Does 1d12+5 for ability scores work OK?

[Asisreo]

Everyone could start with 3d6 - 17 floor 1 so that way, everyone is balanced in relation to each other and the DM can easily challenge the players.

 

[Bardic Dave]

Everyone could start with 3d6 - 17 floor 1 so that way, everyone is balanced in relation to each other and the DM can easily challenge the players.

I prefer 4d6(drop one)-17 floor 1 myself. Rolling that extra die is just so satisfying!

 

[hbarsquared]

Since most dice sets have only one of each die, I often need to roll 1d4+1d6+1d8 to generate stats.

 

[Bacon Bits]

I'm a fan of using a deck of cards. Easy way to ensure consistent stats with radomness.

4,4,5,5,5,6,6,7,7,7,8,8 or 4,4,5,5,5,6,6,6,7,7,7,8 are good starting decks. Shuffle and deal into 6 piles of 2 cards. Then you can exchange any two cards.

You can make it high power just by adjusting the deck. 4,4,5,5,6,6,7,7,8,8,9,9 makes powerful characters.

 

[Mistwell]

I’m sure this has probably come up before but has anyone tried 1d12 +5 for rolling Ability Scores?

What is the downside of this?

Ps my Maths is atroash...atroch...attroshouce…..and so is my spelling

You would do better with 4d4+2, for a bell curve with a range of 6 to 18.

 

[Sword of Spirit]

I don't like random stats but if I had to, I would use playing cards.

Use the 2s-6s. Deal out into 6 piles of 3.

You can use as is or remove 2 of them so that everyone has the same sum. The downside as with most randomized methods is that it is possible to have a stat higher than 15. The likelihood of this is lower than other methods and having a high stat means others will be lower.

This is interesting, because the only way I ever do random scores is if everyone's stats add up to the same number. (No player should have an inferior character to their peers for the rest of the game because of a few random rolls in character generation. Same reason I don't roll hit points--random number generation is for temporary results, otherwise might as well roll randomly to determine you class, subclass, ASI, feats, spells known, etc--but at least with those others you theoretically get equivalent features.) Looks like this card system would be a fairly simple way to get random but equal stats.

What do you mean about "remove 2 of them"?

 

[Blue Orange]

As other people have said, the more dice the more bell-shaped the curve and the less likely extreme values are.

One interesting variant I haven't seen is rolling two dice of different numbers of sides: this produces a flat distribution at the center and bumps the ends of the distribution a little. For example, d10+2d4 gives a 3-18 distribution that's slightly more spread-out than 3d6 but less spread-out than 2d8+2.

 

[Flamestrike]

3d4+3 would be better.

More of a bell curve with ranges in the 7-15 region and an average score of 10.5.

You want to avoid any system that allows a roll of over 15 to start (i.e 4d6d1).

 

[Ancalagon]

Whenever how to pick ability scores comes up as a topic, I feel it is my sacred duty to pass on the wisdom of the superior method, the redric roller

Redrick Roller

This was inspired by an older EN world thread: D&D 5E - Array v 4d6: Punishment? Or overlooked data

He proposes keeping the rolls in order, but I think it's even better if you allow moving the numbers around (so no one is stuck with 8 con...). This way, you don't have full "control" like you do with point buy or stat arrays, but you also end up with PCs that are more closely balanced - I think some combo of numbers are slightly better than others, but no one is going to have god-like stats.

And finally, it makes the standard human, in some cases, a great pick.

 

[necrotizer]

This is interesting, because the only way I ever do random scores is if everyone's stats add up to the same number. (No player should have an inferior character to their peers for the rest of the game because of a few random rolls in character generation. Same reason I don't roll hit points--random number generation is for temporary results, otherwise might as well roll randomly to determine you class, subclass, ASI, feats, spells known, etc--but at least with those others you theoretically get equivalent features.) Looks like this card system would be a fairly simple way to get random but equal stats.

What do you mean about "remove 2 of them"?

You would have to "remove 2 of them" because if you used every suit for cards 2 through 6 you would have 20 cards (4 times 5 equals 20). To sort into 6 equal stacks of 3 (per his suggestion), you need to have 18 cards.

You could remove one of the 6's and one of the 2's, for example, and the total ability score for every player would always be 72. I simulated this earlier, however, and found that it results in disappointingly average ability scores. You might have better luck removing two of the fours.

But if I were to run a playing card method, I think I would do something like what @Bacon Bits said: allow the player to swap one pair of cards between stacks.

I like the look of this:

1. Player receives a stack of 18 playing cards containing: 3 sixes, 3 fives, 4 fours, 4 threes, and 4 twos.
2. Randomly deal the cards into 6 piles (3 cards each) to generate 6 numbers.
3. Before totaling the value of each pile, the player may choose any two cards to swap.

This generates a range of 6-18. The sum of all ability scores will be 69 (i.e., 11.5 average), but it has a (slightly greater than) 50% chance of allowing the player to swap for (at least) a 17.