shield)
• Dueling fighting style (+2 damage)
• Level 4, max out Strength and Con instead of taking a feat

Just for clarity, the low stat PC is Larry, the high stat PC is Henry.

Larry: STR 20, CON 14, DEX 6, INT 10, WIS 9, CHA 11 PC: 51%
Henry: STR 20, CON 18, DEX 12, INT 13, WIS 14, CHA 12 PC: 63%
----------------------------

Larry: STR 18, CON 16, DEX 6, INT 11, WIS 11, CHA 8 Hell Hound: 68%
Henry: STR 20, CON 17, DEX 10, INT 12, WIS 14, CHA 15 Hell Hound: 54%
----------------------------

Larry: STR 16, CON 15, DEX 10, INT 12, WIS 12, CHA 12 Hell Hound: 73%
Henry: STR 20, CON 20, DEX 8, INT 12, WIS 13, CHA 15 PC: 55%
----------------------------

PC with lower points average win: 40%
PC with higher points average win: 55%

A point buy character (just for grins)
STR 18, CON 18, DEX 8, INT 12, WIS 12, CHA 9 Hell Hound: 52%

My conclusion? If you're a 4th level fighter that got separated from the party and run into a hell hound, run.

That and my sample size is too small ... maybe I need to tweak my program a bit. Hmm ... my wife is working today.

 

[Lanefan]

My conclusion? If you're a 4th level fighter that got separated from the party and run into a hell hound, run.

Er...if you're a 4th-level fighter Dwarf that got separated from the party and run into a hell hound running ain't gonna help very much: the hell hound can easily outrun you and probably dance a few circles around you in the process. This is true no matter what your stats are.

That and my sample size is too small ... maybe I need to tweak my program a bit. Hmm ... my wife is working today.

What would be more interesting, though probably impossible to simulate in any way other than playing it through, would be to take a party of 6 characters with a widely-disparate lot of stats between them and see what their individual long-term survivability looks like through a simulated long campaign. For more fun, repeat the simulation but each time vary the classes e.g. on the first run the lowest-stat guy is the Rogue, on the second run that set of stats goes to the Fighter, etc.

The 6 characters might for example start with:
#1 - 18-16-15-14-12-12
#2 - 18-15-12-11-10-7
#3 - 16-15-14-13-12-11
#4 - 15-14-13-12-10-8 (standard array)
#5 - 13-13-13-12-12-12 (bland boring average point-buy or rolls)
#6 - 14-12-11-9-8-6 (barely beats my bad-roll cutoff standards)

My own experience tells me there isn't much if any difference in their long-term survivability odds; but that's in a 1e variant game. I'm not sure how it'd play out in 5e where individual character death seems less likely.

Lanefan