The Gaming Den

But you still can do dumpstat 'that' if the grids line up properly. and there's a chance that you might not be able to play the character you want. (Which is why we banished 2nd edition to the Dead Zone)

This is the point-buy system the people I've been playing with have used. With 28 points, you can see why no one chooses 5d4x7.

-Crissa

Crissa wrote:With 28 points, you can see why no one chooses 5d4x7.

I sure can. Le's roll up a couple of stat lines:

10	8
9	15
14	12
13	11
14	15
13	13
11	15
Total:
15	24

The average point cost of a roll is pretty small. Half the rolls are worth 2 points or less, while your chances of exceeding a 16 on any score are less than 30%. That point cost system will give you all 14s and a 16, which is substanially better than you could expect to actually roll. Of course, that still doesn't equal being a 20 Int Wizard with a 14 Con that you can also mysteriously afford, but whatever.

-Username17

-Username17

But that's just ... the regular point buy with 10s already purchased. Nothing's fixed there, and in fact it's exacerbated by the ability to buy a 20 and sink even more resources into that single stat.

NineInchNall wrote:But that's just ... the regular point buy with 10s already purchased. Nothing's fixed there, and in fact it's exacerbated by the ability to buy a 20 and sink even more resources into that single stat.

Yeah, it's 40 point buy where you have to buy at least a 10 and you can buy above 18. The other alternative is a low-random die rolling scheme that will give a natural 20 somewhat less than one in very one thousand attempts.

-Username17

On the topic of balancing builds of characters,

How fair is 8's, or 13's at higher level, across the board, and 30 points to spend 'free buy' (i.e. 1 point for each +1 to an attribute), no stat going over 18, or 23 at higher levels

You can get a bunch of average stats, since spending points can get:

18, 16, 16, 14, 10, 8

One quirk of Frank's system is that average PC stats will go up as party size increases. So a group of 6 will have better average stats than a group of 2.

Josh_Kablack wrote:

kjdavies wrote:
This gives a mean of 12.5 (rather than the 12.24 you would expect with 4d6s3 * 6),
..
Keith

But the standard rolling method for D&D 3.x IS NOT JUST 4d6 DROP LOWEST (see 3.5 PHB page 8), so anything balanced against that average is statistically inferior to just having players roll and the right choice for any given player is always to roll randomly instead, which makes any such system worthless unless it's mandatory for all players. And as it results it lower numbers: making it mandatory is a nerf.

At least be honest that your system is resulting in weaker PCs on purpose.

How so? Lower mean value? Not usually, no.

It's New Years Eve (day), I'm at work, and I'm bored, so I wrote a script to exhaustively check. If I didn't make a mistake -- and I probably didn't, this is pretty simple -- there are 4,301,918,778,535,382,290 ways for 4d6s3 rolled six times to come up (in order; you can reduce this by a factor of 720 if you ignore permutations) such that they satisfy:

• at least one roll greater than 13
• sum of modifiers greater than 0

The mean value for acceptable sets is 74.5664. Divide by six to get the mean ability score value, which is a little under 12.43.

27-25-23 gives a mean of 12.5, before any bonuses such as the +2 I give to people who roll rather than choose (which bumps it to 12.83).

It's not a big difference, true (about 0.07 or 0.4 difference on average), but it certainly isn't worse than the average acceptably-rolled set. You *can* do better rolling if you get lucky because 27-25-23 constrains the total, but on average you'll probably be better off going with this.

I didn't examine what happens when you reroll sets that are valid according to 'PH3.5 pg 8' because it's entirely not what we're comparing to.

For reference, 4d6s3 results in 1296 possible rolls per ability score, of which there are 16 acceptable values in varying frequency. Given six ability scores, this results in 4,738,381,338,321,616,896 possible sets, of which 4,301,918,778,535,382,290 are acceptable -- almost 91%. The invalid sets that you are 'allowed to reroll' have very little statistical effect.


Keith

I think at one point I just started handing out 18, 16, 14, 14, 12, 10 for a few games. seemed to make quite a few happy enough. (especially the sole Paladin from a few years back)

Yeah, that 5d4 vs PB system proposed is fucking horrible in every detail and respect.

I do what frank does because players like rolling dice for charcter generation and I like fair character generation, BUT as my usual group is 3 to have a diversity of stats I give people 2 sets of rolls each. (so 6 sets to choose from)

The invalid sets that you are 'allowed to reroll' have very little statistical effect.

Well then, excuse my skepticism that dropping the lowest 9% out of the sample set was something that could just be ignored.

It's way better than that. While the effects on the "average" are not huge, the average doesn't mean shit. The statistical effects are titanic. The key is actually the ability to drop the statlines that don't have any features. Because features cost a lot of points in point buy, so the fact that you are absolutely guaranteed to get some is a big deal.

Even at 4d6, drop one, you have a 64.5% chance of getting a 13 or less on any particular die roll. So you have a 7.2% chance of getting a 13 or less on every die roll, which even if it "averaged" 12 or so, would still be a really shitty stat line and cost only 24 points. But you automagically get to drop all of those. And when you're absolutely guaranteed at least one that's 14+, you have to remember that more than 1 in 3 of those are 16+. And everything more than 14 costs substantial extra points.

Remember, the average value on the point buy system Crissa was dealing with is 12.16 - but since it's an 18, a 14, an 11, and 3 10s, that's a completely ridiculous piece of awesome, right? Having an average value of twelve point something doesn't tell you a lot. What matters is how much that average is being weighted by the higher rolls that will allow you to be a good spellcaster or rogue. And the rules in the PHB make sure that by definition the subtle increase in averages is coming from those higher numbers.

Just compare the random set of rolls I made on the first page. The first set was the worst one, it "costs" 25 points in Point Buy. The next set would have been even worse, costing only 19 points - but it gets rerolled automatically. The replacement set costs 33 points. The best set costs 39 points, and the last set costs 28. On four rolls without rerolling, the average cost would have been 27.75. With the rerolls, the average point cost is 31.25. So even though the "average" only changed by seven tenths of a point, the important part, the part where the attributes actually have different relative values, was skewed tremendously in favor of the PCs.

Dropping 9% of the sample size when it's the worst part of the sample has a huge statistical impact. Especially when the upper end of the sample size has such a hugely ballooning cost and value.

-Username17

Red_Rob wrote:One quirk of Frank's system is that average PC stats will go up as party size increases. So a group of 6 will have better average stats than a group of 2.

That's nothing. You should look at the TMNT rules for "We all share the same origins" character creation.

...and it was Frank who alerted me to that level of crazy, too.

Point buy - each character gets 85 points to spend on a 1:1 basis. Your stat max is 18 (before racial bonuses).

To help alleviate MAD, we allow a character to pick primary stats, with a "prereq" of providing the DM with quick flavor text about your character's concept and how the stat works.

In my games we use 18, 16, 14, 11, 10, 10. There has been some bitching amongst the players so I will probably bump the 11 to a 12 for the next CharGen.

I might also give Frank's method a try.

Or the one above.

Game On,
fbmf

For my current (gestalt) campaign I used what I call the Hero Array: 18 18 16 16 14 14

No mesure of balance has ever been heard of (well ... gestalt ...), but this works nicely with the inherently MAD gestalt.

NineInchNall wrote:What I don't understand about the elite array is the damn odd numbers.

It's supposedly the "most likely stat array you get from 4d6 drop lowest" (i.e. the expected value of the highest roll out of 6x(4d6L1) is around 15, the expected value of the 2nd highest roll out of those 6 is around 14, etc.).

Eh, I prefer to just do a point buy, it's simpler, but least when Frank does the idea of rolling, everyone can be held to the same standard, so don't have characters "genetically" superior than everyone else. Nowadays, I had realized I just really like PC's have at least an 18, so just decided to do a 28pt buy, except one of them is an 18, that cannot be broken down for extra points, so my starting array: 18,8,8,8,8,8. Otherwise for 4th edition, I basically did a 35pt buy where the attributes all start at 10 (technically a 37pt buy if want to get technical in how 4th edition starting array looks before points added), so ye can get: 18, 16, 14, 14, 10, 10 or 18, 16, 16, 11, 10, 10, whatever have ya.