D&D 5E Savage Attacker - what does it do?

[clearstream]

APOLOGIES - I was using numbers for my tweaked version without announcing that. Here is that broken out.

Savage Attacker was taken two times out of ~240 ASI choices in a survey of ~27 groups. I think part of that is because... wtf does it do? Using the power of Excel and a simple Monte Carlo simulation, here is an estimate.

The feat allows a character to reroll their damage dice once per turn and use whichever they like. That's easy to estimate when you're rolling say 1d6 (it adds about a point) but is more difficult to assess when multiple dice are involved. By looking at an entire adventuring day (30 rounds) and the number of attacks that could be landing, it can be estimated by actually rolling (two times) and taking the best roll... hundreds of times via the spreadsheet.

Using 11th level characters with 20 on their relevant ability, I found that it added ~1-2 damage per turn, or ~50/day*. Compared with GWM, which adds ~400/day for fighters, it seems worth about one-tenth of an ASI. And because it is not obvious what it does, it represents a trap feat for new players. Hence I recommend the following feat revision -

Duelist (Defensive Duelist merges with Savage Attacker)
Prerequisite: Strength or Dexterity 13 or higher
When you are wielding a one-handed melee weapon with which you are proficient and another creature hits you with a melee attack, you can use your reaction to add your proficiency bonus to your AC for that attack, potentially causing the attack to miss you.
Once per turn, when you roll damage for a melee weapon attack, you can reroll the damage dice and use either total.

The goal here is to make a feat that helps fill a gap in the system - which is feat support for one-handed melee. Switching to reroll the damage instead of only the weapon scales it with the number of dice thrown, so that with Sneak Attack it adds ~2-3/turn or 60-90/day. Still pretty underwhelming, but not practically nothing (as the feat is now). It becomes worth about one-sixth of an ASI. Merging it with Defensive Duelist improves both feats. I don't believe this is the destination point for these feats, but only a stepping stone toward a feat that supports 1H-melee properly. Also, slightly, gives a reason to use a Versatile weapon.


*I'm estimating over an average adventuring day of 2 combats, short rest, 2 combats, short rest, 2 combats; with combats being on average 5 rounds. The average add depends on weapon.

 

[jaelis]

As a formula, a reroll of a single dX die is worth (X^2-1)/6X points of damage. So that is best for a d12 where it is worth just less than +2 damage. (It is +1.986 to be exact.) I know 2d6 is worse, I might be able to get a formula for that too.

So assume a character with a battle axe. +2 damage for a feat isn't totally pathetic, though it seems on the weak side. The fact that it only works once per turn is certainly a limit, though if you get multiple attacks you do gain some advantage through the choice of when to apply it. It would be interesting to look at that too.

---
Added: the average damage for roll 2d6 twice, keep the best, is 5425/648, or about 8.37 hp. So that is a damage increase of +889/648 = 1.37 hp, which is pretty close to the 1.31 hp you get with a d8 weapon. So using SA with a greatsword is about as good as using it with a longsword, and considerably worse than using a greataxe.

--
I think a better fix would be to just have it work every time you roll damage. The fix you proposed is a bit awkward because the reroll benefit is much better for a 2H weapon, while the duelist benefits require a 1H weapon.

 

[CapnZapp]

Your comparison to GWM is at first blush very apples to oranges.

Savage Attacker only works once per turn. Thus your example assumes twenty combat rounds.

Your GWM figure, OTOH, doesn't say how many attacks, only the per day figure. Also, it likely doesn't take advantage or to-hit bonuses into account, which is where it's true strength lie.

(By itself GWM appears much more reasonable than in practical play)

I mean, if you contrast the 3-5 figure for Savage Attacker with the +40 figure actual play shows GWM to yield during nova rounds, who cares that the difference might be smaller during non-critical "easy" fights where the GWM fighter doesn't make an effort?

All these comparisons need to focus on spotlight or they lack real play relevance. What matters for a feat is its ability to steal the spotlight for your character, not some theoretical averages.

A more fruitful comparison would be SA vs +2 Strength/Dex. This shows the feat to be a clear failure, only enticing players without a graph of the math.

My solution is to replace the reroll with extra dice. That is, with a Greatsword (the maximum) you gain +7 dmg per round regardless of your level or stats (unless you miss all your attacks, which pretty much only happens on low levels, where it is okay for the feat to be slightly less powerful)

Easy to analyse, and no longer a newbie trap.

Sent from my C6603 using EN World mobile app

 

[clearstream]

Your comparison to GWM is at first blush very apples to oranges.

Savage Attacker only works once per turn. Thus your example assumes twenty combat rounds.

Yes, I used it once per turn. The way I did the estimate is I looked at the number of attacks that hit over the 30 round day, and then I used Savage Attacker (once per turn) on the dice connected with that attack. I used randbetween(1,N) to create two damage sets for each attack that hits, and then used max(range) to get the highest number and min(range) to get the lowest number. I subtracted min from max and summed the differences to get the per day value... but now that I think about it, I believe I should have halved the per day value to represent the probability that when you rolled once, you rolled the high value.

That's interesting, because I was close to already saying that Savage Attacker should drop the "Once per turn" language. I'll improve the model and come back to this estimate, with a look at dropping "Once per turn".

Your GWM figure, OTOH, doesn't say how many attacks, only the per day figure. Also, it likely doesn't take advantage or to-hit bonuses into account, which is where it's true strength lie.

It takes fiat advantage and to-hit bonuses into account. Advantage is extremely powerful with GWM, but if it's not fiat it's basically up to a DM to allow it. (Fiat means something like Shield Master or Feint.)

All these comparisons need to focus on spotlight or they lack real play relevance. What matters for a feat is its ability to steal the spotlight for your character, not some theoretical averages.

A more fruitful comparison would be SA vs +2 Strength/Dex. This shows the feat to be a clear failure, only enticing players without a graph of the math.

Yes, I do exactly such comparisons. You saw here I said it looks no better than half an ASI right? And buffed it by joining it to another (underpowered) feat.

My solution is to replace the reroll with extra dice. That is, with a Greatsword (the maximum) you gain +7 dmg per round regardless of your level or stats (unless you miss all your attacks, which pretty much only happens on low levels, where it is okay for the feat to be slightly less powerful)

Easy to analyse, and no longer a newbie trap.

What I like about Savage Attacker (in principle) is that it does more, the more dice you roll with one attack. Adding dice makes it good for any attack, which I dislike. What I'm after is something that works very nicely for single/few-attack classes, is a fair benefit for multiple-attack classes, and plays well with Sneak Attack. In short, a dueling benefit that works really well for dueling, but not so well for fighting styles that are already supported.

Adding a die (or two) to an attack per turn is meh: you can do it already with Hunter's Mark and Hex, GFB and Booming. I feel we can do better than that...

 

[jaelis]

What I like about Savage Attacker (in principle) is that it does more, the more dice you roll with one attack.

What do you mean here? You only get to reroll the weapon dice, which are pretty much fixed. The most dice you can reroll is 2d6, and that is less benefit than rerolling 1d12. Is there some way to get more weapon dice on an attack that I'm not thinking of?

It sounds like maybe you want to use it with sneak attack... that doesn't work
https://www.sageadvice.eu/2016/01/2...r-feat-apply-to-sneak-attack-or-divine-smite/
Or is this something you are proposing to change?

 

[clearstream]

What do you mean here? You only get to reroll the weapon dice, which are pretty much fixed. The most dice you can reroll is 2d6, and that is less benefit than rerolling 1d12. Is there some way to get more weapon dice on an attack that I'm not thinking of?

It sounds like maybe you want to use it with sneak attack... that doesn't work
https://www.sageadvice.eu/2016/01/2...r-feat-apply-to-sneak-attack-or-divine-smite/
Or is this something you are proposing to change?

You possibly didn't notice my tweak, in the revised Duelist feat? Up-thread.

 

[jaelis]

So that is a pretty major change, I might perhaps have highlighted it?

I think that it might be too powerful now for paladin smites, they are already pretty effective.

---
I guess not, it still only does like +3 or 4 points. The benefit of the reroll really peters off as you add more dice.

 

[CapnZapp]

Yes, I used it once per turn. The way I did the estimate is I looked at the number of attacks that hit over the 30 round day, and then I used Savage Attacker (once per turn) on the dice connected with that attack. I used randbetween(1,N) to create two damage sets for each attack that hits, and then used max(range) to get the highest number and min(range) to get the lowest number. I subtracted min from max and summed the differences to get the per day value... but now that I think about it, I believe I should have halved the per day value to represent the probability that when you rolled once, you rolled the high value.

Not sure I follow you. I have no insight into your calculations, I did not question your spreadsheets.

What I was talking about is instead: who agreed the 30 round day was the interesting number?

If one feat does +X damage per attack over a 30 round day, and another feat does +2X they look quite comparable (in the sense that the first feat is "half the second").

But if the first feat can't ever do more than +X damage per attack, while the second can be used to do +5X damage over, say, ten rounds, that's a much more useful statistic to me (with my minmaxer hat on).

That tells me the second feat is MUCH better at "hogging the spotlight" as it were - becoming the MVP of the combat of the day that really counts.

It's possible there exists gamers (and minmaxers) that value being able to consistently outdamage another build. But I'm not one of them.

What I'm looking for in a feat (to deem it "OP") is the ability to completely overshadow other builds, to the degree that those other builds feel like trap builds. Not on average. Not throughout the day.

After all, after making the other players look like they were sitting on their hands for That One Combat, chances are good you don't even have to be manipulative to get to a long rest. They usually are well in sight after The One Combats.

That's interesting, because I was close to already saying that Savage Attacker should drop the "Once per turn" language. I'll improve the model and come back to this estimate, with a look at dropping "Once per turn".

Not sure what's so interesting, but hey, if I can help I'm happy... to have helped

It takes fiat advantage and to-hit bonuses into account. Advantage is extremely powerful with GWM, but if it's not fiat it's basically up to a DM to allow it. (Fiat means something like Shield Master or Feint.)

Do you mean "by an ability or power given by the rules" when you say Fiat?

Not sure why you need to bring that up? Let me assure you I'm not talking about advantage given through Inspiration or houserules here - only cold hard crunch interests the cold hard optimizer...

What I like about Savage Attacker (in principle) is that it does more, the more dice you roll with one attack.

Sure, but I happen to put a higher negative value on the diminishing returns than you, so I dislike rolling all those dice for essentially no payback.

Adding dice makes it good for any attack, which I dislike. What I'm after is something that works very nicely for single/few-attack classes, is a fair benefit for multiple-attack classes, and plays well with Sneak Attack. In short, a dueling benefit that works really well for dueling, but not so well for fighting styles that are already supported.

I can respect that.

My Savage Attack does provide exactly the same benefit no matter the number of attacks: up to +7 damage. You might not recollect this, but I offer it as a replacement for GWM. In that light, it's most attractive feature is the hard cap on that number, as opposed to GWM which gave +10 per attack.

Adding something for the duelist I applaud. Adding something for the (melee) rogue I heartily encourage. OTOH, I never considered "Savage Attack" to be the feat for them, though.

Adding a die (or two) to an attack per turn is meh: you can do it already with Hunter's Mark and Hex, GFB and Booming. I feel we can do better than that...

Sure.

Remember, I'm not criticizing your feat per-se. I just got hung up on your comparisons, and how they became the basis of justification for your changes. In short: in my opinion the original SA sucks male donkey reproductive glands and GWM (also book) outperforms it by a factor far greater than merely two.

 

[Stalker0]

What do you mean here? You only get to reroll the weapon dice, which are pretty much fixed. The most dice you can reroll is 2d6, and that is less benefit than rerolling 1d12. Is there some way to get more weapon dice on an attack that I'm not thinking of?

It sounds like maybe you want to use it with sneak attack... that doesn't work
https://www.sageadvice.eu/2016/01/2...r-feat-apply-to-sneak-attack-or-divine-smite/
Or is this something you are proposing to change?

Does it work with the cold damage of a frost brand or the radiant damage of a holy avenger?

 

[clearstream]

Not sure I follow you. I have no insight into your calculations, I did not question your spreadsheets.

What I was talking about is instead: who agreed the 30 round day was the interesting number?

First I must apologise for not making it clear at the outset I was discussing a modified version. I have corrected that. You are exactly right, unmodified the feat is hopeless: it's a trap feat.

The reason for 30 rounds is that the average 5 round combat, twice between short rests, with two short rests before a long rest, gives us a day of 30 rounds. That is the refresh cycle for short and long rest abilities, so it allows us to properly expend those over our estimate. Also, I'm finding that in giving chunkier numbers it matches better my requirements for good estimating (as a project manager, I do a lot of estimating). What I need is information that clearly separates scenarios, without pretending to unachievable (or over effortful) levels of precision.

Additionally, a broader estimate will capture more situations and more tables, instead of putting the spotlight on occasional situations at some tables.

If one feat does +X damage per attack over a 30 round day, and another feat does +2X they look quite comparable (in the sense that the first feat is "half the second").

But if the first feat can't ever do more than +X damage per attack, while the second can be used to do +5X damage over, say, ten rounds, that's a much more useful statistic to me (with my minmaxer hat on).

That tells me the second feat is MUCH better at "hogging the spotlight" as it were - becoming the MVP of the combat of the day that really counts.

In balancing the game, I'm fine if minmaxers can find ways to hog the spotlight. In fact, we want characters to find ways to shine! What I am pursuing is a better game overall. Right now, we have a hole where feat support for 1H melee should be (sorry TWF I'm ignoring you for now, be patient).

What I'm looking for in a feat (to deem it "OP") is the ability to completely overshadow other builds, to the degree that those other builds feel like trap builds. Not on average. Not throughout the day.

After all, after making the other players look like they were sitting on their hands for That One Combat, chances are good you don't even have to be manipulative to get to a long rest. They usually are well in sight after The One Combats.

I don't agree with your assumptions about how the game is widely played. And would reiterate that my goal isn't to stop some feats being better than others, but to bring different strategies into viability for a broad range of D&D groups. Optimisers are just one possible "client" of this design work. They have plenty to play with already.

Do you mean "by an ability or power given by the rules" when you say Fiat?

Not sure why you need to bring that up? Let me assure you I'm not talking about advantage given through Inspiration or houserules here - only cold hard crunch interests the cold hard optimizer...

I mean that in my estimates a character only gets consistent advantage worth counting, if they can make that advantage happen by using the rules. Shield Masters for instance, can generate advantage at least 2/3rds of the time so in my estimates I give them that. Beast Master as another example can use their Hawk to Help nearly all the time, so I give them that. If someones says - "Oh, but my character has advantage" - then they need to explain how it does that consistently enough to include in estimates. If their answer is - "I'm a special snowflake and my DM lets me" - then I'm going to ignore them because I'm thinking about all the other times and tables where that doesn't happen.

And in the end, I have to address the design work to the published rules. When people aren't using those rules, I don't know much about what can happen in their games so I can't address their scenarios.

My Savage Attack does provide exactly the same benefit no matter the number of attacks: up to +7 damage. You might not recollect this, but I offer it as a replacement for GWM. In that light, it's most attractive feature is the hard cap on that number, as opposed to GWM which gave +10 per attack.

Adding something for the duelist I applaud. Adding something for the (melee) rogue I heartily encourage. OTOH, I never considered "Savage Attack" to be the feat for them, though.

Remember, I'm not criticizing your feat per-se. I just got hung up on your comparisons, and how they became the basis of justification for your changes. In short: in my opinion the original SA sucks male donkey reproductive glands and GWM (also book) outperforms it by a factor far greater than merely two.

You may well be right! As you know, I'm trying to take a conservative approach - minimising word changes and limiting additions. (So far, I have included no outright additions in my list of revised feats, and I think only one or two major wording alterations. Lots of single word changes, and slight tweaks of course.)

For example, I noticed we can double what the feat does for many characters by just making it reroll all the damage dice rather than just the weapon. A small change for a decent effect.