I was running a game the other day and one of my players was thinking about what subclass they wanted their paladin to take. They were sticking to the PHB, which is fine, but doesn’t have many options; only three subclasses are available. They wanted to play a worshipper of Selune, a moon god, because they liked stars as a quirk. Great! They’ll choose Devotion or Ancients and be Selune. They agreed, but they found a homebrew Oath of the Moon paladin subclass on reddit that was connected to Selune. She sent me a link and I looked it over, read one line, and rejected it outright. Why? Why would I be so brash as to immediately dismiss a subclass so early? Well, it was simple and immediate.

At 3rd level, when taking the Oath of the Moon, the Paladin learns sleep.

‘What? That’s not that bad’, you may be thinking. ‘Sleep’s a first level spell and it’s not a broken spell or anything.’ Well, it is a good spell, but you wouldn’t be completely off the mark with thinking that a first level spell isn’t wild for a paladin to have.

Ask yourself, what classes learn sleep outside of class features? Well, that’d be wizards and bards, the most spell-reliant classes. They don’t get proficiencies in armor, weapons, or melee attacks, just strong spells and utility, like sleep. In case you’re not aware, sleep has a chance to make a target unconscious if their hit points are low enough. This can end fights quickly with weaker monsters and be a great way to finish off something that’s got a high AC that the martial classes can’t get through. It doesn’t do damage, and it’s conditional on the enemy’s hit points. Why shouldn’t paladins get sleep? Well, the paladin is not a battlefield manipulator. They don’t get spells that slow enemies or have huge debuffs or move enemies around, they have divine smite and martial weapons and heavy armor. Smites do high, single target damage to an enemy.

Another quirk of D&D is that an unconscious enemy is automatically score a critical hit when a melee attack is made within 5 feet of it, a coup de gras type feature. A critical hit doubles all the damage dice that the attack would normally do, so 1d8 becomes 2d8. This accounts for additional damage, most notably a rogue’s sneak attack, which can deal massive damage on critical hits under certain circumstances. Divine Smite deals 2d8 radiant normally on top of a melee attack, so a first level critical hit smite with just a long sword deals 2d6+4d8+modifiers. This can be combined with certain spells that smite, held actions, and other brutal ways to destroy enemies. Important to note: paladins, unlike the rogue’s sneak attack, can smite as much as they want with as many attacks as they have. A sleeping creature can take two attacks while still being unconscious, so the numbers become very big

Picture this: A Nycoloth comes in front of the party, a powerful CR 9 creature with a fly speed, tons of immunities, and non-magical resistance (and magical resistance). After one turn, they can do, when they hit, 4d12+10 (36) slashing magical damage. This guy is strong and was designed for four level 9 characters. The paladin steps up and blows a third-level spell slot, dealing 4d8 radiant damage and a resisted 1d12+7 slashing with their great ax. That is too much martial damage for a primary spell caster because it’s not, it’s for a fighter with a d10 hit die and a possible AC of 16-20. Sleep itself takes 5d8, up to 40, and if that roll is higher than the creature’s hit points, they become unconscious. That means after 3 turns, a level 5 paladin can beat a CR 9 Nycaloth if they roll high and use sleep after. That is wild. It makes the paladin able to end fights entirely on their own. Do not do this, unless there is special reason why. When you give spells and features to classes that don’t normally have, know what that means for these classes, and what snaps, and whether you want it to snap.

I’ve fallen into this trap; giving a Ranger turn undead might be thematically appropriate, but damn is that powerful for a martial class. There is no main problem with giving a Sorcerer subclass access to the Barbarian’s rage feature but think about combinations of what the Sorcerer gains from this and the feature given. This is true even with novel features, you have to constantly ask yourself What does this do to the class? I recently made a martial Sorcerer subclass (no rages) but I gave the Sorcerer unarmored defense, proficiency in martial weapons, and a weaker form of stunning strike. These all meshed with my idea of the character, blending Kensei Monk and a class that had no martial subclasses. I liked this a lot, but when it came to a final ability to give them the ability to give all martial weapons the finesse property. I scrapped this idea because while cool and thematic, it breaks the way equipment works. The obvious best weapon for this feature is the Warhammer, a weapon that has the versatile property. It does massive bludgeoning damage but is balanced to be a strength weapon. You need to invest in strength, a relatively limited stat outside of combat. Paladins, fighters, barbarians, clerics, all martial classes who use strength as a primary stat become hogwash when it comes to weapons if you can just be a spellcasting Sorcerer who can still cast spells by switching to one-handed wielding with the Warhammer. This was replaced with just a basic clash of blades, increasing AC with a reaction and granting an opportunity attack. This is within the rules of the game and doesn’t snap any mechanics unintentionally. Do not allow the rules of play to be broken unless you are absolutely sure it doesn’t turn into a first-order optimal strategy.

An addendum that has to be stated: there is a type of paladin that can learn sleep – Oath of Redemption. They can learn sleep as a class spell at 3rd level, but this is a very special case of the world and the mechanics overlapping, because Oath of Redemption Paladins have sworn an oath to uphold honor, fight non-lethally, and attempt to solve conflict with peaceful resolution. While a Redemption Paladin can use their sleep for a nasty 4d6+4d8+4d6+6d8+3 radiant damage, they would be going against the character that would have taken that oath. Without this in-world restriction, like a paladin of Selune, it becomes a first-order optimal strategy to sleep a creature at the highest possible level and destroy them before they can begin to dream.

A first-order optimal strategy is the best way to do something in a game that is the simplest and most direct way to solve a problem in a game. Button mashing, a powerful combo, spamming, all of these are this kind of strategy and if all combats can be solved with 2 smites and a sleep you have broken the game. If the optimal weapon is not subjective for a feature, you have broken the game. Breaking the game can be fun, but in homebrew, you are interested in releasing and want to be part of a balanced game, think carefully and avoid breaking the game by using these abilities.