NPC Best Practices: Reflavor Existing or Build and Estimate CR?

[Urpriest]

DMs, in your experience what is the best way to build combat stats for NPCs? In 4e I'd typically take an existing monster and refluff it, but in 5e the DMG suggests you can build an NPC essentially like a PC, and then use the monster rules to estimate its CR. When you need to stat up an NPC, which do you typically do? Do you use the monster math from the DMG, or the blogofholding version (http://blogofholding.com/?p=7338)?

[MrStabby]

I use PC classes as inspiration, but never explicitly build an NPC as a PC.

I might take an NPC and give it some sneak attack type damage and a cunning action type ability to capture some roguishness but I would use different damage and HP (make more defensive, less damaging) than one would expect for a given level.

Then I never bother with CR. I have the NPC I want - I don't need CR. Also the MM way of generating CR is less useful than just eyballing how hard something is. There is so muchmissed in the MM that it gives wildly different answers to what you might expect in reality.

[Marcloure]

Take an existing monster, tweak its stats, add new features / remove the ones that doesn't fit. I also usually add some new actions that I think captures the creature's fantasy better than just attacks.

[LordShade]

I usually start with a monster statblock and make adjustments. If an NPC were going to join the party, however, I would build something off of a PC class progression (but probably with fewer abilities).

[MaxWilson]

DMs, in your experience what is the best way to build combat stats for NPCs? In 4e I'd typically take an existing monster and refluff it, but in 5e the DMG suggests you can build an NPC essentially like a PC, and then use the monster rules to estimate its CR. When you need to stat up an NPC, which do you typically do? Do you use the monster math from the DMG, or the blogofholding version (http://blogofholding.com/?p=7338)?

I normally build like a PC, and then ignore all of the CR math and give out XP as if it had CR = level.

(Or if it's a monster like a dragon that already has a CR, I eyeball it. Sometimes I take that amount of XP and add it on top of the existing XP.)

Key insight here: players don't mind when you're generous. At the same time, I genuinely feel that CR = lvl is a pretty fair ballpark estimate in the sense that between a level N (N)PC and a typical MM CR N monster, it's hard to predict who would win in a fight, because (N)PCs have lots of strong abilities that don't show up on the CR table, like Uncanny Dodge and Battlemaster maneuvers and Dimension Door.

So if I give out 10,000 XP for facing a 13th level enemy NPC, at worst I'm being generous and at best I'm being fair.

The Blog of Holding math cannot be trusted BTW, it's built on faulty assumptions.

[Rukelnikov]

I usually use build as PC for important humanoid NPC characters that I expect will appear in multiple encounters.

If playing with Feats and MC I usually add the monsters proficiency bonus to AC, and 1 or 2 hp/hd. I do not consider them higher CR though.

When making a new monster, most of the times I take a statblock that I like and add it some unique ability, or take a monster with a cool mechanic and change its stats to fit the levels I'm going for. I haven't given XP in years so I only need a rough estimate of the creatures CR for encounter difficulty reasons, and for that the CR of a somewhat similar stated minion is usually enough of a guideline

Adendum, in general its better to use your own judgement to decide how challenging a monster will be, not only because many of the ratings in the manuals feel off, but also because different parties will be more or less challenged by different abilities, against a ranged heavy party, a flying speed won't mean that much, against one with limited ranged prowess it will make a lot of difference.

[quindraco]

DMs, in your experience what is the best way to build combat stats for NPCs? In 4e I'd typically take an existing monster and refluff it, but in 5e the DMG suggests you can build an NPC essentially like a PC, and then use the monster rules to estimate its CR. When you need to stat up an NPC, which do you typically do? Do you use the monster math from the DMG, or the blogofholding version (http://blogofholding.com/?p=7338)?

The problem with the monster math from the DMG is that very few monsters match their own CR, and WOTC has no apparent interest in addressing the dichotomy. For example, Guards math out to CR 1/4, but are listed at CR 1/8.

I appreciate that link you provided based on backsolving CRs assuming the MM is correct, rather than the DMG. I don't honestly know which is more accurate - I can tell you with certainty the DMG rules are absurd and unlikely to have any basis in reality. For example, the DMG monster math only has CR increase for higher saves based on how many saves the monster is proficient in, completely ignoring the actual size of its proficiency bonus or how good its stats are. That's one of the reasons Intelligence never impacts defensive CR - being better at a save doesn't matter if what made you better at it was a higher ability. Intelligence only impacts CR when a creature uses it for attack or defense (such as when it relies on save DCs based on its Intelligence).

You should generally not make NPCs as PCs and then backsolve their CR, though, because NPCs fundamentally live and operate in a different world from PCs - it's common and typical for NPCs to have abilities PCs don't have and can't get, and the PCs at your table basically have to carefully avoid thinking about it, lest they find a bunch of tribal warriors and plead to be taught their incredibly effective combat style. Or gladiators. Or whatever. 5E takes a lot of cognitive dissonance to play. Your best bet is modifying existing monsters.

EDIT: Geryon's DPR is higher than the 97 the blog of holding link claims, so I'm not sure how they analysed him. Even if you ignore his additional DPR against targets restrained by his claw, and you assume his lair blast only hits one target, he deals a baseline of 105: 23 from his claw, 27*2 from his stinger twice (once on his turn and once at the end of a player turn), 28 from his lair blast. Since that's 8 more than 97, I don't know how they got there - probably a division operation that got floored, I'm guessing. If you assume a restrained target stays restrained, Geryon's DPR goes up by 27 on his second and third rounds of combat, which is +54/3 = +18 DPR. If you assume he only manages the +27 once, that's +9 DPR. The DMG suggests (under the way to handle draconic breath weapons) you should assume an AOE hits two targets, as well, increasing his DPR by 28 (from the second target of the lair blast).

Geryon also has more than 300 effective hit points - even if you disregard the DMG's methods for handling traits, his regeneration absolutely behaves the way the DMG suggests it should, which is that it stacks with his real health in terms of how hard he is to kill (the DMG suggests using the three round rule, so his regeneration brings his HP to 360). I'm skeptical of the blog's methodology now - I'm concerned how many monster special qualities they ignored that change the numbers in the statblocks, even if you stick to only the obvious ones.

[MaxWilson]

For example, the DMG monster math only has CR increase for higher saves based on how many saves the monster is proficient in, completely ignoring the actual size of its proficiency bonus or how good its stats are. That's one of the reasons Intelligence never impacts defensive CR - being better at a save doesn't matter if what made you better at was a higher ability. Intelligence only impacts CR when a creature uses it for attack or defense (such as when it relies on save DCs based on its Intelligence).

Another interesting oddity is that Legendary Resistance affects CR, but you can tell from both the DMG text and from e.g. air elementals that having a metric ton of condition immunities does not affect CR.

The occasional discrepancies between DMG CR and MM CR can be attributed to the final step in the CR calculation process, "tweak based on playtesting" (and implicitly, on DM judgment). But they are less irksome than the enormous number of things where no DMG guidance is given nor any methodology from when guidance can be derived. If I want to check the CR of a Gorgon, how does the petrifying breath affect its CR? The DM just has to eyeball it, and of course for save-or-lose abilities like that there really isn't a single CR number that expresses its that level, because CR is a proxy for HP and DPR, and petrification breath bypasses HP.

TL;DR the occasional discrepancies aren't a problem but the holes are annoying and have to be eyeballed.

[Unoriginal]

Reflavoring/adjusting existing NPCs is the simplest method, creating them with the NPC rules can work out, but IMO using PC creation rules for NPCs is much too much work compared to the results, even in the cases where it actually works out.



The Blog of Holding math cannot be trusted BTW, it's built on faulty assumptions.

Would you mind expanding on that a bit more? It's the first time I hear about that blog.

[EggKookoo]

The PC chassis is best used for creatures that need a smooth progression of power and ability. Very few NPCs fall into that category. Even if an NPC is going to "power up" and show up later in the campaign, you don't need the kind of interpolated gradient that PC leveling provides. Just rebuild the creature at a higher CR.

If you're going to make an NPC that levels smoothly the way a PC does, and assuming you're aware of the pitfalls of running a DMPC, it's arguably better to base one off the sidekick template than a true PC template, although to be fair there's not much difference there.

Regarding wholesale creation or modification of existing NPCs/monsters, once you get a firm handle on how to make NPCs, there's really not much difference. I do both, and to me the two approaches blend into each other.

[quindraco]

Doing my best to faithfully interpret the DMG's rules for calculating CR when applied to Geryon, he's CR 23 (and it doesn't matter for this if his Lair hits 1 target or 2 - that raises his OCR 1 step from damage and then lowers it 1 step because it changes his attack bonus gap from expected from an even number to an odd number), for what it's worth, and that's without letting him summon minotaurs:

HP: 300 + 10*3 (Legendary Resistances) + 20*3 (Regeneration) = 390
AC: 19 + 1 (Constrict on his claw) + 2 (Magic Resistance) + 2 (Invisibility as an at-will action) + 2 (Proficiency in 4 saves) = 26
DCR: 23

Attack Bonus: 16
DPR: 27*2 (stinger) + 23 (claw) + 18 (assuming one target is restrained by the claw for 2 rounds, copying the DMG rules for Swallow) + 28 (Lair action cold damage) = 123; it's 151 if you assume the Lair blast hits two targets, but due to the attack bonus, the OCR is the same either way.
OCR: 23

What's really bizarre about this calculation is that Constrict is worth +1 to AC only, even though he has the attack volume to attack a constricted target (with both of his stinger attacks), and he can fly and teleport, meaning he can boost his DPR by dropping PCs.

[Urpriest]

The Blog of Holding math cannot be trusted BTW, it's built on faulty assumptions.

Unoriginal already asked, but FWIW I'm also curious about this. Do you just mean something like what quindraco pointed out, that they flubbed the DPR calculations in some cases? If so, how prevalent is that? Most of the areas where he actually came to firm conclusions were for low-level monsters where I would assume this isn't that easy to get wrong.

Thanks for the feedback all. Light tweaks/reflavorings of monsters is what I'm used to doing from 4e, so it's good to know that's a popular method. I'm not sure I'd be comfortable just 100% homebrewing a monster/NPC in this edition if the math balancing them is as controversial as people are saying.

[KorvinStarmast]

Take an existing monster, tweak its stats, add new features / remove the ones that doesn't fit. I also usually add some new actions that I think captures the creature's fantasy better than just attacks. That's my method, now that I've messed about with other trials. This is faster/more efficient. :smallcool:

[MaxWilson]

Would you mind expanding on that a bit more? It's the first time I hear about that blog.

Sure. The Blog of Holding is famous for some strong claims, like "Based on my number crunching, it looks like the DMG’s central monster creation chart, “Monster Statistics by Challenge Rating”, isn’t in line with the Monster Manual, and if you try to follow it you will get monsters that don’t look much like Monster Manual monsters." (Source: http://blogofholding.com/?p=7283)

But the methodology doesn't actually back up this claim. Notably, Blog of Holding doesn't examine a single MM monster's DMG CR to see whether it matches up with its MM CR! Instead it eyeballs a bunch of correlations between individual stats, like AC vs. DPR, to see if it can find anti-correlations.

Another example of Blog of Holding logic: "In my last post, I demonstrated, via too many charts, that the monster-creation math in the 5e Dungeon Master’s Guide doesn’t match the monsters in the Monster Manual. As an example: the average CR 1/4 monster has 13 hit points. The DMG suggests that they should have 43 hit points. Imagine your level-1 party mobbed by skeletons with 43 HP each instead of 13!"

That isn't what the DMG says at all of course. It says that if it has 13 HP, it should be about CR 1/4 (if its offensive output ad AC are also in the CR 1/4 range), and of course the MM shows that this is pretty much true. CR is an output from the monster design process, not an input.

The absolute most wrong thing about Blog of Holding's logic is in this post (http://blogofholding.com/?p=7357), where they pass along an interesting and quite useful factoid from JeremyCrawford and then reject it because it doesn't fit their preconceptions. Here's the key excerpt:


Mike asked Jeremy about the table in the DMG for homebrewing monsters. It’s public record that I have my doubts about that table. Jeremy provided this interesting fact: apparently, the canonical formula for determining monster CR is encoded in an internally-used Excel spreadsheet. (We’ve seen this spreadsheet in action in the Mike Mearls Happy Fun Hour.) The table in the DMG was made after the spreadsheet, and was an attempt to reverse-engineer and simplify the spreadsheet’s formulas for DM’s home use: it’s not used as part of the process for creating for-publication monsters.

My research indicates that while D&D monster stats are internally consistent and carefully designed, their power levels don’t line up well with the DMG chart. This leads me to believe that the spreadsheet does what it intends to do, while the DMG chart does not.

I invite you to ponder this further mystery from the Deep Dive podcast:

In the podcast, Jeremy Crawford talked about how the D&D team approaches what he calls “action denial” attacks: paralysis, charm, etc. This is something I’ve wondered about. My initial research suggested that these attacks didn’t have much of an effect on CR, which I though was strange: taking out a combatant for a few turns seems like it should be a powerful ability.

Jeremy used paralysis as an example to illustrate the team’s approach. First, find the lowest-level spell which inflicts a condition. For paralysis, that’s Hold Person. Next, you translate that into damage by finding the damage output of the simplest pure-damage spell of that level. Hold Person is level 2, and its comparable damage spell is Scorching Ray, which does 6d6 (21) damage. Thus, the ability to paralyze is worth 21 “virtual” damage for Challenge Rating calculation purposes.

I see the logic in that, but it doesn’t scale the way I’d expect it to. If you’re fighting four opponents, I’d expect the ability to keep one of them paralyzed to be worth, say, 33% extra virtual hit points (the hit points you won’t be losing to the paralyzed opponent) – or some similar scaling benefit. If the ability is expressed as flat damage, 21 hit points damage is dominant at low levels and negligible at high levels.

And indeed, when checked against monster data, it doesn’t seem as if monsters are balanced exactly as Jeremy describes. Check out the chart below. This includes all ten CR-5 or lower monsters with a paralysis attack: the carrion crawler, chuul, ghoul, ghast, grell, pentadrone, mummy, revenant, thri-kreen, and yeti. (Above CR 5, the 21 virtual hit points of paralysis would rarely be relevant to CR because monsters generally have a more damaging attack options.)

...The green triangles represent the average damage of paralyzers, once you take the 21 points of “virtual damage” into account. As you can see, adding this would skew monster damage way too high – doubling their effective damage in most cases! If this damage were really being taken into account the way Jeremy describes, all these low-level monsters would have to have very low-damage attacks – a ghoul with a claw doing 1 damage, for instance – to make room for all that virtual damage.

This is one of those unaccountable situations, like many others that we’ve discovered while investigating how monsters are built, where the WOTC instructions for building a monster don’t match the monsters we see in the manual.

Of course you can see how the bolded section is fallacious. The monsters don't have to have low-damage attacks; they just have to have low damage OR low AC OR low HP so that the total CR comes out right.

Take the Carrion Crawler for example as an iconic paralyzer. With AC 13, 51 HP, and only 11 damage at +4 to hit on its most damaging attack (or DC 13 on its primary save), it only earns an offensive CR of 1 and a defensive CR of 1/2, which leaves its CR 2 unexplained. But if we add in its paralysis, according to the method Crawford mentioned, then we roughly equate paralysis to Scorching Ray (21 damage) or to a generic 2nd level spell (3d10) on DMG pg. 284, giving it a damage of either 32 or 26. Assuming 3d10 (26 total) gives offensive CR 3, defensive CR 1/2, for an average CR of 1.75, approximately 2. Assuming Scorching Ray (32 total) gives offensive CR 4, defensive CR 1/2, for an average CR of 2.25, again approximately 2.

This means that the most useful factoid on the Blog of Holding is one which Blog of Holding itself dismisses! Special abilities do affect CR, and this is roughly how WotC's internal spreadsheet works: convert special abilities to damage equivalents by analogy to spells.

If Blog of Holding actually bothered to test its own hypothesis by looking at individual monsters they might have realized that, but instead they just used the deeply fallacious approach of looking at parameters pairwise and eyeballing correlations when it's a four-or-more-parameter calculation.

[Unoriginal]

[...]

This means that the most useful factoid on the Blog of Holding is one which Blog of Holding itself dismisses! Special abilities do affect CR, and this is roughly how WotC's internal spreadsheet works: convert special abilities to damage equivalents by analogy to spells.

If Blog of Holding actually bothered to test its own hypothesis by looking at individual monsters they might have realized that, but instead they just used the deeply fallacious approach of looking at parameters pairwise and eyeballing correlations when it's a four-or-more-parameter calculation.

Thanks for explaining. It does indeed seems that the author didn't actually do the calculations they say they did, at minimum.

[da newt]

For the OP - do you mean how do you build an person to be a foe to the party / PCs in combat to ensure you balance the difficulty of combat and reward XP correctly?

I ask because for some folks NPC = a non-player character that the DM plays to help the plot along.

It sounds like you are asking about building a humanoid with PC like class / features to act as a BBEG or other foe.

[quindraco]

That info about setting conditions equivalent to damage based on spell level is fascinating, and explains a lot. Let's get some spell level equivalents going.

Deafened: 2 for Blindness/Deafness
Blinded: 1 for color spray (resisted with hit points, duration 1 turn) or 2 for blindness/deafness (resisted with save, duration 1 minute no concentration)

Charmed: 1 for humanoids and beasts (charm person and animal friendship), 3 otherwise (hypnotic pattern)
Poisoned: 1 (Ray of Sickness)
Frightened: 1 (cause fear, wrathful smite)

Incapacitated: 1 (Tasha's Hideous Laughter)
Stunned: 5, somehow (technically Contagion can cause it, but there are extra steps; at level 7, Divine Word and Symbol can both do it, but Divine Word is resisted with both save and then hit points, and Symbol takes a minute to cast; the lowest level spell that lets you simply cast a spell as an action and the target(s) save or stun is level 9. Psychic Scream), despite being not as bad as Paralyzed
Paralyzed: 2 (Hold Person - following Charm's precedent, humanoid restrictions are fine. 5 for Hold Monster)
Petrified: 6 (Flesh to Stone)
Unconscious: 1 (Sleep, hit point resist) or 6 (Eyebite)

Grappled: No spell.
Prone: 0 (Sapping Sting)
Restrained: 1 (Entangle)

Exhaustion: 4 (sickening radiance, wears off)

This list has some serious absurdity to it - Blinded is a lower level than Deafened, and Unconscious is lower than Paralyzed which is lower than Stunned which is lower than Petrified, even though the approximate order of how bad they are for you is Stunned < Petrified < Paralyzed < Unconscious. You can help the list by removing hit point resists, which raises blinded to 2 and unconscious to 6, but the incapacitated...unconscious chain is still messed up.

[EggKookoo]

It is interesting, isn't it?

So an Intellect Devourer's Devour Intellect attack has a DC of 12 and can apply a stun effect. Stun is the equivalent of 5th level (as weird as that seems). The DMG says a 5th level spell deals 44 damage to a single target. That damage is CR 6, but the DC of 12 brings it down to 4. The Devourer's defensive CR is 1/8, so we end up with something slightly over CR 2 in the end.

I think Ability score damage should factor in more, but it does add up.

Edit: Wait, I'm wrong. The Devourer's offensive CR really should be 5 if the above is correct. That results in an average CR of just over 2.5, which I would tend to round to 3. It gets worse if we're meant to combine the stunned-equivalent damage and the 11 points from Devour Intellect. That puts its combined offensive CR to 6, which solidly puts the Devourer's overall CR to 3. Which, honestly, is what the thing feels like...

[MaxWilson]

Edit: Wait, I'm wrong. The Devourer's offensive CR really should be 5 if the above is correct. That results in an average CR of just over 2.5, which I would tend to round to 3. It gets worse if we're meant to combine the stunned-equivalent damage and the 11 points from Devour Intellect. That puts its combined offensive CR to 6, which solidly puts the Devourer's overall CR to 3. Which, honestly, is what the thing feels like...

Sold! You've persuaded me. Remind me to give out 700 XP next time my players run into an Intellect Devourer. It's still a nasty, deadly little creature, but you're right, they at least deserve more XP.

[Urpriest]

For the OP - do you mean how do you build an person to be a foe to the party / PCs in combat to ensure you balance the difficulty of combat and reward XP correctly?

I ask because for some folks NPC = a non-player character that the DM plays to help the plot along.

It sounds like you are asking about building a humanoid with PC like class / features to act as a BBEG or other foe.

Indeed, sometimes I forget that 5e distinguishes NPCs for use out of combat from humanoid monsters for use in combat. I mean the latter, yeah. I'm used to 3.5 where everybody has combat stats in principle, so they're the same thing. (And in 4e the question is moot because combat encounters are always reflavored or homebrewed monsters, there's no other way to build them. 5e is unusual in that the rules suggest you can do either.)

So for example, the last time I ran a game in 5e there was a fight with a necromancer at the end of the dungeon. I couldn't find a monster in the MM that felt suitable, so I did what the DMG suggested and built a Wizard using the PC rules, then estimated the CR, tweaking the level until it was expected to be an appropriate challenge for the party. Because the DMG recommends pretty high hp for monsters and at that time I assumed the guidelines were correct, I made sure the Wizard had cast False Life before the fight. One thing I'm curious about here is whether that was actually the right thing to do. Blogofholding points out, I think correctly, that the hp of MM monsters tends to be lower than the recommended hp in the DMG. While the formulas might still be fine once you take all the adjustments into account, I'd assume that the more adjustments that apply the "swingier" the fight gets: if you have very low hp made up for by high AC then the monster could die very quickly or resist many turns depending on luck, if you have high offensive CR and low defensive CR then whether or not the monster TPKs can come down to the initiative roll. Luck always plays a role in D&D of course, but if the game gets too swingy it's much less fun, so it would be good to know a safe "baseline" to perturb around.

And before someone says "estimate based on how strong your party is" - that was one of the first times I ran a 5e game, I don't have the experience with the system to do that yet.

[Eldariel]

I normally build like a PC, and then ignore all of the CR math and give out XP as if it had CR = level.

(Or if it's a monster like a dragon that already has a CR, I eyeball it. Sometimes I take that amount of XP and add it on top of the existing XP.)

Key insight here: players don't mind when you're generous. At the same time, I genuinely feel that CR = lvl is a pretty fair ballpark estimate in the sense that between a level N (N)PC and a typical MM CR N monster, it's hard to predict who would win in a fight, because (N)PCs have lots of strong abilities that don't show up on the CR table, like Uncanny Dodge and Battlemaster maneuvers and Dimension Door.

So if I give out 10,000 XP for facing a 13th level enemy NPC, at worst I'm being generous and at best I'm being fair.

The Blog of Holding math cannot be trusted BTW, it's built on faulty assumptions.

I pretty much second all of this. I generally prefer using PC statblocks since they're more interesting and natural, and can give the PCs an idea of what they didn't pick. Plus they feel fair since everything the NPC can do, the PCs can do just as well. Plus it lets me "play" various builds that I've got lying around and I would never get to try in an actual game since the amount of games I get to play vs. what I get to DM is very lopsided.


That info about setting conditions equivalent to damage based on spell level is fascinating, and explains a lot. Let's get some spell level equivalents going.

Deafened: 2 for Blindness/Deafness
Blinded: 1 for color spray (resisted with hit points, duration 1 turn) or 2 for blindness/deafness (resisted with save, duration 1 minute no concentration)

Charmed: 1 for humanoids and beasts (charm person and animal friendship), 3 otherwise (hypnotic pattern)
Poisoned: 1 (Ray of Sickness)
Frightened: 1 (cause fear, wrathful smite)

Incapacitated: 1 (Tasha's Hideous Laughter)
Stunned: 5, somehow (technically Contagion can cause it, but there are extra steps; at level 7, Divine Word and Symbol can both do it, but Divine Word is resisted with both save and then hit points, and Symbol takes a minute to cast; the lowest level spell that lets you simply cast a spell as an action and the target(s) save or stun is level 9. Psychic Scream), despite being not as bad as Paralyzed
Paralyzed: 2 (Hold Person - following Charm's precedent, humanoid restrictions are fine. 5 for Hold Monster)
Petrified: 6 (Flesh to Stone)
Unconscious: 1 (Sleep, hit point resist) or 6 (Eyebite)

Grappled: No spell.
Prone: 0 (Sapping Sting)
Restrained: 1 (Entangle)

Exhaustion: 4 (sickening radiance, wears off)

This list has some serious absurdity to it - Blinded is a lower level than Deafened, and Unconscious is lower than Paralyzed which is lower than Stunned which is lower than Petrified, even though the approximate order of how bad they are for you is Stunned < Petrified < Paralyzed < Unconscious. You can help the list by removing hit point resists, which raises blinded to 2 and unconscious to 6, but the incapacitated...unconscious chain is still messed up.

Grappled is actually a condition caused by among others Bigby's.

But yeah, the method seems extremely rough: e.g. Hold Person as an example of Paralysis and then compared to Scorching Ray completely ignores that Hold Person:
- Takes Concentration
- Is limited to one creature type
- Offers a save each turn

Which make it a very situational Paralysis effect. Hold Monster, a level 5 spell, is a better point of comparison but even there comparing it to equal level damage spell ignores that Hold has Concentration and Repeat Saves making it "lower level" than it would otherwise be. And of course, for such a comparison to be valid in the first place, the spells of a given level would have to be balanced against one another.

[Urpriest]

I pretty much second all of this. I generally prefer using PC statblocks since they're more interesting and natural, and can give the PCs an idea of what they didn't pick. Plus they feel fair since everything the NPC can do, the PCs can do just as well. Plus it lets me "play" various builds that I've got lying around and I would never get to try in an actual game since the amount of games I get to play vs. what I get to DM is very lopsided.

For those suggesting this (NPC CR=roughly PC level): do you just never use published humanoid monsters? If you use them too, don't your PCs get confused that some NPCs enemies have access to much higher-level spells than others? (For that matter, this would apply for spellcasting monsters too.)

Also, do you have to do something extra to make the NPCs durable enough? I feel like PC hitpoints are much lower than those of equal-CR monsters.

[Eldariel]

For those suggesting this (NPC CR=roughly PC level): do you just never use published humanoid monsters? If you use them too, don't your PCs get confused that some NPCs enemies have access to much higher-level spells than others? (For that matter, this would apply for spellcasting monsters too.)

Also, do you have to do something extra to make the NPCs durable enough? I feel like PC hitpoints are much lower than those of equal-CR monsters.

Well, if I'm running a module and it has some premade humanoids, I might use them if I don't have anything appropriate lying around and am not interested in drafting a PC version up (though even there, I generally throw a special ability or two from the class the statblock is emulating onto it; I rarely run the same statblock twice). I'm totally okay with just throwing higher level spells and effects at the PCs too: e.g. in RoT that's simply a bunch of level 10+ Wizards in the final fight. My Glasstaff was a level 8 Conjurer, Hamun Kost a level 10 Necromancer, Nezznar a level 11 Transmuter. I've found NPCs are a great place to use the various suboptimal (for PC use) spells: Cloudkills, Stinking Clouds, Confusions, etc.

As for durability, well, the same as with PCs, NPCs tend to have abilities that make them harder to take down than would appear on the surface. For those Wizards for instance, I generally have them make copious use of the various defensive spells (False Life, Mirror Image, Blink, Shield, etc.) and of course position carefully and defensively (and where appropriate, always be at a ready to teleport away or to reposition). As the classes get more defensive numbers, the active defenses of course become more sparse but generally, just as with PCs, I've found that NPCs can get durability with abilities and positioning. And it's totally okay for characters to go down quickly if they position poorly or get surprised or some such.

[Urpriest]

Well, if I'm running a module and it has some premade humanoids, I might use them if I don't have anything appropriate lying around and am not interested in drafting a PC version up (though even there, I generally throw a special ability or two from the class the statblock is emulating onto it; I rarely run the same statblock twice). I'm totally okay with just throwing higher level spells and effects at the PCs too: e.g. in RoT that's simply a bunch of level 10+ Wizards in the final fight. My Glasstaff was a level 8 Conjurer, Hamun Kost a level 10 Necromancer, Nezznar a level 11 Transmuter. I've found NPCs are a great place to use the various suboptimal (for PC use) spells: Cloudkills, Stinking Clouds, Confusions, etc.

As for durability, well, the same as with PCs, NPCs tend to have abilities that make them harder to take down than would appear on the surface. For those Wizards for instance, I generally have them make copious use of the various defensive spells (False Life, Mirror Image, Blink, Shield, etc.) and of course position carefully and defensively (and where appropriate, always be at a ready to teleport away or to reposition). As the classes get more defensive numbers, the active defenses of course become more sparse but generally, just as with PCs, I've found that NPCs can get durability with abilities and positioning. And it's totally okay for characters to go down quickly if they position poorly or get surprised or some such.

Ah ok, so to clarify, when you use NPC CR=PC class level, it sounds like you're just using that CR for determining XP, not for determining whether the fight is an appropriate challenge? (Come to think of it it sounds like that's what MaxWilson was saying he does too.)

In that case, how do you figure out what's an appropriate challenge? Or do you just have enough experience at this point that you can do it by eye?

[Eldariel]

Ah ok, so to clarify, when you use NPC CR=PC class level, it sounds like you're just using that CR for determining XP, not for determining whether the fight is an appropriate challenge? (Come to think of it it sounds like that's what MaxWilson was saying he does too.)

In that case, how do you figure out what's an appropriate challenge? Or do you just have enough experience at this point that you can do it by eye?

Well, this is just me, not Max, but eyeball if relevant but I largely don't care about what's an appropriate challenge: I throw creatures that feel like they make sense into the world and let the PCs determine what's appropriate for them to face and what isn't. I do sometimes eye it approximately but it is my experience that the PCs can usually handle themselves. And if they **** up their scouting or threat assessment or whatever and don't get lucky enough to win anyways, they die or get captured or whatever makes sense for the opponent.

Like when I write or adapt a campaign skeleton, I determine NPC levels based on what I know about their story and history, and what kind of level would make sense for them; I recall I had Sildar Hallwinter as a level 7 Fighter and Reidoth the Druid at level 8 too for example (pardon all these LMoP examples, it's my gut feeling that it's the path most people are familiar with). In general, people with a long history in important positions locally speaking I put around Tier 2 and regionally important characters or characters with exceptionally long history around Tier 3 and globally/interplanarry important ones around Tier 4 (that's of course just a vague guideline: there are people who lack power but are important and people who have power but aren't important).

I generally start PCs off on level 3 so most NPCs are beatable straight-up (though in a fair fight and with equal numbers, most of them can wipe the floor with the party).

[MaxWilson]

For those suggesting this (NPC CR=roughly PC level): do you just never use published humanoid monsters? If you use them too, don't your PCs get confused that some NPCs enemies have access to much higher-level spells than others? (For that matter, this would apply for spellcasting monsters too.)

Also, do you have to do something extra to make the NPCs durable enough? I feel like PC hitpoints are much lower than those of equal-CR monsters.

I certainly do use published humanoid monsters: Githyanki Warriors, various Orcs, etc. I sometimes even use some MM NPC stat blocks like Bandits. I never use stuff like MM Archmages though--archmages are rare and significant enough that they get the full treatment.

I don't understand the question about confusing players. No one has ever expressed such confusion to me, and I don't know why it would be confusing that not all people or monsters are the same. The ones with powerful magic generally have a reputation which precedes them, either individually or as a class, and I've never had anyone go, "Huh? But [some other monster] only has weak magic!" I don't understand the problem you're asking about.

(N)PCs with class levels are somewhat glass cannon-ey compared to MM monsters but I view that as a feature, not a bug. 5E monsters have wildly inflated HP--in AD&D it takes one Fireball to kill a troll, in 5E it takes three, not counting saves. 5E PCs only have moderately inflated HP--two Fireballs would down most Tier 2 (N)PCs. Having fewer HP doesn't matter that much because they also have higher AC than most monsters, or Counterspell, or Skulker + Pass Without Trace + Cunning Action, or Mobile feat, or whatever other combo they rely on for defense, but it does have the nice effect that when you manage to actually hit them they do tend to go down faster, instead of laughing off the hits with absurdly inflated HP.

But to answer your question, I do roleplay such NPCs as intelligent, and so they generally do intelligent things like use partial/total cover, use spells like Shield and Dimension Door when outnumbered, recruit monsters as bodyguards, avoid knowingly attacking a superior force head-on, etc. They want to live as badly as the PCs do. If I just wanted combat speedbumps I would not be using NPCs in the first place, I'd be using Stirges or Manticores or something.


Ah ok, so to clarify, when you use NPC CR=PC class level, it sounds like you're just using that CR for determining XP, not for determining whether the fight is an appropriate challenge? (Come to think of it it sounds like that's what MaxWilson was saying he does too.)

In that case, how do you figure out what's an appropriate challenge? Or do you just have enough experience at this point that you can do it by eye?

That part is actually easier when the enemy has levels instead of CR. By definition, M Nth level NPCs are an approximately fair fight for M Nth level PCs: you can't really predict who's going to win without knowing more details. (They could be mirror images of each other!) But Lanchester's Square Law says combat power scales roughly as the square of the quantities involved, so 3 NPCs have only about half the combat power of 4 same-level PCs (3^2/4^2 = 9/16, just over 1/2). So you can have what superficially looks like a strong enemy force of three near-peer NPCs, but the math still heavily favors the PCs to win the battle, which might be what you want.

You do still have to decide what you mean by "appropriate challenge" (how heavily do you want to favor the PCs?), but once you know what you're shooting for it's easier to do than it would be with monsters, because you're comparing apples to apples instead of apples (levels) to oranges (CR).

You might want to change the NPCs' level relative to the PCs. PC power is sort of linear in level, but power tends to jump a bit at tier boundaries, and furthermore if you've optimized the NPCs (e.g. a Rogue 2/Bladesinger 7 with Greater Invisibility and Stealth Expertise to go with Cunning Action (Hide)) you'll also know in what ways they punch above their weight class. When in doubt, run yourself some practice combats with made-up PCs to get a feel for tactical options, counterplay, counter-counterplay, and outcomes. (But never underestimate the players' ability to surprise you with brilliant creative solutions or bad tactics that get them killed! That's on them mostly but it does mean you should always be prepared for them to TPK no matter how easy you THOUGHT the fight was. Have a plan for what you'll do if they do TPK.)

So yeah, pretty much I eyeball things. Just like I do with monsters actually, using that same "CR ~= level" heuristic, while also being conscious of monsters that clearly punch above their CR, like Banshees and Star Spawn Manglers and anything that can spam Hold Person, like Neogi Masters.

[da newt]

Using bad guys built like PCs often results in very swingy combat encounters. Most PC builds are glass cannons compared to monsters - PCs tend to have high offense capability (especially casters) and low HP compared to monsters of the same CR (PC AC and other defense can be all over the place - some lo some hi). Most monsters are built so that they are a little low on offense and have a bunch of hp so that combat is more predictable.

When you replace a mediocre offense / Hi hp monster with a Hi offense / Lo hp PC, combat is often decided at initiative / during the first round or two, and depending on who get the upper hand early, could be a cake walk for the PC foe or they could get curb stomped quickly.

When building PC foes I usually try to nerf their offense a bit (by build or DM tactics) and boost their defense so that they are less of a glass cannon and more of a grinder. As for CR vs lvl {for balance}, I try to find the closest MM / Volo / other book published NPC to use as a guide and modify from there when I can. I think that lvl = CR results in too high a CR, and lvl * 2/3 = CR or lvl * 3/4 = CR seems to be more accurate, but it also depends on if you are building optimized PCs and giving them lvl appropriate magic items etc ...

But then you get into what is the real difficulty of X CR vs the party issues too (many argue book defined 'deadly' encounters are nowhere near deadly).

In other words, I agree w/ Max. (although it might be handy to note Max prefers combat where the party has an equal chance of victory and TPK - many DMs prefer the majority of combats are less deadly)

[MaxWilson]

In other words, I agree w/ Max. (although it might be handy to note Max prefers combat where the party has an equal chance of victory and TPK - many DMs prefer the majority of combats are less deadly)

Yeah, it's worth noting that while my games tend to be combat-light, it's partly because I don't find attrition all that interesting and I would prefer to (but don't necessarily) skip over combats where nothing is at stake except attrition. I started my most recent adventure (https://forums.giantitp.com/showsinglepost.php?p=25042859&postcount=23) in medias res by basically saying, "You're in the middle of a dungeon crawl. Everybody starts at half HP and half special abilities, but up 3000 XP." I didn't want to play through all the little fights that would have gotten the PCs those 3000 XP.

YMMV and your style may be different from mine.

[Urpriest]

I certainly do use published humanoid monsters: Githyanki Warriors, various Orcs, etc. I sometimes even use some MM NPC stat blocks like Bandits. I never use stuff like MM Archmages though--archmages are rare and significant enough that they get the full treatment.

I don't understand the question about confusing players. No one has ever expressed such confusion to me, and I don't know why it would be confusing that not all people or monsters are the same. The ones with powerful magic generally have a reputation which precedes them, either individually or as a class, and I've never had anyone go, "Huh? But [some other monster] only has weak magic!" I don't understand the problem you're asking about.

I think your later description of how you structure your campaigns clarified this. What I have in mind is not plot-relevant NPCs that the players would have heard of, but just class-leveled enemies they fight on the way. The druid who leads a squad of lizardfolk, the quirky martial arts cultists, the hobgoblin mage tossing fireballs behind the back lines. If you're the kind of DM who lightly railroads the party through the proscribed mix of encounters of various CRs each day, then the players can pick up on broad patterns like which spell levels these kinds of guys have access to. Using published monsters those levels would typically be higher than what the same DM would create with a CR=level guideline.

The thing is of course, you're not that kind of DM. You avoid using those kinds of encounters, so there's no pattern for players to notice. It's just a different style.

[MaxWilson]

I think your later description of how you structure your campaigns clarified this. What I have in mind is not plot-relevant NPCs that the players would have heard of, but just class-leveled enemies they fight on the way. The druid who leads a squad of lizardfolk, the quirky martial arts cultists, the hobgoblin mage tossing fireballs behind the back lines. If you're the kind of DM who lightly railroads the party through the proscribed mix of encounters of various CRs each day, then the players can pick up on broad patterns like which spell levels these kinds of guys have access to. Using published monsters those levels would typically be higher than what the same DM would create with a CR=level guideline.

The thing is of course, you're not that kind of DM. You avoid using those kinds of encounters, so there's no pattern for players to notice. It's just a different style.

Yeah, and even if I did do that, the only thing they'd pick up on would be that the 5th level druid yields a lot more XP (1800) than the equivalent lizardfolk shaman (400 XP), not that the lizardfolk shaman has better spells. BTW I think this indicates that the lizardfolk shaman should have its CR raised--anyone who can spam you with 8 Giant Poisonous Snakes and then another 8 Giant Poisonous Snakes is worth more than a measly 400 XP.

[verbatim]

When in a pinch and I don't have level appropriate official/homebrew statblocks for something I'll take something that fits what I want thematically and change their proficiency bonus and # of hit die to scale to the party's level, making subjective base stat/ac changes along the way as neccesary.

[EggKookoo]

My totally scientific and mathematically bulletproof method of designing encounters to party level goes like this.

I take the APL. Then I turn to the charts starting on page 90 of Xanathar's Guide to Everything. I look for the Challenge Rating that's 1:1 for the APL. I form an encounter with a set of creatures with that CR, and the number of creatures matches the party count. That's about the most basic it gets.

If I want to mix things up, I increase the number of creatures to make the fight harder, or decrease them to make the fight easier. Regardless, I use creatures whose average CR still works out to be that number.

For example, I have a party of 4 lv6 PCs. The CR at the 1:1 column is 2, so I can hit the party with four CR 2s and expect a fun, if not super-challenging fight. If I want to make the fight harder, I could go with three CR 2s and two CR 1s, or two CR 2s and four CR 1s. The 2/2/1/1/1/1 encounter will actually be a little tougher than the 2/2/2/1/1, because action economy reigns supreme in 5e.

I also consider combat pets to be full party members when doing these calculations. If I have an artificer battle smith with a steel defender, and a chainlock with investment and voice of the chain master and, say, a pseudo-dragon (in addition to two other party members), I calculate the encounter as though there were 6 party members. So that's six CR 2 enemies as a baseline, and I adjust from there.

In practice, I don't do this that often. But I do use it as a rough guide for determining how hard a given encounter will be. I usually just eyeball encounters based on gut feeling and how things would work in the fiction. But I like to have a sense of how hard the fight will be so I can provide fair-play clues to the players. I kind of reverse-engineer it out of the above method (and even then, it's mostly a high-level pass rather than strict number-crunching).

[Bobthewizard]

I fly more by the seat of my pants than most of this suggests. So I don't know if this will be helpful but this is how I do it.

When I come up with a potential combat encounter, I base the NPCs on simplified PCs. I try to make the total enemy HP and total enemy damage per round about equal to the party's. Then the AC, and bonus to hit average about the same as the party. Saves are usually worse than the party because save or suck spells are fun for players to cast but terribly boring for a player to get hit by. I'll usually give the minions one interesting ability - cunning action, trip or disarm maneuvers, sneak attack, smite, misty step.

The boss is usually a spellcaster, but I don't stat out all their spells. I try to come up with, "What are they going to do for 3-4 rounds before they get taken out?" and "Does this spellcaster have shield and absorb elements?" Solo bosses get legendary saves and actions even as low as level 3-4.

I can come up with the stats for a 10-15 NPC combat in about 5 minutes. Then I can spend my time on "What is the goal of this combat?" "How do I engage the players?" My combats tend to last 10-15 rounds with multiple rooms, waves of enemies, innocent bystanders, and conflicting objectives.

[PhoenixPhyre]

To be honest, I do a mix of things. First, I don't use CR for anything other than to set the proficiency bonus. I don't calculate encounter difficulty[1] and I don't award XP[2], so yeah.

Most of my homebrew creatures (humanoid or otherwise) follow the pattern:
1. Is there a published one that's close enough, maybe with renaming a weapon (but not bothering to recalculate anything)? If so, use it.
2. Is there a published one that's almost close enough, but needs a new feature or changes to spell lists[3]? Use it, make those changes. Don't bother recalculating CR.
3. Is there one I can use as a framework, even though I'm going to have to throw out all the features? Do it. Maybe think about adjusting the CR if it really doesn't match and would throw off the proficiency-related things.
4. If absolutely necessary, I'll build one from scratch. Even then, I don't really build PC-type NPCs--I'll estimate a rough level, but stats and proficiencies get set by the character's needs. I'll generally throw in 2-3 "key" features if they're strongly associated with a class (rage, sneak attack, etc). If I'm doing this, 90% of the time those associated classes are homebrew and I'm doing a soft play-test[4].

Generally though, I don't do much for the stats. I know what the attack bonus should be for monsters of the "strength category" (like CR, but way more nebulous) I'm shooting for and so I just use that (not worrying about getting the internal math right).

[1] Encounters are eyeballed based on what should appear there based on the narrative. I'm no tactician and my dice like my players, so I don't really worry about being too hard. They took a CR 22 ithillich at level 11 without real difficulty, mostly by managing to start combat at knife-fighting range via trickery and the paladin going first and smiting the ever-living crap out of it on turn 1 (roughly 70% of its HP before anyone else acted). And that one had a trick up its sleeve to give it a full HP reset. None of my parties are anywhere near the baseline, so the encounter difficulty/XP/adventuring day tables don't do me much good. Which is fully expected by the devs--they've said that those tables are a crutch for new DMs to get a ballpark estimate, not some system expectation. They're reverse-engineered from play tests, not some goal the devs shot for.
[2] I use a "session-tick-mark" leveling--every X sessions you all go up a level. It goes 1, 2, 3, 3, ... 3; only levels 1 and 2 take less than 3 sessions. I'll probably back off of this a bit for future campaigns, maybe going to 1, 2, 3, 4, 5, 5, 5, 5, ..., 5.
[3] And I don't restrict myself to the class-based spell lists. Instead I use my thematic lists and some gut feelings about how that character should be acting.
[4] I'm a forever DM, usually playing with new players, so I've got a backlog of classes I want to test. So I'll pull specific mechanics that I want to test on to NPCs. Not a great play-test because NPCs are not PCs, but it gives some fit-and-finish feedback.