Why do we round down when multiclassing EK/AT?

[BarneyBent]

Something I just realised while doing some theorycrafting, and feel like a dolt for not noticing it before.

According to the multiclass rules, you round down when multiclassing across 1/3rd casters (Eldritch Knights and Arcane Tricksters). But when you play these as single classes, you round UP to get their effective spellcasting level.

For example, a level 3 EK has 2 first level spell slots, which is what you’d expect from a level 1 full caster. A level 4 EK has 3 first level slots, which you’d expect from a 2nd level full caster. In other words, the 1.333etc you get when diving the EK level by 3 is rounded up to 2. This pattern holds across the full spell slot progression of EK and AT.

But if I was to multiclass EK/AT together, all of a sudden I start rounding down, and my 4 (or 5) EK levels are now only good for 1 full caster level.

The net result is that an EK 5/AT 5 has the same number of spell slots as a plain old EK 4, both being second level casters. That is just bizarre to me.

Similarly, an EK 4-5 or AT 4-5 taking a full caster level gains no spell slots at all.

This feels like an oversight. Rounding down makes sense for half casters as it follows the same pattern as their single class spell slot progression. But third casters round up. It actively discourages multiclassing third casters with other spellcasters.

Does anybody houserule this? Would it be OP to do so? Have I overlooked some important balance reason why this would be the case?

[Malifice]

A: Because there is supposed to be a price for Multiclassing.

It's by design.

[BarneyBent]

A: Because there is supposed to be a price for Multiclassing.

It's by design.

A price that applies to no other spellcasting classes? A price so disproportionate that it essentially removes most of the point of multiclassing the third-casters together? There’s almost no point taking AT with EK - you’d be better off taking a different Rogue or Fighter subclass.

[twomp01]

Note: The same is more or less true for half casters: Rangers/Paladins get their 2nd level spells at level 5, instead of at level 6.

[Malifice]

A price that applies to no other spellcasting classes? A price so disproportionate that it essentially removes most of the point of multiclassing the third-casters together?

The same rule applies to 1/2 casters.

A Paladin 3/ Ranger 3 is caster level 2 (3 x 1st level spells per long rest). If he had 6 levels in one of those two classes instead of the 3/3 split, his casting level is 3rd (4/2 slots).


There’s almost no point taking AT with EK - you’d be better off taking a different Rogue or Fighter subclass.

That's not true; you just get the best spellcasting return taking levels in increments of 3 (or leaving the class to advance in a different level at 3rd, 6th, 9th etc).

An EK 6/ AT 3 has the same spellcasting as a 9th level member of either class, and as long as he takes his remaining levels in those two classes in increments of 3 (his next 3 levels gained increase either AT or EK) he loses nothing (in terms of slots/ day) over what he would have should he have instead been single classed.

A 5/5 split is bad (caster level 2, when a single classed character is CL 3). You get 3 slots/ day, instead of 4/2. Of course, you also have extra attack AND uncanny dodge.

A 7/3 split however is no different (in terms of slots/ day) than would be a 10th level single classed 1/3 caster. Both are CL 3 (4/2 slots).

TL;DR - Take your levels in those classes in '3's if you dont want to miss a level of casting here or there. Ditto with Paladins and Rangers (except in increments of '2').

[BarneyBent]

Note: The same is more or less true for half casters: Rangers/Paladins get their 2nd level spells at level 5, instead of at level 6.

Yeah, you’re right though it makes much less of a difference. It’s only ever 1 level off either way for a half caster, whereas for a third caster you lose all benefits of multiclassing them together unless they are both taken to levels that are multiples of 3.

Not only that, but it totally removes synergy from two classes that should, on paper, have it.

You have to sacrifice to multiclass as it is. Applying such a severe penalty is very tough. Even if you just say “round to the nearest whole number”, that applies a small nerf of similar magnitude to the half-casters, without totally wrecking the class’s spell slot progression.

[Malifice]

Yeah, you’re right though it makes much less of a difference. It’s only ever 1 level off either way for a half caster, whereas for a third caster you lose all benefits of multiclassing them together unless they are both taken to levels that are multiples of 3.

If you want to prioritise spellcasting, take them in levels of 3.

If you want to prioritise feats, take them in levels of 4.

Or simply bail on EK at 6th (CL2 and 2 x feats, plus extra attack).

Out of curiosity is the formula [Ranger + Paladin]/2 (round down) or [Paladin/2 round down] + [Ranger/2 round down].

It's not entirely clear from the RAW.

[BarneyBent]

If you want to prioritise spellcasting, take them in levels of 3.

If you want to prioritise feats, take them in levels of 4.

Or simply bail on EK at 6th (CL2 and 2 x feats, plus extra attack).

Out of curiosity is the formula [Ranger + Paladin]/2 (round down) or [Paladin/2 round down] + [Ranger/2 round down].

It's not entirely clear from the RAW.

Yes, you can do that, you can work around this obviously. But it makes it much more difficult than it otherwise would be.

Regarding that formula, that’s actually a great question, and if you apply the first one to both half and third casters you have a much better balance IMO.

[Chronos]

Extreme example: An Eldritch Knight (level 3n+2) decides that he'd like to focus more on spellcasting instead of martial combat, and multiclasses to wizard. He'll actually end up with less spellcasting by switching to a primary spellcasting class than if he had stayed with the incidental-spellcasting class.

[Malifice]

Extreme example: An Eldritch Knight (level 3n+2) decides that he'd like to focus more on spellcasting instead of martial combat, and multiclasses to wizard. He'll actually end up with less spellcasting by switching to a primary spellcasting class than if he had stayed with the incidental-spellcasting class.

No, it should stay the same.

At 5th level he has 3 slots/ long rest.

He MC's to Wizard at 6th. He now calculates his caster level as (5th level EK = CL 5/2) or CL 1 plus (1st level Wizard) = +1.

The formula spits out a CL of 2. So 3 slots/ long rest. Same slots he had before dipping Wizard.

This might hold true if he took a Level of Paladin instead of EK, but you only add in your second and subsequent classes to the formula if they have the 'spellcasting' class feature; which Paladins dont get till 2nd level.

So even should our 5th level EK (3 slots/ LR) take a single level of Paladin, he remains at 3 slots per long rest. He ignores his Paladin level for the purposes of Multi-classing spell slots, and uses the Spells progression chart for Eldritch knight of 5th level (and not the MC spell progression chart).

When he hits 2nd level of Paladin, he will finally have the Spellcasting class feature from 2 or more classes, so the rules for MC'd spellcasting apply, and he has to use the formula for MC'd spellcasters.

His CL at this point becomes 5/2 (round down) or 1, plus 2/1 or 1 for a total of CL 2 (so he retains 3 slots per long rest).

He didnt gain any slots from those 2 Paladin levels, but he didnt lose any either.

One more level in EK grants him CL 3 (and 4/2 spells per LR) as would 2 more levels in Paladin.

It works. At no point do your spell slots/day go down as a MC'd spell caster.

[tieren]

@malifice

Everything you put was true, but I think you are missing the comparison he was trying to make between an EK5/Wiz1 and an EK6.

As you point out the EK5/Wiz1 stays with the same spell casting power, but if the PC had instead of dipping taken another level of EK he would have 2/4 slots. So by taking the full caster dip he ends up with less spell casting than taking the 1/3 caster level.

Personally I don't think the whole point of the MC is to get more spell slots. You get the advantage of more spells known, access to new schools for spells, and all of the base class features. To cry you didn't also get a new spell slot too seems a bit off to me.

[Snowbluff]

If it was rounded more aggressively, EK at 5 would be pretty crazy as a multiclass. 2 CL and a feat and extra attack and action surge and a fighting style...

2 CL at four is a full blown half caster dip. That's the ASI and a spell level boost, both at once.

[Malifice]

@malifice

Everything you put was true, but I think you are missing the comparison he was trying to make between an EK5/Wiz1 and an EK6.

As you point out the EK5/Wiz1 stays with the same spell casting power, but if the PC had instead of dipping taken another level of EK he would have 2/4 slots. So by taking the full caster dip he ends up with less spell casting than taking the 1/3 caster level.

No, he wouldnt have. He'd still have 3 slots per long rest as a 6th level EK.

An EK 5/ Wizard 1 has 3 slots a long rest. The exact same number of slots an EK 6 would have.

[djreynolds]

Just grab 3 level increments if possible.

It tough to go from level 1 EK and AT. I had 12th level character who ended up 7th EK and 5th AT... I really wanted war magic (pre scag) I leveled up each equal, until I was 6ek/5at.

It's doable.

[Willie the Duck]

This feels like an oversight. Rounding down makes sense for half casters as it follows the same pattern as their single class spell slot progression. But third casters round up. It actively discourages multiclassing third casters with other spellcasters.

Does anybody houserule this? Would it be OP to do so? Have I overlooked some important balance reason why this would be the case?

I do not feel that it is an oversight. Not, mind you, that the designers have been at all forthcoming with regards to questions of 'why' (and if they ever start being so, I have a lot of questions regarding what they envisioned for one-handed quarterstaves :smalltongue:). Still, I strongly suspect that it might have been some knee-jerk fear of multiclassing, and the potential abuses that it might bring. Same reason why a AT who multiclasses into fighter doesn't get heavy armor proficiency or an EK who multiclasses into rogue gets one less skill--the designers thought that this would be why someone might do casual dips..as opposed to the cleric dips and cha-based character super-builds and what ended up being the Multiclassing most-talked-abouts.

I think there might have been some issue about simplicity-to-learn as well. Remember that this edition is supposed to be like the redbox editions--if a bunch of 10-year-olds get the books and try to learn to play without adult supervision, they should at least approach/approximate the same game as everyone else is playing. If they put in some formula like "(1/6)*[6*(brd,clr,drd,sor,wiz)+3*(rgr,pal)+2*(ftr{EK}, rog{AT},mnk{4e})]," they'd have the same hatchet-job-mess my childhood group had trying to parse 1e AD&D's initiative. This is simple, straight-forward, and you have to find really specific examples (AT5/EK5/Pal3/Rgr3, or the like) for it to be much more onerous than the already slow progression of partial casters.

As to houserule? Yes, of course! If you the player want to try something like this, and this bit is what's standing in your way, make your case! In general, my attitude is that the rules designers are supposed to be cautious, but as a DM I have the power of veto if something turns abusive, so I don't have to be as cautious the rest of the time.

[OverLordOcelot]

@malifice

Everything you put was true, but I think you are missing the comparison he was trying to make between an EK5/Wiz1 and an EK6.

As you point out the EK5/Wiz1 stays with the same spell casting power, but if the PC had instead of dipping taken another level of EK he would have 2/4 slots. So by taking the full caster dip he ends up with less spell casting than taking the 1/3 caster level

Where are you getting your numbers from? EK6 has 3/0 from the class-specific table, while EK5/Wiz 1 has 3/0 from the multiclass table. At 7th, EK 7 and EK6/Wiz1 both go up to 4/2 slots. He won't have any 2nd level spells until he gets to Wiz3 or EK 7, but I wouldn't call that 'less spell casting'.

[Asmerv]

Where are you getting your numbers from? EK6 has 3/0 from the class-specific table, while EK5/Wiz 1 has 3/0 from the multiclass table. At 7th, EK 7 and EK6/Wiz1 both go up to 4/2 slots. He won't have any 2nd level spells until he gets to Wiz3 or EK 7, but I wouldn't call that 'less spell casting'.

This is still silly though, in my opinion.

Lets say we have two EKs, A and B, both at level 5.
A decides to stay the course, trains both with sword and spells in his/her downtime, keeps taking EK levels.
B decides to give up all martial training and focus purely on his spellcasting, and will multiclass wizard.

A and B are both Caster Level 2 at 5.
A takes a level in EK, B takes a level in Wizard. Even though B dropped all martial training and focused purely on spellcasting, they are both still Caster Level 2.
A takes another level in EK, B continues to focus purely on spellcasting and takes another level in Wizard. Even after two levels of Wizard, they are still the same Caster Level, as A is now Caster Level 3 and B is also Caster Level 3.

It takes three whole levels of Wizard to start pulling ahead. I find that a bit nonsensical.

[OverLordOcelot]

Lets say we have two EKs, A and B, both at level 5.. Even after two levels of Wizard, they are still the same Caster Level, as A is now Caster Level 3 and B is also Caster Level 3.

It takes three whole levels of Wizard to start pulling ahead. I find that a bit nonsensical.

The wizard jumps ahead in overall spell casting ability immediately, more than doubling the number of cantrips, gaining ritual casting, and more than doubling the number of spells known with no limits on which school they can be. If you're that worried about caster level specifically, then always take EK levels in multiples of 3, don't take just enough EK to get the extra attack feature. I recommend not focusing narrowly on one specific mechanical feature, you're almost always going to find things to be annoyed at if you do.

[OverLordOcelot]

I do not feel that it is an oversight. Not, mind you, that the designers have been at all forthcoming with regards to questions of 'why' (and if they ever start being so, I have a lot of questions regarding what they envisioned for one-handed quarterstaves :smalltongue:). Still, I strongly suspect that it might have been some knee-jerk fear of multiclassing, and the potential abuses that it might bring.

If you look at rogue and fighter abilities per level, I think it's pretty obvious why they did the caster breakpoints as 'only at 3 levels' and not at 2 levels or with some fractional formula. Each of the 3n-1 levels gives a big ability that's strong enough to be a "I will go to here and no more", so tying spell improvement to the level after each super good level means that there's always a reason to take one more level of this class instead of a level of multiclass. If you gave the spell level improvement at the same time as you get uncanny dodge or go up to 2x or 3x attacks, it produces a really strong incentive to split AT or EK at 2/5/8/11 and avoid taking the 'bad' levels at 3/6/9/12.

2: Cunning Action/Action Surge
5: Uncanny Dodge/Extra Attack
8: ASI/ASI
11: Reliable Talent/Extra Extra Attack
14: Blindsense/ASI
17: Spell Thief/Action Surgex2

[Malifice]

This is still silly though, in my opinion.

Lets say we have two EKs, A and B, both at level 5.
A decides to stay the course, trains both with sword and spells in his/her downtime, keeps taking EK levels.
B decides to give up all martial training and focus purely on his spellcasting, and will multiclass wizard.

A and B are both Caster Level 2 at 5.
A takes a level in EK, B takes a level in Wizard. Even though B dropped all martial training and focused purely on spellcasting, they are both still Caster Level 2. .

Yeah, but B has far more spells to chose from with those slots (EK spells known, plus those from Wizard), and a much wider variety to choose from (the entire Wizard list, and not just limited to Evocation and Abjuration any more), in addtion to gaining several more cantrips.

He is (by any metric you choose) a much better caster than A, who instead took another level of EK.

[BarneyBent]

Yeah, but B has far more spells to chose from with those slots (EK spells known, plus those from Wizard), and a much wider variety to choose from (the entire Wizard list, and not just limited to Evocation and Abjuration any more), in addtion to gaining several more cantrips.

He is (by any metric you choose) a much better caster than A, who instead took another level of EK.

By that logic, Cleric1/Wizard 19 is a better caster than a Wizard 20. A Cleric 1/Druid 1/Wizard 18 is even better. The best caster in the game would be a Cleric 1/Druid 1/Bard 1/Wizard 17. Level 20 caster with cantrips and spells of a Cleric 1, Druid 1 and Bard 1 on top.

Point is, multiclassing full-casters doesn’t have a spellcasting progression penalty, even though you get all the same benefits as you describe. So why does multiclassing third casters (and half, to a lesser extent) get this penalty? You already lose out on a bunch of proficiencies when multiclassing to keep things balanced. Just seems an extra nerf on top of all the other ones that only applies to part-casters, and punishes those with the fewest spell slots the most.

Can you work around it based on your build? Sure. But it’s silly having to. It doesn’t add any more complexity, you just have to stick with the class an extra 2 levels to receive the same benefits you’d get from the class as a single class.

If you could point to equivalent nerfs that apply to multiclassing martial builds only, or full casters, then sure. But a nerf that unfairly targets just two subclasses seems rough, and also impacts on two other classes, while not affecting any other class, seems unbalanced to me.

Incidentally, I do think the approach of adding all ET and AT subclasses together and then dividing by 3 and rounding down is more balanced than dividing each by 3 on their own. But that only works specifically when using AT x/ET x. When multiclassing with other spellcasters, it should be round to nearest whole number. Still a minor nerf but prevents shenanigans like an AT 4/ ET 4 being a 4th level spellcaster.

[bid]

By that logic, Cleric1/Wizard 19 is a better caster than a Wizard 20. A Cleric 1/Druid 1/Wizard 18 is even better. The best caster in the game would be a Cleric 1/Druid 1/Bard 1/Wizard 17. Level 20 caster with cantrips and spells of a Cleric 1, Druid 1 and Bard 1 on top.
Slippery slope.

Trying to understand is not the same as arguing to win.

[BarneyBent]

Slippery slope.

Trying to understand is not the same as arguing to win.

I’m just making the point that extra spells and cantrips known is a benefit that ALL multiclasses into a spellcasting class (minus cantrips for half casters) get. So to say that it makes up for a nerf to the spellcasting progression of third casters doesn’t hold.

The argument that it’s ok for them to not progress spellcasting level despite taking a level in a full caster because they get everything else any other class would gain by taking the same multiclass is silly. It’s like saying “Hey, you don’t get A, but you can’t complain because you get B”, when everybody else gets both A and B.

[Asmerv]

Yeah, but B has far more spells to chose from with those slots (EK spells known, plus those from Wizard), and a much wider variety to choose from (the entire Wizard list, and not just limited to Evocation and Abjuration any more), in addtion to gaining several more cantrips.

He is (by any metric you choose) a much better caster than A, who instead took another level of EK.

You're a more versatile caster, sure. But you are not a much better caster, especially not by any metric I choose as I already chose caster level as my metric and have shown that despite investing two levels into a full-caster vs. a 1/3 caster you get the same increase in spellcasting *power*.

Simply rounding up also causes funny behavior though. A 4/4 EK/AT ends up at caster level 4 if rounded up individually.

I feel like the best way to go about this is to sum all 1/3 caster class levels, divide the sum and round up, and do the same for 1/2 classes rather than divide and round individually.

[Tanarii]

Point is, multiclassing full-casters doesn’t have a spellcasting progression penalty, even though you get all the same benefits as you describe. So why does multiclassing third casters (and half, to a lesser extent) get this penalty? You already lose out on a bunch of proficiencies when multiclassing to keep things balanced. Just seems an extra nerf on top of all the other ones that only applies to part-casters, and punishes those with the fewest spell slots the most.
You're asking the question the wrong way around. The question is: why do single class 1/3 casters (and to a lesser extent 1/2 casters) get a buff to their spell casting ability? Why do they get slot and Max spell level prepared/known ahead of where they should?

It's not like any of them need that kind of buff.

[Willie the Duck]

If you look at rogue and fighter abilities per level, I think it's pretty obvious why they did the caster breakpoints as 'only at 3 levels' and not at 2 levels or with some fractional formula. Each of the 3n-1 levels gives a big ability that's strong enough to be a "I will go to here and no more", so tying spell improvement to the level after each super good level means that there's always a reason to take one more level of this class instead of a level of multiclass. If you gave the spell level improvement at the same time as you get uncanny dodge or go up to 2x or 3x attacks, it produces a really strong incentive to split AT or EK at 2/5/8/11 and avoid taking the 'bad' levels at 3/6/9/12.

We're pretty much in agreement except that I'll stick with my using the wording of 'I suspect' instead of your 'obvious why they did,' because -- while what you have stated are all true things that are good reasons for doing what was done -- we don't really know that that was the designers' reasons, and they are (frustratingly) silent on those sorts of questions.


Can you work around it based on your build? Sure. But it’s silly having to. It doesn’t add any more complexity, you just have to stick with the class an extra 2 levels to receive the same benefits you’d get from the class as a single class.

They could certainly have found another way to do these things. Restructured what you got at what level to avoid the issue OverLordOcelot and I are talking about, make things straight-up additive, figure out a way to make a 'adding fractions' formula comprehensible to the lowest reading-level audience for which the books are intended, etc. It could have been done.

The only I'd say in contrast to that is that it's hardly the only place that multiclassing has issues. Simply put, there are winners and loser in the multiclass system, as it was implemented. I wouldn't even put an EK/AT as one of the top contenders for getting the shaft. That honor goes to non-rogue-martial5/other-non-rogue-martial5 (fighter5/paladin5, barbarian5/ranger5, etc.) -- that great big honking class feature that you stuck with each class for five whole levels to gain does not synergize with that other great big honking class feature that you stuck with each class for five whole levels to gain. Beyond that, Cleric1/Wizard 19 is usually better than cleric19/wizard1 (and both knock cleric10/wizard10 out of the water), paladin MCs better with sorcerer or warlock than with the (usually) more-thematically-appropriate fighter or cleric, all Cha-based classes interlock together better than most anyone else, and taking a 1-level dip in druid hamstrings you more than a 1-level monk dip (the monk feature might only turn on if you are unarmored, but the druid one actively constrains your armor choices*).
*DM ruling dependent, as is the whole of druid armor restrictions.

Simply put, the designers could have put more effort into the design of the multiclass system to make it more universally equitable, but they didn't. I won't call it a bad decision, as I think anything they did come up with could well have caused as many complaint-worthy issues as what we ended up with has done, but still it is a notable weak spot in the edition.

[BarneyBent]

We're pretty much in agreement except that I'll stick with my using the wording of 'I suspect' instead of your 'obvious why they did,' because -- while what you have stated are all true things that are good reasons for doing what was done -- we don't really know that that was the designers' reasons, and they are (frustratingly) silent on those sorts of questions.

I see this argument, but that same argument applies to a single-classed AT/EK. I don’t get how applying the nerf when multiclassing solves any balance issues that didn’t already exist. Keep in mind that when rounding up, 3n+1 is the spell slot breakpoint.

For an AT you are talking about ASI, Evasion, ASI, Versatile Trickster, ASI, ASI. All pretty great. For an EK you’re talking about ASI, War Magic (basically the subclass’s signature ability), Eldritch Strike (also subclass’s signature ability), Indomitable x2, ASI, ASI.

Having spell slot breakpoints align with other breakpoints doesn’t unbalance things in single class. Why multiclassing?



The only I'd say in contrast to that is that it's hardly the only place that multiclassing has issues. Simply put, there are winners and loser in the multiclass system, as it was implemented. I wouldn't even put an EK/AT as one of the top contenders for getting the shaft. That honor goes to non-rogue-martial5/other-non-rogue-martial5 (fighter5/paladin5, barbarian5/ranger5, etc.) -- that great big honking class feature that you stuck with each class for five whole levels to gain does not synergize with that other great big honking class feature that you stuck with each class for five whole levels to gain. Beyond that, Cleric1/Wizard 19 is usually better than cleric19/wizard1 (and both knock cleric10/wizard10 out of the water), paladin MCs better with sorcerer or warlock than with the (usually) more-thematically-appropriate fighter or cleric, all Cha-based classes interlock together better than most anyone else, and taking a 1-level dip in druid hamstrings you more than a 1-level monk dip (the monk feature might only turn on if you are unarmored, but the druid one actively constrains your armor choices*).
*DM ruling dependent, as is the whole of druid armor restrictions.

Simply put, the designers could have put more effort into the design of the multiclass system to make it more universally equitable, but they didn't. I won't call it a bad decision, as I think anything they did come up with could well have caused as many complaint-worthy issues as what we ended up with has done, but still it is a notable weak spot in the edition.

Sure, but those examples you gave are due to the nature of those classes as single classes. Some classes are more heavily front loaded than others. Some have mechanics early on that go well slapped on another class. These multiclasses being better or worse isn’t because the game actively takes away things from the classes (barring some first level proficiencies).

That’s the issue I have. It’s that a deliberate nerf has been applied to these classes and these classes only for no good reason other than maybe the designers were lazy. And I don’t think that’s a very good excuse.

The solution needn’t be complicated. “For half and third casters, add levels of same progression together and divide the total by 2 for half and 3 for third-casters. For half-casters, round down. For third casters, round to the nearest whole number. Then add the results to your overall spellcasting level”.

Done.