D&D 5E Fixing Challenge Rating

[tomedunn]

I've put the bones of the system up on GitHub:

GitHub - mikemearls/5e_point_encounters: A system for balancing encounters in 5e using point values

A system for balancing encounters in 5e using point values - mikemearls/5e_point_encounters

github.com

The math is still early, so expect changes as I spin up some code to run a deeper analysis of the monsters and characters in the 5e SRD. Hit me up here with any questions or comments.

This is very similar to the encounter building system used in Pathfinder 2e, even to the point where monster point values are doubling every two levels. I ran an analysis of the PF2 encounter building rules and showed their baseline monster stats do double in combat power roughly every two levels.

However, a lot of that power comes from the fact that their attack bonuses and armor classes scale include their level in them. This isn't the case with monsters in DnD, which means their points value should scale slower than that. By my calculations they should double, roughly every four levels instead (past CR 5 at least).

There's still the question of how, or whether to account for the number of monsters. From my understanding, the PF2 rules get around this by how their critical hits work, which DnD 5e doesn't have. However, people generally seem to like the encounter building rules in Xanathar's Guide to Everything which doesn't adjust for the number of monsters, so maybe doing so isn't all that necessary. It certainly seems to cause more troubles than its worth based on how I see other people apply it, but from a mathematical perspective I find it quite elegant.

 

[mearls]

Will you be working with others, or open to chats with other designers? I'd love to pick your brain!

I'm thinking of something along those lines, likely a Patreon with a few different tiers, but we'll see.

 

[mearls]

I think any realistic CR system has to include the PCs. A PCs level is just a terrible metric for what a specific party can bring to the table. What you really want is for each PC to be run through the system (AC, average damage utilizing their "normal" attack plan, etc), and then they get a PCR number.

From there you can match Cr to PCR to get a better approximation of challenge to THAT specific group of players, which is the real goal of the system.

Great point!

Long-term, I think something like that should be possible. The math will be a bit different on the two sides of the equation, but I think we can bridge.

We expect monsters to go to 0, so we need to measure how much damage they do each round and how many rounds it takes for the characters to put them at 0.

For PCs, we assume survival except in extreme cases, so we're more interested in figuring out resource loss.

It's a subtle distinction, but the net of it means that the measures need to be comparable but arrived at by different methods.

The nice thing is that, as you point out above, it means we can start to measure specific characters and potentially come up with points for any monster from any source.

 

[mearls]

How do you gauge the balance and point power of spells vs martial stuff, how is a 3 attacks equal to a 5th-6th level spell.

This is the basic sticking point. Each class having its own power floor and power peak makes things messy. It's legacy code that a lot of people run, so we have to address it.

For the first pass, I converted all the spells of a caster of a given level into a very rough output per action. That works for a first pass, but it will not work for a final assessment. From there, I'm going to model out a few different techniques to see what gets closest. For instance, on my next pass I'm going to cherry pick the optimal value for each facet of a character and see how that maths out.

So that baseline becomes a character that gets the best of what any character can have:

• Barbarian hit points
• Top tier spell cast with each action
• Etc

The purpose of this test is to get a sense of the best case scenario for the PCs. I think that having a power ceiling (or something close it) could be helpful.

 

[mearls]

There's still the question of how, or whether to account for the number of monsters. From my understanding, the PF2 rules get around this by how their critical hits work, which DnD 5e doesn't have. However, people generally seem to like the encounter building rules in Xanathar's Guide to Everything which doesn't adjust for the number of monsters, so maybe doing so isn't all that necessary. It certainly seems to cause more troubles than its worth based on how I see other people apply it, but from a mathematical perspective I find it quite elegant.

My analysis so far uses actions as the basic currency. IME, the challenge with ramping up the number of monsters comes down to:

• Overwhelming the party with a horde of little guys, because the PCs simply don't have enough actions to deal with them all in a timely manner
• The PCs using action denial to nuke a single creature. Legendary actions were a brute force hack around that.

My current encounter work uses actions as the basic currency. Very roughly speaking:

• DMs think of encounter building in terms of literal combo creation. These orcs act quickly and throw out ranged debuffs. These orcs act in the middle of the round and focus on area control. The last guys up are pure damage per action. The DM's encounter building focuses on finding a way to combo up interesting monsters according to their associated actions.
• To go deeper into the 40k analogy, the DM has some suggested guidelines by monster category to help deal with this. This is still very early, but it uses party size to suggest the maximum number of minions, bosses, and so on. A complex boss might be rated as appropriate for parties of 5+ characters, and so on.

 

[Stormonu]

I think any realistic CR system has to include the PCs. A PCs level is just a terrible metric for what a specific party can bring to the table. What you really want is for each PC to be run through the system (AC, average damage utilizing their "normal" attack plan, etc), and then they get a PCR number.

From there you can match Cr to PCR to get a better approximation of challenge to THAT specific group of players, which is the real goal of the system.

Don't forget their saves as well. If the monsters were to use abilities that target's the group's weak saves that can have a dramatic difference than attacks against the PC's strong saves (and/or resistances).

My biggest beef with the current CR system is it doesn't account for non-hp attacks or abilities (stun, banish, walls) very well, so you get wildly off on creatures like Ghouls or Shadow's Strength attacks.

Also, it always assumes the two groups want to close and engage. Evasive/Hidden/Hit-and-Run tactics on either side's part for certain monster roles should possibly be considered. Running up and beating on a bugbear sentry plays out a lot differently than getting ambushed by one, for example. Same with a dragon on the wing using strafing attacks vs. one landing and engaging. I guess these would be "Encounter Factors", but should be something brought to the DM's attention when building an encounter rather than expecting them to think of it for themselves and hoping they apply such factors to the difficulty.

 

[Trekiros]

Ooh, I'm interested to see where this ends up going in the future.

I'm curious, by any chance, have you watched this video, and if so, what are your thoughts on it? I made it last year, and it was very positively received, and it's also about trying new approaches to this little math problem.

 

[mearls]

Ooh, I'm interested to see where this ends up going in the future.

I'm curious, by any chance, have you watched this video, and if so, what are your thoughts on it? I made it last year, and it was very positively received, and it's also about trying new approaches to this little math problem.

That's pretty amazing. Also incredible stuff for your sixth video.

I think a lot of what you have there is spot on. I'm definitely going to use Battle Sim to check predictions the system makes versus the reality. That's an amazing community resource!

 

[M.T. Black]

Hi Mike,

Welcome to the site and I'm sure everyone looks forward to hearing more from you.

This looks like a pretty interesting approach. At some point, it would be good to have it embedded in a public google sheet - that would make it very easy to play with and I'm sure you'd get lots of feedback.

For the last couple of years, I've mostly relied on @SlyFlourish's Lazy Encounter Benchmark when I'm putting together encounters - even those in my published adventures. You can see it here: https://slyflourish.com/the_lazy_encounter_benchmark.html. I'd welcome a more sophisticated system, though.

cheers,
MTB

 

[John Lloyd1]

If I'm reading this right, I would need to recalculate each monster point value if the party composition changes?

For context, I'm running AL at a LFGS and I never know my exact party composition until the start of the session. And then some may turn up late. So, I would prioritise being able to make those adjustments on the fly.