Author Topic: Optimization by the Numbers: Discussion  (Read 6230 times)

Offline Agita

  • He Who Lurks
  • Moderator
  • *****
  • Posts: 2705
  • *stare*
    • View Profile
Optimization by the Numbers: Discussion
« on: February 19, 2012, 05:27:08 PM »
This thread is for discussing Optimization By The Numbers. Post any broken formatting or missing symbols that need fixing here, or else discuss the available data and T_G's conclusions from it.
Please send private messages regarding board matters to Forum Staff instead.

Offline Maat Mons

  • DnD Handbook Writer
  • ****
  • Posts: 1203
  • What is a smile but a grimace of happiness?
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #1 on: February 19, 2012, 06:56:47 PM »
I think it'd be better if the table was actually a table.  Like this:

(click to show/hide)

Offline Agita

  • He Who Lurks
  • Moderator
  • *****
  • Posts: 2705
  • *stare*
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #2 on: February 19, 2012, 07:09:38 PM »
I think it'd be better if the table was actually a table.  Like this:
Thanks, I'll add that. I already made it an actual table in the pdf version, so I was too lazy to code it for the thread too. :p

EDIT: Or not... the fully-formatted table breaks the character limit wide open. I'll add it in a new post instead.
EDIT2: Done.
« Last Edit: February 19, 2012, 07:15:07 PM by Agita »
Please send private messages regarding board matters to Forum Staff instead.

Offline Tubercular Ox

  • Lurker
  • *
  • Posts: 37
  • I'm new! And always will be!
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #3 on: April 06, 2012, 09:32:27 PM »
One thing I never understood about this analysis:  Shouldn't the average saves of monsters be bi-modal?  All the monsters based on types with Good fort saves should average to a higher level than monsters based on types with not-Good fort saves.  In the worst case scenario (a 50/50 split between good fort monsters and bad fort monsters) averaging the two groups together creates a number that represents neither group adequately.

If all you have is a fort-based attack, that's fine, but if you have one attack for each defense type, remembering which monster types have inherently weak Fort saves (or inherently strong Ref saves) may skew the analysis a bit.

If the distribution *isn't* bimodal, it requires a bit of an explanation why, and before I read the entire monster manual myself, I was wondering if anyone had a ready one available?  It seems a bit of a cop-out if any monster with a bad Fort save automatically gets a huge Con bonus or something.

Offline Agita

  • He Who Lurks
  • Moderator
  • *****
  • Posts: 2705
  • *stare*
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #4 on: April 07, 2012, 05:31:09 AM »
I believe the main reason is practicality. Theoretically, one should also distinguish even more finely on, say, attack bonuses (you've got weapon-users, which will generally have higher attack bonuses, pouncers, which will generally have lots of attacks at low bonuses, casters, which won't use their attack bonuses at all in most cases...). It would no doubt be even better if the difference was taken into account. The issue is the effort to return ratio isn't very good.

For what it's worth, in the case of saves specifically, consider this: If your save DCs can reliably affect the average save at a given CR, then chances are you'll always be targeting an average or below-average save if you can pick and choose which ones to target.
« Last Edit: April 07, 2012, 05:32:45 AM by Agita »
Please send private messages regarding board matters to Forum Staff instead.

Offline Tubercular Ox

  • Lurker
  • *
  • Posts: 37
  • I'm new! And always will be!
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #5 on: April 07, 2012, 02:45:54 PM »
The difference between attack and saves is that a monster should always be choosing to use his best attack, but the player has the choice of targeting their weakest save. 

If your save DCs can reliably affect the average save at a given CR, then chances are you'll always be targeting an average or below-average save if you can pick and choose which ones to target.

That's the part that's wrong.  In a bimodal distribution, if you stick with one save, chances are you'll always be targeting an above-average or below-average save.  The chances of targeting an average save should be rather slim.

Let me throw out some complete speculation.  20th level, the effect is most extreme.  The spread between a strong save and a weak save is 6 points (+12 vs. +6)  If the MM has 50/50 strong/weak saves, then monsters with strong saves are going to average 3 points higher than average, and monsters with weak saves are going to average 3 points less than average.  Same argument goes for the spread at any level.  Plus or minus 2 at level 10, plus or minus 1 at level 1.

I did some flipping through the SRD and the results are as expected but sometimes weird.  Elementals generally have a strong save WAY above average, and a weak save WAY below average (strong/weak depends on elemental type).  Giants (w/ strong Fort saves) had fort saves way above average, ref saves way below average, but will saves spot on average.  Given that both Giants and Elementals have strong saves way above average, I expected Devils to be depressingly hard to tag... but it turns out, despite having all strong saves, their averages were generally only the expected amount above average for each save -- one or two points depending on CR.  This was so weird to me I kept looking: Lycanthropes (statted up as Warriors in the SRD -- strong Fort saves) fit the pattern as predicted: Fort saves above average, will/ref saves below average.  Fey, OTOH (with weak Fort saves), have (as expected) Fort saves that are weaker than their ref/will saves, despite Fort saves having the highest average on the chart.

So... I'm wrong.  Giants and Elementals show that individual influences other than base save bonus can easily overwhelm differences created by strong/weak base save types.  But the Fey example shows that loading out on ref/will saves and expecting the average case to always turn in your favor doesn't work. 

I don't think anyone actually needed a three paragraph analysis to reach that conclusion.  I had fun doing it though...

Offline Agita

  • He Who Lurks
  • Moderator
  • *****
  • Posts: 2705
  • *stare*
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #6 on: April 07, 2012, 03:00:25 PM »
Those discrepancies can be at least partially explained through the amount of HD each creature has. Elemental and Giant hit dice suck, so Elementals and Giants generally have a lot of HD for their CR, whereas Outsiders have great HD and generally awesome abilities on top of that, so their HD are a lot closer to their CR. As an example, an Elder Earth Elemental is CR 11, but has 24 HD, so the relation between its good and bad saves is accordingly blown up. A Barbed Devil, by comparison, is also CR 11 but has only 11 HD, so its Fort save is accordingly lower than the Elemental's.
Basically, there's huge loads of factors, so that it's just not very practical to try and do a bimodal distribution. What the list probably should include, however, is some measure of variance for each value, so that one might get a sense of how much individual values may deviate from the average.
« Last Edit: April 07, 2012, 08:59:18 PM by Agita »
Please send private messages regarding board matters to Forum Staff instead.

Offline Tubercular Ox

  • Lurker
  • *
  • Posts: 37
  • I'm new! And always will be!
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #7 on: April 07, 2012, 08:48:20 PM »
Thank you.  The bit about hit dice was the piece that I was missing that makes all the numbers make sense for me.  I just don't pay enough attention to spot those things on my own.

Offline Sinfire Titan

  • Hustler 3
  • Retired Admin
  • *****
  • Posts: 1443
  • You have one round to give a rat's ass.
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #8 on: August 13, 2016, 06:55:08 PM »
Looking at the numbers what, if any, WotC books are missing from the chart's averages?
Concerned about how moderation works here? Please PM this account.

Offline Agita

  • He Who Lurks
  • Moderator
  • *****
  • Posts: 2705
  • *stare*
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #9 on: August 14, 2016, 06:39:19 AM »
The main table was compiled from all monsters in the SRD, which means it would contain the first Monster Manual, the Expanded Psionics Handbook, and the Epic Level Handbook. The linked second chart incorporated more sources, but it's been a dead link for years - since well before I ported this.
Please send private messages regarding board matters to Forum Staff instead.

Offline Sinfire Titan

  • Hustler 3
  • Retired Admin
  • *****
  • Posts: 1443
  • You have one round to give a rat's ass.
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #10 on: August 14, 2016, 08:00:47 AM »
Looks like I have a project then.
Concerned about how moderation works here? Please PM this account.

Offline Agita

  • He Who Lurks
  • Moderator
  • *****
  • Posts: 2705
  • *stare*
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #11 on: August 14, 2016, 09:49:52 AM »
If you plan on compiling additional sources and are up for some extra work, I'd recommend also dividing monsters into somewhat heuristic categories like caster monsters, critters with middling AC and lots of hp and vice versa, and things like that. When you have monsters that are really hard to hit but don't take many hits to go down and average them with easy to hit monsters that take a lot of damage, the averages are liable to not reflect the actual cases very usefully.
Same with monsters that don't necessarily have high AC or high hp but benefit from a miss chance - it's misleading to count a displacer beast as having AC 16 and 51 hp, for example. Same with certain gimmick monsters whose toughness is greatly dependent on whether you use a certain mode of attack or not, like regenerators or incorporeal creatures. Those either need some kind of approximate weighting to account for those factors or should probably be ignored as outliers.
Please send private messages regarding board matters to Forum Staff instead.

Offline Sinfire Titan

  • Hustler 3
  • Retired Admin
  • *****
  • Posts: 1443
  • You have one round to give a rat's ass.
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #12 on: August 14, 2016, 06:21:50 PM »
If you plan on compiling additional sources and are up for some extra work, I'd recommend also dividing monsters into somewhat heuristic categories like caster monsters, critters with middling AC and lots of hp and vice versa, and things like that. When you have monsters that are really hard to hit but don't take many hits to go down and average them with easy to hit monsters that take a lot of damage, the averages are liable to not reflect the actual cases very usefully.
Same with monsters that don't necessarily have high AC or high hp but benefit from a miss chance - it's misleading to count a displacer beast as having AC 16 and 51 hp, for example. Same with certain gimmick monsters whose toughness is greatly dependent on whether you use a certain mode of attack or not, like regenerators or incorporeal creatures. Those either need some kind of approximate weighting to account for those factors or should probably be ignored as outliers.

My ability to sort those out may be limited to just listing a few extra numbers. I also wanted to look into Attacks/round, Damage, and Ability Scores.
Concerned about how moderation works here? Please PM this account.

Offline SorO_Lost

  • Epic Member
  • ****
  • Posts: 7197
  • Banned
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #13 on: August 14, 2016, 07:19:06 PM »
While you're at it I'd love to have a truly definitive answer on energy immunity/resistance. And how about a DR addition too, it'd compliment the HP/AC figures to figured out how much you'd deal on average and if you grab their bypasses maybe we'll see if Cold Iron is better than Silver and exactly how effective DR/Magic is.

Offline Sinfire Titan

  • Hustler 3
  • Retired Admin
  • *****
  • Posts: 1443
  • You have one round to give a rat's ass.
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #14 on: August 14, 2016, 11:00:10 PM »
While you're at it I'd love to have a truly definitive answer on energy immunity/resistance. And how about a DR addition too, it'd compliment the HP/AC figures to figured out how much you'd deal on average and if you grab their bypasses maybe we'll see if Cold Iron is better than Silver and exactly how effective DR/Magic is.

Good suggestion.
Concerned about how moderation works here? Please PM this account.

Offline awaken_D_M_golem

  • Epic Member
  • ****
  • Posts: 7639
  • classique style , invisible tail
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #15 on: August 15, 2016, 04:08:21 PM »

Looks like I have a project then.

 :thumb ... :bow


Hard monster numbers , for the win.
Your codpiece is a mimic.

Offline awaken_D_M_golem

  • Epic Member
  • ****
  • Posts: 7639
  • classique style , invisible tail
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #16 on: August 16, 2016, 04:18:54 PM »
I can't find the googledoc or in wayback. 
It might be inaccessible by now, or my google-fu has failed++.


site:realmshelps.net "cr 4" ... got 164 results.
site:realmshelps.net "cr 19" ... got 67 results.
site:realmshelps.net "cr 1/6" ... only got 5.


I'm pretty sure this is the book list they use:
http://www.realmshelps.net/books/books1

So anyway, that might be easiest access.
Your codpiece is a mimic.

Offline Garryl

  • DnD Handbook Writer
  • ****
  • Posts: 4503
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #17 on: August 18, 2016, 01:47:46 PM »
I compiled a lot of the numbers for Monster Manual skeletons and zombies here for a different project. HP, BAB, and saves are all based purely on HD, and initiative's purely a function of Dexterity (adjusting for whether it's a skeleton or a zombie), so they aren't in there. Touch and flat-footed AC aren't listed, but you can set up a formula easily enough based on the final AC, size, and the base Dex score if you want to calculate it.

Offline Chemus

  • Legendary Member
  • ****
  • Posts: 1929
    • View Profile
Re: Optimization by the Numbers: Discussion
« Reply #18 on: August 20, 2016, 03:35:57 PM »
The main table was compiled from all monsters in the SRD, which means it would contain the first Monster Manual, the Expanded Psionics Handbook, and the Epic Level Handbook. The linked second chart incorporated more sources, but it's been a dead link for years - since well before I ported this.

Is this what you wanted?

Edit: Nemmind; it doesn't do anything, though the creator's contact info is at the bottom:
Quote
'jose dot jtp at gmail dot com'. Please include CritterFilter in the title so I know it's not spam

I'm gonna dig a little to see if I can get at the js's, but I'm kinda ignant on how pages are scripted.

Edit2: I'm unable to get the scripts; WB appears to only have the archive of the page and little to no resources. Gonna see what google has. Edit2Edit: Which is exactly nothing.
« Last Edit: August 20, 2016, 04:15:53 PM by Chemus »
Apathy is ...ah screw it.
My Homebrew