12
\$\begingroup\$

My sorcerer just hit 5th level, so he needs to consider whether he's going to take Fireball, however, he really likes his Chromatic Orb and branding is important. So I started thinking about if there might be a point when Chromatic Orb could realistically deal more damage than Fireball.

The Chromatic Orb spell says:

You hurl an orb of energy at a target within range. Choose Acid, Cold, Fire, Lightning, Poison, or Thunder for the type of orb you create, and then make a ranged spell attack against the target. On a hit, the target takes 3d8 damage of the chosen type.

If you roll the same number on two or more of the d8s, the orb leaps to a different target of your choice within 30 feet of the target. Make an attack roll against the new target, and make a new damage roll. The orb can’t leap again unless you cast the spell with a level 2+ spell slot.

Using a Higher-Level Spell Slot. The damage increases by 1d8 for each spell slot level above 1. The orb can leap a maximum number of times equal to the level of the slot expended, and a creature can be targeted only once by each casting of this spell.

and Fireball says:

A bright streak flashes from you to a point you choose within range and then blossoms with a low roar into a fiery explosion. Each creature in a 20-foot-radius Sphere centered on that point makes a Dexterity saving throw, taking 8d6 Fire damage on a failed save or half as much damage on a successful one.

Using a Higher-Level Spell Slot. The damage increases by 1d6 for each spell slot level above 3.

The average per target damage for fireball at 3rd level is 28 (3.5*8) assuming no targets make their save, and the damage increases per slot at a rate of 3.5 per spell level.

Were I to upcast Chromatic Orb to 3rd level, the damage would increase to 5d8 and thereby have an average damage of 22.5 (4.5*5), and the damage increases at a rate of 4.5 per spell level.

Given the baseline disparity, it seems unlikely I can ever increase the damage of Chromatic Orb to even equal that of Fireball, however, I'm a Sorcerer who usually has access to the Innate Sorcery feature which allows Advantage for attack rolls. And critical chromatic orbs are a lot of damage given the slot cost in 1st tier play.

For the purposes of this question, I am trying to figure out at what spell level is the break even point between Chromatic Orb and the Fireball spell with regards to per target damage. For the purposes of calculations, assume the following:

  • Assume all targets are in some sort of cluster that allows for them to be valid targets for a single casting of either spell (e.g. Chromatic Orb cast at 3rd level, can hit as many as 4 targets, so assume fireball hits 4 targets as well).
  • Chromatic orb will hit its target unless the attack roll is a 1 when rolled with Advantage (i.e. barring the 1/400 chance, a break in Chromatic Orb's jumps will only happen if the caster fails to roll doubles on the d8s)
  • All targets of the fireball fail their save (i.e. full average damage for each target)

I initially started trying some calculations for this, but found myself getting confused when trying to factor in Chromatic Orb's jumps only if doubles are rolled on the damage dice. Can someone help me figure out if a break even point exists between these spells? If one does not exist with these base parameters, at what percent of successful saves vs. fireball is necessary to help make a break even occur?

\$\endgroup\$
6
  • 3
    \$\begingroup\$ Should an answer include cases in which the targets are vulnerable or resistant to the damage type, or are we assuming a chromatic orb cast as fire damage only, for a direct comparison? \$\endgroup\$ Commented Jan 17 at 22:27
  • \$\begingroup\$ I understand the thought behind assuming that fireball hits the same number of creatures (for purposes of comparison), but... Fireball can affect something like 52 medium sized creatures on the ground under perfect (albeit unlikely!) circumstances, so it feels a bit forced. Am I missing the point? \$\endgroup\$ Commented Jan 18 at 6:04
  • \$\begingroup\$ In the interest of clarify, let's keep discussion of how to connect this Q to real scenarios in chat. Keep it there, unless it can graduate to a full frame challenge answer, though with the specificity in the question, do consider whether it'd be better as its own question. \$\endgroup\$ Commented Jan 19 at 17:03
  • 1
    \$\begingroup\$ I think you need to take onto account more then just the damage output, but also the tactical layout. Chromatic orb allows you to hit any creature within 30 feet of the last creature hit, this means at level 4 you could get 4 targets and target 1 and 4 could be 120 feet apart, in addition you can pick your targets to ensure you hit the ones with a lower AC while Fireball you are restricted only to the area of target \$\endgroup\$ Commented Jan 21 at 16:02
  • 1
    \$\begingroup\$ @RichardC I am aware of this, however, if I can hit 4 enemies with a fireball then I can also hit those same 4 enemies with a Chromatic Orb. Given the variables on this, I elected to simplify things and just evaluate the damage per the criteria described. \$\endgroup\$ Commented Jan 21 at 17:29

2 Answers 2

14
\$\begingroup\$

Under the stated conditions, chromatic orb pulls ahead at level 6

The expected damage is proportional to the expected number of damage dice

Consider each potential damage die labeled by \$(c, d)\$, where \$c\$ is the index of the creature hit and \$d\$ is the index of the die within those that hit that creature.1 By linearity of expectation, the expected damage of chromatic orb is equal to the sum of the expected damage contribution \$E\left[x_{c,d} \right]\$ of these possible damage dice, even though they are not independent:

$$ E\left[X\right] = \sum_{c, d} E\left[x_{c,d} \right] $$

In turn, since whether we got to use each potential damage die is already determined before we roll it, it's not possible for the number it rolls to be correlated with whether we got to use it. Thus, the expected damage contribution of each die is the probability \$p_{c, d}\$ of us getting to use it times its average damage of 4.5:

$$ E\left[x_{c,d}\right] = 4.5 p_{c, d} $$

which makes the overall damage

$$ E\left[X\right] = \sum_{c, d} 4.5 \; p_{c, d} = 4.5 \sum_{c, d} p_{c, d} $$

So the expected total damage is simply 4.5 times the expected number of damage dice.

The expected number of damage dice

The full distribution of the number of damage dice is a reasonable computation. Here is an example using my own Icepool Python package:

from icepool import d, map def attack_roll(to_hit): """Returns a die with 2 = crit, 1 = hit, 0 = miss.""" def hit_type(x): if x == 20: return 2 elif x >= to_hit: return 1 else: return 0 return d(20).map(hit_type) def step(total_d8s, can_chain, hit_type, base_d8s): """Updates the total number of d8s and whether chaining can continue.""" if not can_chain: return total_d8s, False next_total_d8s = total_d8s + hit_type * base_d8s if hit_type * base_d8s >= 9: # pigeonhole principle next_can_chain = True else: next_can_chain = d(8).pool(hit_type * base_d8s).largest_count() >= 2 return next_total_d8s, next_can_chain print('| Level | Chromatic orb | Fireball | Ratio |') print('|-------|---------------|----------|-------|') for level in [3, 4, 5, 6, 7, 8, 9]: base_d8s = level + 2 result = map(step, (0, True), attack_roll(2).highest(2), base_d8s, repeat=level + 1) orb_mean = result.marginals[0].mean() * 4.5 fireball_mean = (level + 5) * (level + 1) * 3.5 print(f'| {level} | {orb_mean:0.1f} | {fireball_mean:0.1f} ' + f'| {orb_mean / fireball_mean:0.2f}') 

You can try this in your browser here. You can change the attack roll by changing attack_roll(2).highest(2); for example, hitting on 6s but with Elven Accuracy would be attack_roll(6).highest(3).

The results:

Level Chromatic orb Fireball Ratio
3 74.2 112.0 0.66
4 128.1 157.5 0.81
5 197.0 210.0 0.94
6 272.1 269.5 1.01
7 351.7 336.0 1.05
8 439.1 409.5 1.07
9 536.0 490.0 1.09

If spell slots continued indefinitely above level 9, eventually chromatic orb would fall behind again due to the the miss chance putting an asymptote on the expected number of hits. Then again, you can also only fit so many creatures in a fireball's radius.

Final notes

Note that this gives the correct expected damage (because linearity of expectation holds regardless of independence), but you can't just sample the number of dice and roll that many d8s to get the correct distribution of damage (because this can be affected by correlation -- though I expect it to be fairly close in this case).

If you prefer, you could instead apply linearity of expectation on the damage dealt to each target rather than individual dice, in which case you arrive at something like Eddymage's answer. Equivalently, you could start with the fact that expected damage is proportional to the number of damage dice and apply linearity of expectation a second time on the expected number of damage dice per target.


1 \$c \in \left\{1 \ldots \ell + 1\right\}\$ and \$d \in \left\{1 \ldots 2 \left( \ell + 2 \right)\right\}\$, where \$\ell\$ is the spell slot level, and the double number of damage dice is due to the possibility of crits.

\$\endgroup\$
4
  • \$\begingroup\$ The random variable representing the damage is subtly different between the last target and all the others since the earlier ones necessarily exclude any results where there are no duplicate numbers. I suspect that this is a symmetrical change so it wouldn’t change the expected value but it should be checked. \$\endgroup\$ Commented Jan 18 at 4:08
  • 1
    \$\begingroup\$ Any of the \$c\$ could be the actual last target, not just \$\ell + 1\$, which are exactly those "missing" results where there are no duplicate numbers. In fact, there is no symmetry required; for example, linearity of expectation works just as well for computing the expected value of an exploding die. \$\endgroup\$ Commented Jan 18 at 5:06
  • 1
    \$\begingroup\$ +1 I just had to commit this to memory... To kill the goblins while they zorb, / from 3 to 5 just blast with fire, / but at 6 it's either-or. / Let 7 to 9 be fueled by ire, / as you hurl kaleidoscopic orbs. \$\endgroup\$ Commented Jan 19 at 9:19
  • 1
    \$\begingroup\$ I'm having a very difficult time understanding this answer. You're using what I assume is a summation function, but I'm unclear on why. Is it feasible to present this algebra in lieu of more complex functions and Python coding? It's possible that's infeasible, but as written, I can't assess if this answer is correct or not. \$\endgroup\$ Commented Jan 20 at 15:25
4
\$\begingroup\$

The break-even point is at 6th slot

Some preliminary computations:

  • Succeeding in attacking with advantage with a target AC of 2 has a probability of 0.9975.
  • The probability to get a nat 20 with advantage is 0.0975.
  • The expected damage of an attack is \$p\cdot D\$, where p is the probability to hit and D is the expected output of the dice.

Basic case: Level 3 Slot

At level 3, the expected dice damage of the 1st hit is \$D=23.6250=4.5\cdot5\cdot1.0975\$, where the 1.0975 accounts also for critical hit. The maximum number of further leaps is 3, for a grand total of possible hits equal to 4, and the i-th leap occurs if the (i-1)-th hit, starting from i=2.

The probability to have at least 2 equal values on the 5 dice for the i-th leap is \$P_{2}(l_i)=1-P_{\neq}(l_i,5)\$, where \$P(l_i,5)_\neq\$ stands for the probability of having all different values. One has $$ P_{\neq}(l_i,5)= \frac{8\cdot7\cdot6\cdot5\cdot4}{8^5}, $$ since one has 8 possibilities for the 1st die, 7 for the second, 6 for the 3rd and so on. Moreover, this probability is the same for each leap, so we'll call it just p. Checking also the dice from the critical hit leads to a probability of $$ P_2(l_i) = (1-0.0975)(1-P_\neq(5)) + 0.0975(1-P_\neq(10)) $$ where the notation has been shortened. Denote with h the probability to hit: therefore, the expected damage for the 3rd level Chromatic Orb is $$ \begin{eqnarray*} E[dmg] &=& hD + h^2P_{2}(l_1)\cdot D + h^3P_{2}(l_1)\cdot P_{2}(l_2)\cdot D+ h^4P_{2}(l_1)\cdot P_{2}(l_2)\cdot P_{2}(l_3)\cdot D\\ &=& hD + h^2pD + h^3p^2D+h^4p^3D \\ &=& D\sum_{i=0}^3h^{i+1}p^i\\ &=& 74.2 \end{eqnarray*} $$

Generalization for further upcasting

The general formula for \$P_\neq\$ with k leaps, so for the further upcasting, is $$ P_\neq(k) = \left(\displaystyle\prod_{i=8}^{9-k}i\right) \Big/ 8^k $$ which naturally leads to the formula for \$P_2\$. For k=9, \$P_2=1\$, obviously. Hence, the comparison with Fireball is now straightforward:

Level Chromatic orb Fireball
3 74.2 112.0
4 128.1 157.5
5 196.9 210.0
6 272.0 269.5
7 351.6 336.0
8 439.0 409.5
9 535.8 490.0

The break-even point is at level 6, from a pure damage point of view.

As observed in HighDiceRoll's answer, Fireball will overcome Chromatic Orb in impossible scenarios, when slot levels increase towards infinity. plot of the damage output: the second break even point occurs under impossible circumstance

Other considerations

The two spells are quite different in usage: Fireball is the signature spells for Wizards and the like and can hit several enemies at once, but, unless one increments their initiative bonus for casting at the beginning of the fight, it requires precise situations for avoiding friendly fire. Moreover, this spell deals only fire damage, unless some game feature (or the DM) allows to change it. On the other hand, Chromatic Orb is a more versatile from this point of view, since the caster can chose the damage type and hence adapt it to the enemy, but it deals less damage and hits a lower number of creatures.

The major consideration, at least for me

For my characters, I prefer role playing and personalization, and since it seems to me that it is the same for you, I suggest to take whatever you like to play, disregarding min/max strategies.

\$\endgroup\$
17
  • \$\begingroup\$ I run also several simulations that confirm the theoretical results, but I did not included them in this answer. \$\endgroup\$ Commented Jan 20 at 22:49
  • \$\begingroup\$ There is a possible wrinkle, which is whether and how the critical damage dice are counted when determining whether we are eligible for a leap. In my answer I assumed the crit dice get added to the original pool for that target, thus potentially increasing the chance of a leap, but I can see other possible interpretations. \$\endgroup\$ Commented Jan 20 at 23:46
  • \$\begingroup\$ @HighDiceRoller I see what you mean, it should be the reason for the discrepancy in expected damage. Tomorrow I will check that. \$\endgroup\$ Commented Jan 20 at 23:57
  • 1
    \$\begingroup\$ @Pyrotechnical yes, but I did not consider advantage here, just that each hot is auccesfull. If it is so important, I can consider it as well as the possibility to miss. Do you find the explanation enough clear for the computation process? \$\endgroup\$ Commented Jan 21 at 7:18
  • 1
    \$\begingroup\$ @Obie2.0 And this is a complaint that you should do under the question and not under just one of the answers. \$\endgroup\$ Commented Jan 22 at 7:51

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.