Introduction

Chapter 9 of the Dungeon Master’s Guide (DMG) contains a section called “Creating a Monster” which outlines how to calculate CRs for homebrew monsters. By the rules own admission, this calculated CR is intended to act more as an initial estimate than a final value, with the final CR determined through playtesting.

Having CR ultimately determined through playtesting makes sense given the complexity of the task at hand. However, it also raises the question: how accurate are the CR calculations presented in the DMG?

In this post, I aim to answer this question by taking a detailed look at how CRs for monsters in official 5th edition products compare to those calculated using the method outlined in the DMG. You can find a summary of the monster dataset used for this analysis here.

Calculating CR

Without going into too much detail, here’s a high level summary of how to calculate CR according to the DMG. First, determine the monster’s effective hit points (HP), armor class (AC), average damage per round (DPR), and attack bonus (AB) or saving throw difficulty class (DC). Second, compare those with values listed in the Monster Statistics by Challenge Rating table to determine the monster’s defensive CR (D-CR) and offensive CR (O-CR). And finally, average the monster’s D-CR and O-CR together to get the monster’s CR.

For simple monsters, like a brown bear, that have no special abilities, resistances, immunities, and such, this is a straight forward process. However, for more complicated monsters adjustments need to be made to the monster’s HP, AC, DPR, and AB in order to make the calculations work. I’ll talk a bit more about these adjustments later on.

Defensive CR

Defensive CR (D-CR) is calculated from a monster’s effective HP and AC. The most common adjustments to a monster’s HP comes from resistances and immunities to multiple damage types, health regeneration, healing, and legendary resistances. HP is by far the stat that gets the most adjustments, and this is especially true for higher CR monsters.

For AC, the most common adjustments are abilities that give attackers disadvantage on attacks, reactions that temporarily raise AC, and proficiencies in multiple saving throws. Like adjustments to HP, these adjustments play a larger role at higher CRs than they do at lower CRs.

Figure 1: Hit points and armor class vs CR (adjusted and unadjusted).

Figure 1 shows how published creatures compare, on average, against the HP and AC targets listed in the DMG.

For HP, while average effective HP closely matches the DMG target, the average HP consistently falls short. For AC, though, the average AC generally match the DMG target, while the average effective AC comes out consistently higher than expected with the biggest difference coming from monsters CR 17 and above.

The differences in effective AC might seem like a large discrepancy, but AC only acts as a minor adjustment to a monster’s overall D-CR so the overall impact is relatively small. This impact is reflected in the calculated D-CR, which is shown in the Fig. 2 below.

Figure 2: Calculated defensive CR mean and 60% confidence interval vs CR.

Overall, D-CR appears to be on target, with the average D-CR for published monsters coming within +/- 1 of target. It’s worth noting that the biggest differences come from the extreme ends of the CR spectrum, with very low CR monsters showing consistently lower D-CR than expected due to low effective HP, and very high CR monsters showing consistently higher D-CR than expected due to high effective AC.

Offensive CR

Offensive CR (O-CR) is calculated from a monster’s effective DPR and AB. DPR does have adjustments which typically take the form of damage that relies on specific conditions being true, such as the Assassin’s Assassinate feature which allows them to score a critical hit against surprised creatures. However, I did not differentiate these sources of damage when calculating DPR for each monster and so, for the time being, they are not split apart from the effective DPR. The most common adjustments to AB are abilities that give advantage on attack rolls.

Figure 3: Damage per round and attack bonus vs CR (adjusted and unadjusted).

Figure 3 shows how published creatures compare, on average, against the DPR and AB targets listed in the DMG.

Effective DPR closely matches the expected DPR values from the DMG for all but the highest CRs which tend to be a bit lower than expected. The DPR values shown here only include effective DPR. As mentioned earlier, there are adjustments made to DPR but these tend to be rare and relatively minor and its unlikely the unadjusted DPR differs significantly from the effective DPR values.

Effective AB shows a similar story to effective AC. It is generally higher than the expected value and that difference is largest at higher CRs. However, unlike AC, there is little to no difference between effective AB and unadjusted values.

The combined impact of effective DPR and AB is shown in Fig. 4 below which shows how the average O-CR compares with the expected values in the DMG.

Figure 4: Calculated offensive CR mean and 60% confidence interval vs CR.

On average, the calculated O-CR closely matches the expected target and is rarely more than +/- 1 away across the full range of CRs. The width of the distribution is also generally narrower than what was observed for D-CR, except for CRs between 12 and 20 where the distribution is generally wider. This widening of the distribution is due to the increased frequency of legendary monsters which tend to deal a bit more damage than non-legendary creatures within this CR range.

CR

With D-CR and O-CR both calculated, the CR for each monster is calculated by taking the average of the two. The results of this are shown in Fig. 5 below.

Figure 5: Calculated CR mean and 60% confidence interval vs CR.

As with D-CR and O-CR, the calculated CR for published monsters closely matches the listed values on average. It’s interesting to note that the calculated CR is more accurate than either of the component D-CR and O-CR.

Conclusion

In conclusion, the methods for calculating CR in the DMG are generally pretty accurate at predicting a monster’s final CR. Effective AC and AB were generally about +2-4 higher than expected, but not enough to seriously overpower typical monsters.

The large difference between effective HP and actual HP values for monsters poses an interesting problem that I’m sure many DMs grapple with regularly. It indicates that a significant portion of most monster’s effective health is dependent on the PCs strengths and weaknesses as well as how the PCs choose to engage the monster. However, that’s a discussion for another time.