CR use

[zinycor]

For the GMs here, do you use the CR rules in order to balance your encounters? How important is CR for you?

Personally I only see CR as to give me a very rough idea on how difficult the encounter will be . But very often "easy" encounters end up being memorable battles and "deadly" encounters end up being defeated in a couple of turns.

So, how much do you use CR at your table?

[TrueAlphaGamer]

These days, I mostly make my own monsters/enemies, so I rarely rely on doing all the CR calculations when balancing encounters, especially since CR is notoriously esoteric and fickle when it comes to math and level-appropriate challenge (also since my groups are 3 players instead of 4).

[MaxWilson]

For the GMs here, do you use the CR rules in order to balance your encounters? How important is CR for you?

Personally I only see CR as to give me a very rough idea on how difficult the encounter will be . But very often "easy" encounters end up being memorable battles and "deadly" encounters end up being defeated in a couple of turns.

So, how much do you use CR at your table?

I use them for calculating XP requirements, and for rough guesstimates of raw combat power. That is, I don't consciously use it, but at the same time I'm sure that I'm aware on some level that Glabrezus are CR 9, and when I eyeball an encounter for an 8th level party, I'm more likely to make it a Glabrezu and a Young Red Dragon than a Babau and a Fire Elemental because I am aware of their CRs.

Without CR, the alternative would be memorizing HP, AC, and Attack/Damage for each monster, and there would be more mental math before I could tell if a Babau + Fire Elemental was too weak.

I have intentions at some point to drill down on CR and figure out what it really measures, but as of today I treat it roughly as a summary of HP x DPR. I factor in special abilities separately, in my head.

[Kurt Kurageous]

For the GMs here, do you use the CR rules in order to balance your encounters? How important is CR for you?

Personally I only see CR as to give me a very rough idea on how difficult the encounter will be . But very often "easy" encounters end up being memorable battles and "deadly" encounters end up being defeated in a couple of turns.

So, how much do you use CR at your table?

I am with you. CR is a rule of thumb, not a precise tool at all no matter how much space the DMG devotes to it.

[Unoriginal]

CR is useful as a shorthand for damage capacity vs survival capacity in a straight, nuanceless fight. Basically it's like the weight-in before a boxing match: it just let you see if the people are boxing in the same weight class, and it informs on some of the factors important in a fight, but it doesn't tell you how the match will go.

[PhoenixPhyre]

CR is designed to be (and really only useful for) a first-pass filter to narrow down the stat blocks available for an encounter and to choose the number of stat blocks. There's even developer commentary that they expect DMs to depart from the guidelines based on the needs of their groups, and that they intentionally developed the encounter-balance guidelines (which use XP as an input, which comes from CR) for new DMs (new to a group or new to DMing period). Experienced people are expected to eyeball most of it based on their experience with that group.



I have intentions at some point to drill down on CR and figure out what it really measures, but as of today I treat it roughly as a summary of HP x DPR. I factor in special abilities separately, in my head.

I'll save you the effort, as I already did that.

CR is a rough measure of two things (for a balanced offense/defense monster), under some assumptions[1] that are not universally met.
1. A balanced CR X monster will survive, on average, about 3 rounds under direct fire from a party of level X characters. The components here are HP, modified by AC and things that provide defensive bonuses. For example, regeneration provides effective HP, as do immunities and resistances to common damage types[2]. AC reduces incoming damage (note: this assumes that most of the relevant incoming damage is attack-based, as saving throws are not calculated into this statistic).
2. A balanced CR X monster is unlikely to, on average damage rolls, drop any character of level X from 100% to 0% in a single round. The components here are average adjusted DPR[3] and attack bonuses, although save DCs can apply if the creature's damage output is dominantly save-based (which is rare).

Spells and other non-direct-damage are (irregularly) factored in as if they were damaging spells of the same level--this is rarely done because it is rarely relevant to the top 3 rounds.

So CR is basically a safety threshold--CR X means that it has a decent chance of getting off its "cool abilities" in a fight while not being able to one-shot a squishy (except on multiple crits). So it's generally "safe" to use CR < level creatures, while higher ones require a more careful look or other considerations. In reality, most published monsters are offensive-leaning--they have higher offensive CR than defensive CR. There are exceptions to that trend.

[1] The big assumptions:
* All party members are the same level
* Minimal optimization[4] and no variant features
* No combat-effective magic items
* for survival, it seems they assume a "balanced" 4-man-band party
* Simple "you and him fight until one dies" tactics on both sides

[2] Because of the assumptions made in [1], piercing resistance to non-magic BPS damage is expected to happen due to other effects (e.g. smite) or spells. By high levels, the effects of these drop off. Is this a perfect assumptions? No. But it's a core assumption of CR.

[3] adjusted by traits and features that increase damage output, such as the "share damage" features of creatures like Cloakers. Things like a banshee's wail...aren't taken into consideration at all from what I can see.

[MaxWilson]

I'll save you the effort, as I already did that.

CR is a rough measure of two things (for a balanced offense/defense monster), under some assumptions[1] that are not universally met.
1. A balanced CR X monster will survive, on average, about 3 rounds under direct fire from a party of level X characters. The components here are HP, modified by AC and things that provide defensive bonuses. For example, regeneration provides effective HP, as do immunities and resistances to common damage types[2]. AC reduces incoming damage (note: this assumes that most of the relevant incoming damage is attack-based, as saving throws are not calculated into this statistic).
2. A balanced CR X monster is unlikely to, on average damage rolls, drop any character of level X from 100% to 0% in a single round. The components here are average adjusted DPR[3] and attack bonuses, although save DCs can apply if the creature's damage output is dominantly save-based (which is rare).

Spells and other non-direct-damage are (irregularly) factored in as if they were damaging spells of the same level--this is rarely done because it is rarely relevant to the top 3 rounds.

So CR is basically a safety threshold--CR X means that it has a decent chance of getting off its "cool abilities" in a fight while not being able to one-shot a squishy (except on multiple crits). So it's generally "safe" to use CR < level creatures, while higher ones require a more careful look or other considerations. In reality, most published monsters are offensive-leaning--they have higher offensive CR than defensive CR. There are exceptions to that trend.

[1] The big assumptions:
* All party members are the same level
* Minimal optimization[4] and no variant features
* No combat-effective magic items
* for survival, it seems they assume a "balanced" 4-man-band party
* Simple "you and him fight until one dies" tactics on both sides

[2] Because of the assumptions made in [1], piercing resistance to non-magic BPS damage is expected to happen due to other effects (e.g. smite) or spells. By high levels, the effects of these drop off. Is this a perfect assumptions? No. But it's a core assumption of CR.

[3] adjusted by traits and features that increase damage output, such as the "share damage" features of creatures like Cloakers. Things like a banshee's wail...aren't taken into consideration at all from what I can see.

It's got to be slightly more than just those two things though. In particular, #2 (unlikely to drop anyone from full HP to 0) is a ceiling on effectiveness, not a floor. Furthermore, there are no PCs over level 20, and yet there are CRs over 20. I'm interested in (some day) deciphering what they mean, and how CR scales. I know CR scaling is clearly nonlinear, but without looking at the DMG table I couldn't really tell you how much tougher a by-the-book CR 6 monster is than a CR 8 monster.

I can immediately tell you that 8 monsters are about 77% more deadly than 6 monsters, because they have 33% more staying power * 33% more DPR = 77% more combat power overall. But for CR I don't have that kind of intuitive grasp yet of what it means, if it even means anything.

My gut tells me that CR vastly overestimates the importance of to-hit bonuses, and underestimates the importance of a monster's first 40 HP. It's possible that I really should be ignoring CR entirely when I do my rough mental estimates of difficulty and looking purely at HP and DPR.

[da newt]

"I can immediately tell you that 8 monsters are about 77% more deadly than 6 monsters, because they have 33% more staying power * 33% more DPR = 77% more combat power overall."

Could you show how 2 more monsters = 77% more "deadly" overall please?

I'm not sure I understand how you are looking at things. Do you mean 77% more likely to kill one PC, 77% more likely to TPK, or something else?

Certainly the two extra monsters add more attacks/damage per round AND more things that need to be killed (hp pool), but I have trouble believing a party of 8 PCs is 77% more deadly that a party of 6 - does that mean that a party of 8 is 400% more deadly than a party of 4?

[MaxWilson]

"I can immediately tell you that 8 monsters are about 77% more deadly than 6 monsters, because they have 33% more staying power * 33% more DPR = 77% more combat power overall."

Could you show how 2 more monsters = 77% more "deadly" overall please?

I'm not sure I understand how you are looking at things. Do you mean 77% more likely to kill one PC, 77% more likely to TPK, or something else?

Certainly the two extra monsters add more attacks/damage per round AND more things that need to be killed (hp pool), but I have trouble believing a party of 8 PCs is 77% more deadly that a party of 6 - does that mean that a party of 8 is 400% more deadly than a party of 4?

I mean that, for example, they're likely to inflict 77% more damage on the PCs before they die. This is due to Lanchester's Square Law: https://military.wikia.org/wiki/Lanchester%27s_laws, which AFAICT is also where DMG difficulty multipliers come from for "adjusted XP", although see below.

Simple illustration: let's say the PCs have enough firepower to kill 1 monster per round, and each monster can deal 10 HP per round damage to some PC. If there are six monsters, the monsters will all be dead after 6 rounds, and they'll inflict 60 + 50 + 40 + 30 + 20 + 10 = 210 damage in the process. If there are eight monsters, they'll all be dead after eight rounds, and they'll inflict 80 + 70 + 60 + 50 + 40 + 30 + 20 + 10 = 360 damage in the process. 360/210 = 1.7ish.

In real life, there are important factors not considered by Lanchester's Laws, such as how closely the monsters and PCs are clumped up, how vulnerable they are to AoEs like Hypnotic Pattern and Fireball, whether the monsters can switch targets easily if a PC starts Dodging, etc. (There are also less-important factors like "who wins initiative?" that still matter somewhat in small fights.)

Because Lanchester's Square Law is a rule of thumb that doesn't apply perfectly, especially when AoEs come into play, military analysis sometimes splits the difference between Lanchester's Linear Law and Lanchester's Square Law and computes the 3/2 power instead of the square. 8^1.5 is about 22, whereas 6^1.5 is about 14. That's approximately why 6 orcs costs 1200 adjusted XP but 8 orcs costs 2000 adjusted XP. Of course, WotC simplified the multipliers to round numbers to make them easier to compute, but you can see that both methods conclude that 8 orcs is almost twice as dangerous as 6 orcs in terms of how many HP they're going to knock off you before you die, because those two extra orcs get in a lot of attacks while you're busy dealing with the first six orcs.

And this is why it's so important to have access to spells like Fireball and Hypnotic Pattern--because the game assumes that you'll have some access to artillery-ish spells. If you don't have artillery, then instead of 3/2 power the difficulty scales as the square, which means without crowd control AoEs mobs are even deadlier to you than the game predicts they should be. Although, Fireball won't help you if the orcs are spread out and chucking javelins, or if they are hobgoblins using longbows from a dispersed formation, so the game always underpredicts the difficulty of dispersed mobs with ranged weapons.
Any further questions?

[BloodSnake'sCha]

I use CR to know how much XP I need to give in the end of the session.

I see no other real use for it. It is a bad way to know how hard or easy an encounter is.

[PhoenixPhyre]

It's got to be slightly more than just those two things though. In particular, #2 (unlikely to drop anyone from full HP to 0) is a ceiling on effectiveness, not a floor. Furthermore, there are no PCs over level 20, and yet there are CRs over 20. I'm interested in (some day) deciphering what they mean, and how CR scales. I know CR scaling is clearly nonlinear, but without looking at the DMG table I couldn't really tell you how much tougher a by-the-book CR 6 monster is than a CR 8 monster.


It's actually quite linear until close to CR 20. Beyond that, it's also linear, just at twice the rate. Specifically, the equation for monster effective HP (up to CR 19, which is above the curve due to dragons and big-bad boss fiends) is eHP = 14.472*CR + 15, with an R^2 of 0.96. DPR follows the equation (even closer up to CR 13, then with a big upward swing at CR 14-15, but not actually too big) DPR = 6.093*CR + 3.0505 with an R^2 of 0.97.

So every CR increases HP by 14.5 and DPR by about 6. Which almost exactly matches the table values of 15 HP and 6 DPR.

If you want to look at the actual as-printed monsters, the spreadsheet linked in my signature has all the data for the big-3 monster books.

And #2 is actually as follows:
An offensive-CR X monster does enough damage at the low end of its DPR range to drop a "squishy" (CON +1 d6 HD) creature of level X-1 to zero from full in one round. So an (offensive) CR 2 monster does enough damage to take down a level 1 wizard in one round, assuming it hits and does average damage. It's actually calibrated carefully so it can't outright kill that character, even if everything crits (but deals average damage). A CR 1 monster can't one-round KO a full-health squishy, but will come darn close and can ORKO a wounded one.

So it really is both a floor and a ceiling (at different points in the DPR range).



I can immediately tell you that 8 monsters are about 77% more deadly than 6 monsters, because they have 33% more staying power * 33% more DPR = 77% more combat power overall. But for CR I don't have that kind of intuitive grasp yet of what it means, if it even means anything.

My gut tells me that CR vastly overestimates the importance of to-hit bonuses, and underestimates the importance of a monster's first 40 HP. It's possible that I really should be ignoring CR entirely when I do my rough mental estimates of difficulty and looking purely at HP and DPR.

The to-hit bonuses (by the book) are actually pretty close to constant hit-chance, making assumptions about the armor of the characters. Until you get to end-of-campaign boss monsters (ie CR 15+), the hit chances go like
* Tanky: ~30-50%
* Light + high DEX OR Medium : ~50-65%
* Squishy: ~75-85%

There's some variation across level bands, but there's a constant band of "this CR can hit the characters most of the time".

And I think as for the first few HP, there's a break point between "is supposed to die to an average fireball" and "should survive a fireball". Specifically, CR 1 and above monsters won't die to a fireball (on average), while those below will. Note that common "fodder" monsters (orcs, goblins, etc) are all CR < 1. That seems on purpose.

[MrStabby]

I never use CR for balancing encounters. For balancing encounters I usually look at attacks and spells and how closely they line up with party weaknesses, broad brush damage and how closely immunities and resistances line up with the party's preferred damage types.

CR is something I used for inspiration. Let's see what kind of abilities come on at what levels, see what kind of things these monsters can do. Use this to help guide home brewing of monsters.

[EggKookoo]

I use CR as a starting point, but I've found on a 1-for-1 matchup, even-level CR fights are typically pretty easy. I think they're meant to be.

If I want to make a fight harder, I'll usually add more monsters even if the overall encounter CR doesn't change. A party of 5 PCs with an APL of 5 will have a harder time dealing with 6 monster with a total CR adding up to 5 than they will against 5 monsters adding up to CR 5. And 4 monsters adding up to CR 5 will actually be easier, despite the individual monsters being more powerful. There's a limit, of course, but if you stay within a few extra (or fewer) monsters, this kind of thing works.

Also, I try to imbalance the individual CRs if I can. Fighting 5 CR 1 monsters is not the same experience as fighting one CR 3 monster supported by four CR 1/2 monsters, even though in both cases it's a CR 5 encounter with 5 monsters (notably the latter is worth more XP).

[ff7hero]

I've semi-recently adopted Angry's tier based CR-replacement/homebrew guidelines. It's certainly not any worse than CR, but that's not high praise. It is a lot more work though.

[OldTrees1]

CR is a useful rule of thumb until and unless you find a way to not use a rule of thumb.

3E's CR system was simple enough for it to become intuitive for me with practice.
It kinda boiled down to:
ECL ~= CR unless ECL is higher for reasons explained in Savage Species
2[X] = 1[X-1] + 1 [X+1] = 1 [X+2]

5E's CR system is not as intuitive for me yet. Partially because the math is messier. The growth curve is not smooth (see tier jumps) and bounded accuracy made larger encounters messier math than 3E's exponential rule did. However 5E has a tendency to let PCs become OP resilient without entirely feeling overpowered. So the encounters do not need to be accurately or precisely balanced as long as you don't aim too difficult.

I still use the CRs on monsters as a reference in 5E but I don't really bother doing the entire CR math anymore. As long as I don't increase the difficulty too much, the party will not be in real danger.

[da newt]

Max - thanks for the explanation. Is it true that Lanchester assumes both parties have equal offense and defense capabilities (in our case chance to hit and damage per hit vs Ac and Hp)?

[OldTrees1]

Because Lanchester's Square Law is a rule of thumb that doesn't apply perfectly, especially when AoEs come into play, military analysis sometimes splits the difference between Lanchester's Linear Law and Lanchester's Square Law and computes the 3/2 power instead of the square. 8^1.5 is about 22, whereas 6^1.5 is about 14. That's approximately why 6 orcs costs 1200 adjusted XP but 8 orcs costs 2000 adjusted XP. Of course, WotC simplified the multipliers to round numbers to make them easier to compute, but you can see that both methods conclude that 8 orcs is almost twice as dangerous as 6 orcs in terms of how many HP they're going to knock off you before you die, because those two extra orcs get in a lot of attacks while you're busy dealing with the first six orcs.

Any further questions?

Nice. That makes 5E encounters much easier to design now.

Unfortunately there is not a quick way to do mixed CRs unless the ratio is constant. 4 Orcs + 1 Ogre vs 3 Orcs + 2 Ogres for an example of a shifting ratio. Or is there?

[MaxWilson]

Max - thanks for the explanation. Is it true that Lanchester assumes both parties have equal offense and defense capabilities (in our case chance to hit and damage per hit vs Ac and Hp)?

(Thinks) I don't think so. It just assumes that each side is homogenous. Lanchester's equations will behave the same way if one side is English longbowmen and the other side is Confederate riflemen. They'll just have different coefficients, but the scaling won't change. It doesn't require longbowman vs. longbowman, although if you want to compute who's going to win you need to convey them into a common frame of reference, e.g. "longbowmen fire twice as fast as riflemen but are just as easy to kill, therefore treat one longbowman as sqrt(2) riflemen." But if the riflemen can use railroads to fight 1/3 of the longbowmen at a time, Lanchester still predicts that battle will inflict only 1/9 as many casualties on the riflemen as if they fought the whole longbowman army.


Nice. That makes 5E encounters much easier to design now.

Unfortunately there is not a quick way to do mixed CRs unless the ratio is constant. 4 Orcs + 1 Ogre vs 3 Orcs + 2 Ogres for an example of a shifting ratio. Or is there?

That's why I want to analyze CR scaling someday, so I can see if there's a pattern. For now though, I don't know of a way to do this. (But see below.)

For designing mixed encounters, I'm not happy with the DMG simplifications (adjusted XP), so I prefer computing heterogenous mobs by adding up the 2/3 power of all the individual monster XP/100 and raising the total to the 3/2, then multiplying the total by 100. E.g. one beholder and six hobgoblins comes out as ((10000/100)^(2/3)+6*(100/100)^(2/3))^(3/2)*100, which is roughly 14,456 adjusted XP, 45% harder than a regular 10,000 XP beholder on its own.

This seems more accurate than the DMG approach which predicts that the fight is 2.65 times harder (26,500 adjusted XP) if you count the hobgoblins as tactically significant (which they are, but not THAT much), or basically exactly the same difficulty as a lone beholder if you don't. Neither is right. Those hobgoblins can break concentration, take advantage of restrained or paralyzed targets from the eye rays to deal up to ~120 HP per round against a paralyzed squishy, attack squishies hiding in antimagic zones (important for overcoming Darkness and Fog Cloud), grapple PCs and drag them out of antimagic zones, threaten opportunity attacks, and provide 66 extra armored HP for the beholder's side. Saying that they add up to about a third of an extra beholder seems reasonable.

I guess if you precompute the 2/3 power of orcs and ogres you could maybe use that to eyeball changes to ratios. For now I can't do that in my head and I don't do it on the fly, and even if I could I haven't proved that it's accurate in the same way Lanchester's Laws are, just that it's more accurate than the DMG method.