Understanding CR metrics

[ZeroSpace9000]

Hey all, longtime 3.PF DM here. I've recently concluded running a 3.5 module that I converted and ran for my 5e gaming group, and the biggest issue I had was calibrating encounters and monsters. I was always a little unsure about if I was throwing too much stuff at the players, or not enough. So I want some help just understanding the finer points of CR, and encounter build by numbers.

And to everyone to the punch, yes I know that CR is hilariously imprecise. I know there is some degree of eyeballing that's required. I just want a solid idea of how to structure combat encounters, both when several a day are expected, and when there will be just one fight a day.

[CheddarChampion]

It's less official but I recommend looking at the Angry GM's latest blog post.

[MaxWilson]

Hey all, longtime 3.PF DM here. I've recently concluded running a 3.5 module that I converted and ran for my 5e gaming group, and the biggest issue I had was calibrating encounters and monsters. I was always a little unsure about if I was throwing too much stuff at the players, or not enough. So I want some help just understanding the finer points of CR, and encounter build by numbers.

And to everyone to the punch, yes I know that CR is hilariously imprecise. I know there is some degree of eyeballing that's required. I just want a solid idea of how to structure combat encounters, both when several a day are expected, and when there will be just one fight a day.

Kobold Fight Club does all the math for you. I highly recommend it: http://kobold.club/

In a multi-encounter day, anything approximately within the XP budget is fine: three Deadlies with the chance for short rests between them, or one Hard and then six Easies in a row, whatever.

If planning one fight in a day, I recommend building roughly Deadly through triple-Deadly encounters to avoid boredom and a feeling of anticlimax. Choose your own comfort level but I find that triple-Deadly is an almost guaranteed win for the PCs but has a good chance to kill any individual PC who does something exceptionally dumb. (Caveat: I play monsters relatively dumb unless they are genius monsters like mind flayers, because six seconds isn't much time to mentally adjust to PC tactics and monsters, unlike PCs, don't think in bullet time. Genius monsters can be deadly at far less than Deadly difficulty.) At Deadly x10 PCs have to use exceptionally strong tactics (stuff that will break normal encounters, like Conjure Animals or Polymorph spells, or an all-Mobile feat party), run away, use limited-use magic items (Horn of Valhalla), or die.

[Lunali]

CR is all about how long something will take to kill and how much damage it will do in that amount of time. Be sure to read any abilities the monsters have and assess their possible effects before using them at the table.

If you're doing multiple encounters per day, each fight should expend some resources but not a lot of resources. By the end of the day characters/players should feel that it would be dangerous to continue. Be warned that interrupting rests makes players likely to end their day earlier, saving some resources to sleep.

For single encounter days, increased hp or reinforcements joining a battle in progress is better than increasing the number or CR of enemies. If you stick to normal hp and have all the enemies participate for the whole fight, there's very little difference between a fight that challenges a party with full resources and one that wipes them out.

[MaxWilson]

For single encounter days, increased hp or reinforcements joining a battle in progress is better than increasing the number or CR of enemies. If you stick to normal hp and have all the enemies participate for the whole fight, there's very little difference between a fight that challenges a party with full resources and one that wipes them out.

...and the difference is usually not on the DM's side, except for "how well does the DM telegraph the details of the challenge?" Beyond that the difference is all in what decisions the players make. Some people like me would argue fights where your decisions really matter are *the* reason you have fights and that busy adults shouldn't be forced to play through any fights where nothing is at stake except how many HP you lose before winning... But that view is not universal.

It's certainly true that if your goal is to have frequent fights, low PC turnover, and campaigns which take hundreds or thousands of hours of play to reach a climax, you cannot risk TPK in every fight. (I have zero of those three goals: I aim for combat-light, with Dark Sun-style character trees for each player, and ideally at least one dramatic climax every few hours of play if I'm doing my job right.)

[AdAstra]

Making the players feel pressured and under threat will often have a much bigger impact than the numbers arrayed against them. After all, the players can, in most cases, only ever see their own numbers, and what it takes to kill the enemies. You should ask whether you care more about the feeling of challenge or actual presence of it (though these are NOT mutually exclusive). Even if you care more about actual challenge, perceived challenge remains important (though easier to produce) to prevent your players from feeling unchallenged/getting themselves killed.

Powerful encounters fighting dumb will pale in comparison to weaker encounters playing to win and focusing fire, both in perceived and actual challenge (in deadly+ encounters, focusing fire can bring PCs down real quick, so depending on your style refraining from that may be recommended).

Consider non-stat effects and situations when calculating CR. If the monsters have set an ambush, consider that when deciding what difficulty you want to use. Same goes for favorable terrain (such as fortifications or even just cover) and additional objectives for the PCs/enemies (A VIP or some sort of ticking time bomb). This can be fairly easy if you consider how these factors actually affect the PCs. If, say, the enemies are in a fortification/far enough away that melee PCs will have trouble getting to them, you can effectively consider them out of the fight/ severely impaired for however long it takes to circumvent those obstacles.

Also consider monster synergies. Ranged attackers tend to do better when there are other creatures to tie down the PCs, that sorta thing.

[Chronos]

But what do the CR numbers mean? In 3rd edition, for instance, a CR N monster was theoretically an equal match for a single level N adventurer, or a routine fight (almost certainly won, but with some drain of resources) for a full party of level N adventurers. Yes, in practice, that often went right out the window (for starters, the classes were disparate enough that an equal match for a fighter wouldn't be an equal match for a wizard), but it at least provided a framework of what they were attempting.

Now, though, I look at CRs, and have no idea what they mean. If a lion is CR 1/2, does that mean that a first-level character should be able to solo two lions? Clearly not. But what does it mean?

[Protolisk]

But what do the CR numbers mean? In 3rd edition, for instance, a CR N monster was theoretically an equal match for a single level N adventurer, or a routine fight (almost certainly won, but with some drain of resources) for a full party of level N adventurers. Yes, in practice, that often went right out the window (for starters, the classes were disparate enough that an equal match for a fighter wouldn't be an equal match for a wizard), but it at least provided a framework of what they were attempting.

Now, though, I look at CRs, and have no idea what they mean. If a lion is CR 1/2, does that mean that a first-level character should be able to solo two lions? Clearly not. But what does it mean?

It's pretty much the same, according to the DMG.

A single monster with a challenge rating equal to the adventurers' level is, by itself, a fair challenge for a group of four characters.

So a lion is not solo-able by one level 1 character, but instead would be easy (not a fair fight) for a full a party of level 1s.

[MaxWilson]

But what do the CR numbers mean? In 3rd edition, for instance, a CR N monster was theoretically an equal match for a single level N adventurer, or a routine fight (almost certainly won, but with some drain of resources) for a full party of level N adventurers. Yes, in practice, that often went right out the window (for starters, the classes were disparate enough that an equal match for a fighter wouldn't be an equal match for a wizard), but it at least provided a framework of what they were attempting.

Now, though, I look at CRs, and have no idea what they mean. If a lion is CR 1/2, does that mean that a first-level character should be able to solo two lions? Clearly not. But what does it mean?

Nothing really, except that a CR X creature locked in a cage with four level X adventurers is practically guaranteed not to kill any of them by itself--and in practice it doesn't even guarantee THAT (see Intellect Devourers). But adventures don't take place in cages unless PCs forget to use doors or DMs refuse to let them...

So really CR is just a crude correlation with log(monster toughness x damage output) that lets you see which monsters are tougher, in a cage match, than other monsters. But trying to meaningfully compare CR to PC level is pointless. It is not in any way true that a CR 20 monster is too hard for a level 9 party--it depends on the monster and the party.

If you want to evaluate actual monster toughness you have to abandon CR and turn to Monte Carlo simulations, or just eyeball it. CR does not translate to difficulty in the real world.

[AdAstra]

Ok I think we’ve been dancing around the issue. A CR 1 monster is designed such that in general, it’s a Medium challenge for four level 1 PCs. A medium challenge isn’t all that challenging, it just means that the party is likely to lose some meaningful chunk of resources taking it down, be it health to spell slots to special abilities.

Like any system with any degree of complexity involving distilling all the features and traits of a monster down to a single number, this rarely tracks perfectly. It mostly allows you to ballpark it. Also note that this assume four PCs, so more players or player allies that aren’t part of class abilities will offset that.

Also note that comparative numbers are also important. The DMG has you treat encounters that significantly outnumber/are significantly outnumbered by the party as being worth collectively more “effective” xp from the perspective of challenge. Not more actual xp, though, just for the purposes of calculating the encountering

[MaxWilson]

Theoretically, what sort of metric would people want to see? I suppose in theory you could have something like:

Samurai Rating: to say that a monster is SR N means that 1 of these monsters locked in a cage with three featless level N Samurai is ~80% likely to reduce about one of those Samurai to 50% HP before the monster dies.

Then you could also have a

Balanced Party Rating: to say that a monster is BPR N means that 3 of these monsters locked in a cage with a level N featless party of two Samurai, one Evoker who casts a concentration spell (Telekinesis, Hold Monster, Hypnotic Pattern, Slow, Web, Tasha's) and then chucks evocations (Fireball, Magic Missile) on subsequent rounds, and one Swashbuckler are ~80% likely to reduce the entire party to about half HP before the monsters all die.

You could compute this rating using Monte Carlo techniques. Would it be useful to have SRs and BPRs for MM monsters?

[Zuras]

Nothing really, except that a CR X creature locked in a cage with four level X adventurers is practically guaranteed not to kill any of them by itself--and in practice it doesn't even guarantee THAT (see Intellect Devourers). But adventures don't take place in cages unless PCs forget to use doors or DMs refuse to let them...

So really CR is just a crude correlation with log(monster toughness x damage output) that lets you see which monsters are tougher, in a cage match, than other monsters. But trying to meaningfully compare CR to PC level is pointless. It is not in any way true that a CR 20 monster is too hard for a level 9 party--it depends on the monster and the party.

If you want to evaluate actual monster toughness you have to abandon CR and turn to Monte Carlo simulations, or just eyeball it. CR does not translate to difficulty in the real world.

It’s not quite that bad—the problem is that creature stats only provide the final CR of the creature, not a breakdown of the Offensive/Defensive CR.

When you build your own monsters and calculate their CRs using the rules in the DMG, you get a better handle on what creatures are glass cannons and which are bricks. Roughly speaking, a CR 5 creature is supposed to have a decent chance of KO’ing a 5th level PC after 3 rounds of combat, and should be able to last for 3 rounds against an average 5th level PCs damage output.

This is why you have a much higher threat/difficulty multiplier when you hit the third creature. Basically, if the math was done right those 3 CR X creatures have a decent chance of dropping one level X PC with focus fire on turn 1, which can start making things dicey for the players.

Also consider that including glass cannon creatures makes the fights more unpredictable. Even in white room simulations, 4 8th level PCs vs a CR 12 Archmage can end in anything between a TPK and the party taking zero damage, purely based on initiative rolls. In actual play things get even more unpredictable.

CR is only a rough guide, but it does tell you what to look at overall:

How many turns till the likely first PC KO?

How many turns till the enemy loses half or more of their damage-dealing capability?

CR only tells you the white room guesstimates for these, but knowing your party and the monsters you can further adjust for mobility, terrain, and other factors.

A pointless encounter is one that would take 4 or more rounds to down a PC, but the party can kill in two rounds or less using only at-will resources.

A deadly encounter is one where the enemies are likely to knock out one PC in less than the number of rounds they should stay operationally effective.

A beyond deadly encounter is one where actual TPK is possible, meaning there is a 25% or greater chance that the enemies could KO half or more of the party in the number of rounds you expect them to be operationally effective.

[MaxWilson]

This is why you have a much higher threat/difficulty multiplier when you hit the third creature. Basically, if the math was done right those 3 CR X creatures have a decent chance of dropping one level X PC with focus fire on turn 1, which can start making things dicey for the players.

Actually we can derive the difficulty multiplier without making any assumptions about PC HP at all. Three monsters will inflict six times as much damage as one monster before dying. If one Ogre can inflict 30 HP in the time it takes PCs to kill it, the other two ogres will also inflict 30 HP each in that time, and then you'll have two ogres doing damage until the second ogre is killed, and then one.

30 * (1 + 2 + 3) = 30 * 6 = 180.

Six times as much damage, divided among three monsters, means each monster is twice as effective as it is alone. If we treat CR as a rough proxy for "how much HP/resources will the monster consume before dying" (i.e. combat effectiveness) then it makes perfect sense that the multiplier for 3 monsters is x2, and this is true no matter whether the monster is a glass cannon or a tank.

[Zuras]

Actually we can derive the difficulty multiplier without making any assumptions about PC HP at all. Three monsters will inflict six times as much damage as one monster before dying. If one Ogre can inflict 30 HP in the time it takes PCs to kill it, the other two ogres will also inflict 30 HP each in that time, and then you'll have two ogres doing damage until the second ogre is killed, and then one.

30 * (1 + 2 + 3) = 30 * 6 = 180.

Six times as much damage, divided among three monsters, means each monster is twice as effective as it is alone. If we treat CR as a rough proxy for "how much HP/resources will the monster consume before dying" (i.e. combat effectiveness) then it makes perfect sense that the multiplier for 3 monsters is x2, and this is true no matter whether the monster is a glass cannon or a tank.

This doesn’t take AoE effects into account, and assumes the creatures would last at least one round individually, which is why glass cannon creatures throw off the calculations.

At a given CR, however, I like your elegant demonstration of the math showing 3 creatures simultaneously are 2X nastier than fighting them individually in sequence.

[MaxWilson]

This doesn’t take AoE effects into account, and assumes the creatures would last at least one round individually, which is why glass cannon creatures throw off the calculations.

At a given CR, however, I like your elegant demonstration of the math showing 3 creatures simultaneously are 2X nastier than fighting them individually in sequence.

It doesn't really assume that creatures would last one round individually though--if they each die in half a round, then one monster has a 50% chance of getting its damage off before dying (30 * 50% ~= 15), whereas 3 monsters would have have about a 50% chance of the first monster getting its damage off before dying (~15) and then sort-of-kind-of ~100% chance of the second monster getting its damage off before dying (30 * 100% ~= 30) and the third monster definitely getting its damage off in the first round (30), then on the second round the third monster has a 50% chance of getting its damage off (30 * 50% ~= 15). 15 + 30 + 30 + 15 is, again, 6x more than 15, and 6x / 3 monsters = double effectiveness per monster.

Look up Lanchester's Laws in wikipedia: https://en.wikipedia.org/wiki/Lanchester%27s_laws My interpretation of WotC's difficulty multipliers is that they are sort of, kind of, using the modern military guesstimate of 3/2 scaling, which is a compromise used because some modern military power scales with Lanchester's Square Law (like gunfire, or arrows in 5E, or melee attacks once all enemies are close enough) and some scales with Lanchester's Linear Law (artillery fire, or Fireballs, because having more troops increases your offensive power but also increases how much damage you take from AoEs). If I look at WotC's difficulty multipliers it roughly tracks 3/2 power scaling.

However, I agree, at the low end (my example of 3 creatures) and the high end (more than 16 creatures) the WotC difficulty multiplier formula absolutely does not take AoEs into account, so it tends to overestimate the difficulty of large fights up to a certain point, although small XP values for low CR simultaneously leads to undervaluing the difficulty of large fights against many weak creatures, especially if they have missile weapons so that AoEs don't work well against them.


This brings us back to my conclusion: if you want to predict real-world difficulty, you're better off eyeballing it and doing some BOTE calculations than relying on CR. : ) Looking at total enemy HP and total enemy DPR against AC 17 is not a terrible metric: 7 Red Fangs of Shargas is ~350 hp of orcs (AC 15 with disadvantage to attackers), attacking at +5 with advantage for 42d6+42 of damage (~146 points of damage against AC 17), and they have missile weapons and high mobility... yeah, that's going to be a hard fight for a 9th level party, harder than a Purple Worm even though the Purple Worm has a slightly higher total XP cost (13,000 vs. 12,250) especially if the Purple Worm uses stereotypical Purple Worm tactics (eat everything) instead of Tremors-style hit-and-run w/ burrowing.

[Kurt Kurageous]

It's less official but I recommend looking at the Angry GM's latest blog post.

Ditto. Angry is going in his own direction, but he speaks truth about the assumptions underlining CR.

[MaxWilson]

Ditto. Angry is going in his own direction, but he speaks truth about the assumptions underlining CR.

I find Angry GM almost unreadable (signal-to-noise ratio is awful) but you've interested me. Looks like the most recent rant is called Talk Is Cheap (https://theangrygm.com/talk-is-cheap-so-is-thought/). Is that the one I should be reading?

(I read about two paragraphs without seeing anything CR-related and then zoned out, but if there's a good point buried therein I'll take another shot at it.)

[stoutstien]

I find Angry GM almost unreadable (signal-to-noise ratio is awful) but you've interested me. Looks like the most recent rant is called Talk Is Cheap (https://theangrygm.com/talk-is-cheap-so-is-thought/). Is that the one I should be reading?

(I read about two paragraphs without seeing anything CR-related and then zoned out, but if there's a good point buried therein I'll take another shot at it.)

https://theangrygm.com/f-cr-theres-a-better-way-part-2/

Your right about content to text ratio but he points out a pretty big issue with using CR to build good encounters.