Making Savage Attacker a viable Feature

[El_Jairo]

Hey DnD enthousiasts!

I'm currently playing a Half-Orc Barbarian and I find it troubling to see how useless a Great-axe seems, compared to 2d6 weapons.

So I looked over at the Savage Attacker feat, yet I realized that this feature is great before lvl 5 but starts to taper off from there.

Let me remind you what the feature, vanilla does: "Once per turn, when you roll for damage on a melee weapon attack, you can re-roll the weapon damage dice and pick either result."

Which seems underwhelming but being able to re-roll when it matters the most can mean the difference between a kill or not a kill, so it is important for me. Yet it feels more like a half-feature.

So here is the suggested changes I'm going to play-test:

Use it during the whole round
Grant an extra attack when the RR rolls max damage on the dice (or 10+ damage)
ASI for Strength or Dexterity *(maybe, I'm not sure on this one yet)


So the first is a minor tweak that would allow you to RR also on reaction attacks. I did this more for dramatic effect that you can really try to put a foe down that provokes an AoO

The second part is what I feel is where the meat of the change should be. My first thought was when your roll max damage on the dice you immediately make an extra MWA (Melee Weapon Attack) with the same weapon (I would allow another target to be picked). Yet I am worried that if you use low damage die weapons, you can get like 16,66% chance on a bonus attack on a d6 weapon. But I felt that this is offset because those weapons deal less damage. Yet still, if you can add an effect to each attack, this can be exploited.

So I was thinking about rolling 10+ on damage dice, which is only possible for d10, d12 and 2d6 weapons. Or does this feels to restrictive to Savage Attacker only being viable for Heavy Weapons and making the Greataxe have a 25% chance to generate a bonus attack seems way overpowered.
The other thing I am worried about is that together with GWM (Great Weapon Master) it would lead to, too many attacks, so the extra attack should be as a bonus action or a bonus reaction.

The last part (ASI) was just to make more on par with those über feats like GWM or Sharpshooter and to make this feature appeal to more different builds.

You CC is very welcome!

[Man_Over_Game]

A few things to mention:

Savage Attacker is designed to only work off of the weapon's damage dice
Savage Attacker doesn't explicitly state that it requires a melee weapon, only a melee weapon attack (which means it doesn't inherently require a melee weapon, I.E. A Sword).

Why is this important? Because the Wild Shape forms of Moon Druids can deal some pretty hefty damage (5d10 unmodified) on some of their conditional attacks, which can dramatically increase the value of Savage Attacker. Making it better without considering those outliers can lead to some pretty unbalancing changes.

[Brutalitops]

A few things to mention:

Savage Attacker is designed to only work off of the weapon's damage dice
Savage Attacker doesn't explicitly state that it requires a melee weapon, only a melee weapon attack (which means it doesn't inherently require a melee weapon, I.E. A Sword).

Why is this important? Because the Wild Shape forms of Moon Druids can deal some pretty hefty damage (5d10 unmodified) on some of their conditional attacks, which can dramatically increase the value of Savage Attacker. Making it better without considering those outliers can lead to some pretty unbalancing changes.


That actually is kind of cool and a druid taking the feat savage attacker feels very flavourful and cool.

[El_Jairo]

Why is this important? Because the Wild Shape forms of Moon Druids can deal some pretty hefty damage (5d10 unmodified) on some of their conditional attacks, which can dramatically increase the value of Savage Attacker. Making it better without considering those outliers can lead to some pretty unbalancing changes.

Yet this isn't the best use of a re-roll. As the feature states that you can re-roll all weapon dice and pick which result you prefer.
With 5d10 the variance will be lower because you are throwing 5 dice. Sure they have a bigger variance than the standard d6 but still the biggest relative impact will be on a single die roll with high variance. Thus on a d12.

But sure, Savage Attacker will be a decent way to make sure your Wild Shape Moon Druid won't wiff the damage on one of it's attacks per turn.
Which is cool and all, yet it doesn't come close to adding 10 to the damage of each attack you are able to make on your turn.

So yeah, the point in granting a extra attack when you roll 10+ damage is not viable. Yet if I would change it to max result on the weapon damage roll, it doesn't really accomplish that much for high damage weapons.
So just a naked roll to determine when you get and extra attack? It doesn't feel fluffy but statistically makes more sense.

What would you suggest? 10% Chance is too much? I want to stay away from the GWM triggers to grant a bonus attack: rolling a critical hit or reducing an adversary to 0 HP. So technically, I don't really seem to have an obvious candidate. So I will opt to simply roll a d10 and on a 10 you have hit your opponent so savagely, he is reeling back and offering you an opportunity to make another attack with the same weapon.

[Man_Over_Game]

What would you suggest? 10% Chance is too much? I want to stay away from the GWM triggers to grant a bonus attack: rolling a critical hit or reducing an adversary to 0 HP. So technically, I don't really seem to have an obvious candidate. So I will opt to simply roll a d10 and on a 10 you have hit your opponent so savagely, he is reeling back and offering you an opportunity to make another attack with the same weapon.

Could just do something like this:

When a damage die for a melee weapon attack rolls its highest possible value, you can do extra damage equal to half of that amount with that attack (so a 1d12 adds +6 damage). This can only happen once per round.

[Vogie]

You could make it something like "Your weapon damage dice roll can't be lower than your Con Modifier". For example, a d8 weapon with a +4 Con would basically give you a damage die of 4, 4, 4, 4, 5, 6, 7, 8.


If you want to keep it, but tack on a second half of the feat, some other ideas to include could be:

If you roll the same number on both damage die, the attack becomes a critical hit.
If you roll the same number on both damage die, your blows will impact a different creature of your choice within 5 feet of the target. Make a new attack and damage roll.
If you roll the same number on both damage die, the creature can't take reactions until the start of your next turn.
As a bonus action, you can focus a creature you can see within 30 feet of it and make that creature your focus. The target remains your focus for 1 minute, or until either you or your focus drops to 0 hit points. When you make an attack roll against your focus, add a d4 to its damage roll.
Whenever you crit or roll max damage on that target, all further attacks to that target have a +1 damage bonus (stacks up to +5)

[El_Jairo]

Okay, so here is my *former take on the Savage Attacker feature:

Once per round when you roll damage for a melee weapon attack, you can re-roll the weapons damage dice and choose either result.


Increase your Strength or Dexterity score by 1, to a maximum of 20.
When re-rolling the damage dice and if you roll the maximum possible on the dice (and at least 8) on the re-roll, you can make one immediate melee weapon attack as a bonus action on your turn. If this happens outside of your turn, you can make a bonus Attack of Opportunity as part of the reaction on which you chose to re-roll the weapon damage dice.

[Theodoric]

Hey DnD enthousiasts!

I'm currently playing a Half-Orc Barbarian and I find it troubling to see how useless a Great-axe seems, compared to 2d6 weapons.
Well, as a Half-Orc Barbarian you get a lot more use from Savage Attacks racial feature, and your Brutal Critical class feature, by using a greataxe over a maul or greatsword. That, and since you don't have access to the Great Weapon Fighting fighting style (which works better with 2d6 weapons), the opportunity cost is much lower.

You basically have the build for greataxe users already.

[djreynolds]

Is there any way you could write down?

Pre damage and post damage?

I handed the feat out for free on occasion just to see in actual action. It's not bad.

Writing down the actual numbers might shed a brighter light on the feat

[El_Jairo]

Could just do something like this:

When a damage die for a melee weapon attack rolls its highest possible value, you can do extra damage equal to half of that amount with that attack (so a 1d12 adds +6 damage). This can only happen once per round.

I like this because it streamlines the action and doesn't allow for abuse of generating more attacks.


You could make it something like "Your weapon damage dice roll can't be lower than your Con Modifier". For example, a d8 weapon with a +4 Con would basically give you a damage die of 4, 4, 4, 4, 5, 6, 7, 8.


If you want to keep it, but tack on a second half of the feat, some other ideas to include could be:

If you roll the same number on both damage die, the attack becomes a critical hit.
If you roll the same number on both damage die, your blows will impact a different creature of your choice within 5 feet of the target. Make a new attack and damage roll.
If you roll the same number on both damage die, the creature can't take reactions until the start of your next turn.
As a bonus action, you can focus a creature you can see within 30 feet of it and make that creature your focus. The target remains your focus for 1 minute, or until either you or your focus drops to 0 hit points. When you make an attack roll against your focus, add a d4 to its damage roll.
Whenever you crit or roll max damage on that target, all further attacks to that target have a +1 damage bonus (stacks up to +5)

I don't get why CON modifier should matter to a Savage Attacker, fluff-wise I would think of raw strength or dexterous strikes that find the crucial spots. Also this would become broken if u apply this to two weapon fighting, as you can get rid of 1-3 results on the d8, which is like a flat +2 damage output to each attack. Which is the same as the barbarian Rage bonus up to lvl 8.

The condition of rolling the same number on the re-roll is just the same chance as rolling max result but it is inspiring to use.

The focus suggestion is going to be abused in a two-weapon build. Adding a d4 is a lot if you are dishing out multiple attacks, it seems. But I like the reliable boost it gives.

The part of stacking up damage bonus does seem promising, but it could be cumbersome to keep track of.
Does this bonus only apply for you or do allies also benefit from it? Like a debuff kind of thing? I does seem a little situational as when you crit you might outright kill the target and then there is no benefit to be gained...

So your remarks have inspired me to come up with this variant:

Once per round when you roll damage for a melee weapon attack, you can re-roll the weapons damage dice and choose either result.

When re-rolling the damage dice and if you roll the equal or lower to the initial roll: add half of the weapon dice to the weapon damage.

For example, if your weapon damage dice are d12, 2d6 or 3d4, you will add 6 damage.

I like this one maybe the best, as it provides the weapon damage with a guaranteed boost. Giving you the option to mini-nova, once per round. It rewards using Great-weapons as their boost will be more significant. It stays relevant, even if you already rolled high damage, the chances go up, to get a bigger boost out of it.

Is this too much? Maybe just applying the bonus, if you roll lower than the initial damage. Yet then you have a chance that this feature doesn't do anything for you. And with 5e I feel like features need to be powerful enough to be worth taking over a ASI.

[Bloodcloud]

I'd do:

once per turn, you can either:
Re-roll and choose either
OR
Reroll any number of dice from one attack and keep the new roll

Also, triggers on crit automatically.

Because not having it for your crit is just too sad.

[Solid_Snek]

I think Savage Attacker should do what it says plus give a bonus action to attempt to shove a creature back or to the side 5' after using Savage Attacker.

[Skylivedk]

A few things to mention:

Savage Attacker is designed to only work off of the weapon's damage dice
Savage Attacker doesn't explicitly state that it requires a melee weapon, only a melee weapon attack (which means it doesn't inherently require a melee weapon, I.E. A Sword).

Why is this important? Because the Wild Shape forms of Moon Druids can deal some pretty hefty damage (5d10 unmodified) on some of their conditional attacks, which can dramatically increase the value of Savage Attacker. Making it better without considering those outliers can lead to some pretty unbalancing changes.


Is there any way you could write down?

Pre damage and post damage?

I handed the feat out for free on occasion just to see in actual action. It's not bad.

Writing down the actual numbers might shed a brighter light on the feat

It's a horrible feat as it is. Even on a moon druid. Reading this,
https://rpg.stackexchange.com/questions/47172/how-much-damage-does-great-weapon-fighting-add-on-average

You'd think, ok 4 damage, that's... Less than I expected, but lo and behold, it's even less:
https://www.reddit.com/r/DnD/comments/8bl495/when_is_the_savage_attacker_feat_worth_it/?utm_source=reddit-android

So without bothering to do the math in detail: somewhere between 1.5 and 3. Horrible.

https://www.reddit.com/r/dndnext/comments/530395/how_useful_of_a_feat_is_savage_attacker

A proper trap feat. I like the version where you add 50% on top of a max roll, kind of a mini crit. I don't know if it is enough to save this option from looking, smelling and feeling like a trap. If the math doesn't add up tag +1 Str or Dex on top.

[Vogie]

I don't get why CON modifier should matter to a Savage Attacker, fluff-wise I would think of raw strength or dexterous strikes that find the crucial spots. Also this would become broken if u apply this to two weapon fighting, as you can get rid of 1-3 results on the d8, which is like a flat +2 damage output to each attack. Which is the same as the barbarian Rage bonus up to lvl 8.


I was trying not to double dip... although, on reflection, you could use the "Other Martial modifier" (Dex for a Strength attack, or Strength for a dex attack), your proficiency bonus, or tie it to something like Wisdom (Savage Shillelagh?). And yes, you could do something silly like having a +5 to that mod (or +6, if you are using prof), then use Daggers that will always roll a 5 or 6 on a d4, in each hand. lol.

[El_Jairo]

Well, as a Half-Orc Barbarian you get a lot more use from Savage Attacks racial feature, and your Brutal Critical class feature, by using a greataxe over a maul or greatsword. That, and since you don't have access to the Great Weapon Fighting fighting style (which works better with 2d6 weapons), the opportunity cost is much lower.

You basically have the build for greataxe users already.
Sure, but banking on the fact that 5% of the time your critical damage will be higher in exchange for more variance on your damage output that 2d6 weapon? Even the average of the Greataxe is 0.5 lower that the Greatsword.
It seems like a gamble that will pay off only a once per 20 attacks, which is maybe every three to four encounters?


Is there any way you could write down?

Pre damage and post damage?

I handed the feat out for free on occasion just to see in actual action. It's not bad.

Writing down the actual numbers might shed a brighter light on the feat
I did a full statistical analyses and did make some actual rolls and wrote it down.
You don't have to compare a feature like savage attacker in a vacuum but compare it to the other option.

I compared the bonus for using a Great-axe (=GA) with Savage Attacker (SA) with Using a Great-sword (=GS) and picking +2 STR (=2S) instead, here are the results for one attack.
Average damage: GA SA d12 = 8,48611 and GS 2S = 8,0 Delta= 0,48611
Min. Expected dmg: GA SA d12 = 7 and GS 2S = 5 Delta = 2
Median dmg: GA SA d12 = 12 and GS 2S = 8 Delta = 4
Max. Expected dmg: GA SA d12 = 12 and GS 2S =11 Delta = 1

Expected damage is where 75+% of the results lie in.
So for one attack Savage attacker is better than ASI, but I didn't figure in the 5% extra to hit chance, so you could say that +2 STR does 5% more damage than above.

Writing it down does show how much you can gain, which isn't too bad but is only comparing the actual rolls and not the opportunity cost.

I played with the feature last session and we typically don't get a ton of attacks, and once I rolled exactly the same result and once it netted me about +2 dmg.


I'd do:

once per turn, you can either:
Re-roll and choose either
OR
Reroll any number of dice from one attack and keep the new roll

Also, triggers on crit automatically.

Because not having it for your crit is just too sad.
This seems nice.
When you say re-roll any number of dice from one attack, do you also mean the attack roll?

Having it always on on crits is a nice feature indeed.


I think Savage Attacker should do what it says plus give a bonus action to attempt to shove a creature back or to the side 5' after using Savage Attacker.
This would seem not bad indeed: by being so agressive it is very fluffy and it gives the feature more impact on actual tactical combat.

Do you need to re-roll in order to make the shove attack bonus action? I do like that it's a bonus action, so it is competing with other sources of bonus action.

These are some nice ideas, many thanks!

[djreynolds]

Why not steal part of GWM? And tweak it.

Bonus action attack if your reroll kills an opponent.

[Shuruke]

Not sure if this helps at all but it couples well with the great weapon style for fighter etc.

Currently I have a great club fighter 1d8
Two handed
Not heavy so I don't use gwm

I took savage attacker cuz theatrically it fit.

I re roll my two attacks on 1 or 2s on one of these attacks if I get a 1 or 2 on the re roll I use savage attacker.

No clue how this affects the average but I only very rarely actually have to take anything less than a 4 in dice for damage.

Thinking average of 4.5
25% chance of a 1 or 2 on a d8
Re roll that 25% with only a 25% of another 1 or 2

(Probably doing this wrong) but that makes it so. 25% chance re roll
Increases average by 1.78 for 6.28
With another re roll whenever you want once per turn which u can choose to do if u get a 3 or etc to try to get better so roll a 3 re roll have a 25% chance to re roll for better again so odds are really in your favor to keep your dice roll above 4.

Idk how this would work with say a great axe. But I've noticed it doing a lot of work in past 4 sessions. Alot of time if I get a 4 on damage on second attack ill use it just to see if I get higher and generally it works out I've only had to deal 1+5 once and 3+5 twice havent had to take a two.

Makes great club feel fun to constantly be dealing 10-13 damage

(Dm had us roll stats and got a 18 so got to 20 quick )

[Solid_Snek]

This would seem not bad indeed: by being so agressive it is very fluffy and it gives the feature more impact on actual tactical combat.

Do you need to re-roll in order to make the shove attack bonus action? I do like that it's a bonus action, so it is competing with other sources of bonus action.

These are some nice ideas, many thanks!

You may use the bonus action after the re-roll damage. So if you use it on your first attack, you can then use the bonus action. If you use it on the last attack (due to extra attack), you can use the bonus action then.

This way you have a trigger that is specific like other feats.

[Rukelnikov]

Just make all weapon dice explode and call it a day.

[Shuruke]

Hey since this is on savage attacker feat

The way it is worded it says damage dice rather than singular.

Does this mean
Great sword 2d6 u can re roll both

Great sword crit 4d6 you re roll all 4

Great axe half orc barbarian 9 crit with orcish fury feat expending orcish fury dice
6d12 re roll all of them?

It kinda reads as if its like advantage for damage
Grab your dice roll them again take either total

If That's case I don't think it needs a buff tbh

[CTurbo]

Here is how I rule Savage Attacker at my table

When you hit with a weapon attack, you can roll the weapon damage twice and choose which score to take.

It's basically all the time advantage on damage rolls with no attack limit. It does not work on additional damage dice like smite or sneak attack. Also, with a maul or greatsword, you can't roll 4d6 and choose the best 2, you roll 2d6 twice and choose the best set of two.

It is strong, but I have not found it to be OP.

In past editions, I have used what I call a magic Savage Axe that does this.

[El_Jairo]

It's a horrible feat as it is. Even on a moon druid.
...Edit Cut...
A proper trap feat. I like the version where you add 50% on top of a max roll, kind of a mini crit. I don't know if it is enough to save this option from looking, smelling and feeling like a trap. If the math doesn't add up tag +1 Str or Dex on top.
I agree that the statistics on RAW Savage Attacker (SA) feature don't proof to have that umppf your looking for when you hear Savage Attacker!

So I like feats to have you actively interact with the Roleplay by shouting: "I savagely hack into that troll!" and get to make a special roll that could help tip the balance in a fight.

So I went the following SA redux:

+ Increase STR or DEX +1, to max 20.

+ Once per round when you make a mwa, you can reroll the weapon dice and pick the highest score.
Or
If the reroll was equal to or lower, you may add half max roll to the weapon damage total if you pick the reroll result.
(For example 6 if weapon dice are 1d12, 2d6 or 3d4. 5 For 1d10 or 2d5, etc.)

I like the feat because it has the 'always on' 1 ASI.
And the moment you pop the feat on a turn you have greater chances of increasing your damage than before.
Even if you roll high on the initial roll, you might still get the +half dice damage to give a small boost.

What do you think?
Is it too powerful?

I am inclined to believe it would average somewhere around average +5 damage on one attack per round. Nothing game breaking.
I might do the maths later on statistics.
But I will play-test it tomorrow. So I'll let you know what the first alfa testing says.

[JoeJ]

Hey since this is on savage attacker feat

The way it is worded it says damage dice rather than singular.

Does this mean
Great sword 2d6 u can re roll both

Great sword crit 4d6 you re roll all 4

Great axe half orc barbarian 9 crit with orcish fury feat expending orcish fury dice
6d12 re roll all of them?

It kinda reads as if its like advantage for damage
Grab your dice roll them again take either total

If That's case I don't think it needs a buff tbh

Yes, it does say the weapon's damage dice, however many that is. Also it says "reroll" meaning that you've already rolled, so you can use this after you see what the roll was. Finally, it says once per turn, so if you have extra attacks you can only use it on one, but you can use it on your turn and then again if you attack on somebody else's turn (an opportunity attack, for example).

[Man_Over_Game]

I'm not sure what the effectiveness of your example was, but I can confirm that my example (rolling max damage on a damage die deals 50% more damage die) basically equates to +0.5 average damage. This isn't 100% accurate, as you're tacking on MORE damage to what your already maxed damage, so overkill is going to happen quite often (which makes sense for a Savage Attacker)

In comparison, Dual Wielder:
Grants +2 - +4 damage (varying based on the number of Extra Attacks)
Grants +1 AC
Is conditional

So considering Dual Wielder increases damage by the same/more than my proposed version of SA, I think that just dealing double the weapon damage would be a fair fit. It sounds extreme, but it really just balances out to +1 damage on average with a chance to nova. I'd finish it by making it either apply to all of your attacks and making it a half feat.

I'm not saying that my recommendation is better, but it is a bit easier to theorycraft its balance since it's a bit more forward in its numbers.


---------

For some basic calculations:

1d4 has a 25% chance to max out, deals +2 damage, on average deals 2x0.25= 0.5 damage
1d12 has an 8.3% chance to max out, deals +6 damage, on average deals 6x0.083= 0.5 damage

Doubling the bonus damage dealt will result in doubling the average damage per attack.

[Bloodcloud]

Thinking about it, the overkill damage optional rule from the dmg seems like another good add-on.

bonus action trip attack could also be nice.

[El_Jairo]

Why not steal part of GWM? And tweak it.

Bonus action attack if your reroll kills an opponent.
I was thinking along those lines, but I want to keep every feature unique.
I borrowed the gain bonus action attack after dropping a minion and I wasn't convinced by it, because it's conditional in that you need minions to profit from it. Your example is more conditional so it would have a lower impact.
Yet I also want to build this Half-Orc Barbarian into a crit-fisher with GWM, so I don't want to step into GWM territory.

Thank you for your input, djreynolds.


You may use the bonus action after the re-roll damage. So if you use it on your first attack, you can then use the bonus action. If you use it on the last attack (due to extra attack), you can use the bonus action then.

This way you have a trigger that is specific like other feats.
Okay, that seems rather easy to implement and it is easy to use.
Yet it feels a little too tactical for what I imagine the feat to do: just increase your damage output. I do like the idea that your opponent get's shoved because of the impact of your attack.
I might add the following for when you use the Re-roll and use to pick the highest result. On a failed STR test (DC = 8 + PROF + CON mod), the target of your attack is knocked back 5 feet.
Knocking an opponent prone, seems too valuable as it grants tactical advantage to other allies attacking that creature.

Thank you for your input, Solid_Snek.


Just make all weapon dice explode and call it a day.
I do like the simplicity of this proposal, yet it can easily be exploited by using multiple attacks with a low weapon damage, like d4 or d6. So it compensates the fact that you have lower average damage, but this should be compensated by having multiple attacks.
I am rather looking for this feature to make Great-axes and big weapons more viable by giving them more reliability in delivering the damage.

Thank you for your input, Rukelnikov.


Here is how I rule Savage Attacker at my table

When you hit with a weapon attack, you can roll the weapon damage twice and choose which score to take.

It's basically all the time advantage on damage rolls with no attack limit. It does not work on additional damage dice like smite or sneak attack. Also, with a maul or greatsword, you can't roll 4d6 and choose the best 2, you roll 2d6 twice and choose the best set of two.

It is strong, but I have not found it to be OP.

In past editions, I have used what I call a magic Savage Axe that does this.
I like the simplicity of your solutions and I really think that it works very well. Because the damage output of each mwa you make is significantly increased. It scales well as your gain more attack per round which you can make.

What I don't like about this part is that it's always on in combat and you don't feel as much when the feature helped you out. So I am now trying to make it still a decision you have to make, because I like discrete decisions because it makes you feel like your actions matter more.

Thank you for your input, CTurbo.


I'm not sure what the effectiveness of your example was, but I can confirm that my example (rolling max damage on a damage die deals 50% more damage die) basically equates to +0.5 average damage. This isn't 100% accurate, as you're tacking on MORE damage to what your already maxed damage, so overkill is going to happen quite often (which makes sense for a Savage Attacker)

In comparison, Dual Wielder:
Grants +2 - +4 damage (varying based on the number of Extra Attacks)
Grants +1 AC
Is conditional

So considering Dual Wielder increases damage by the same/more than my proposed version of SA, I think that just dealing double the weapon damage would be a fair fit. It sounds extreme, but it really just balances out to +1 damage on average with a chance to nova. I'd finish it by making it either apply to all of your attacks and making it a half feat.

I'm not saying that my recommendation is better, but it is a bit easier to theorycraft its balance since it's a bit more forward in its numbers.


---------

For some basic calculations:

1d4 has a 25% chance to max out, deals +2 damage, on average deals 2x0.25= 0.5 damage
1d12 has an 8.3% chance to max out, deals +6 damage, on average deals 6x0.083= 0.5 damage

Doubling the bonus damage dealt will result in doubling the average damage per attack.
So I am reading it correctly that you would only grant the extra damage when the damage dice roll max?
Or can you still reroll the weapon damage dice once per turn?
Because rerolling damage dice has already an expected boost of 2-4 damage to a normal damage roll.

I want a feature to have an impact, if it only would impact when you already rolled well, this is a win-more feature that has possibly no real impact if it only creates overkill damage.

Thank you for your input, Man_Over_Game.

[Man_Over_Game]

So I am reading it correctly that you would only grant the extra damage when the damage dice roll max?
Or can you still reroll the weapon damage dice once per turn?
Because rerolling damage dice has already an expected boost of 2-4 damage to a normal damage roll.

I want a feature to have an impact, if it only would impact when you already rolled well, this is a win-more feature that has possibly no real impact if it only creates overkill damage.

Thank you for your input, Man_Over_Game.

Whichever fits balance, honestly.

If players are able to do this every attack, all the time, it probably shouldn't give more than +2-3 damage. If the player can only do this once per round, we can afford to make it a special attack that effectively adds +4-5 damage.

If we were to go with my example and having a condition that requires a lot of overkill damage, that actually allows us MORE of a budget due to the fact that, in a sizeable majority of rolls, the bonus damage doesn't matter.

In other words, a +2 consistent damage bonus or a +5 damage bonus that activates on max damage results in having a similar bonus, due to the fact that you only need a small amount of damage (+2, for example) to finish a creature that you've already rolled max damage on.

Make it big and powerful, and then make it throw away a lot of the bonuses. It's still rather thematic, (hitting a goblin and dealing 18 damage in a single blow and cleaving him in two) and all it really ends up doing is giving you +2 damage to mooks and some bonus damage against creatures that can take a lot of hits.

But I can see why you might want to go in a different direction. Some people are disappointed by wasted rolls, and this would end up creating a lot of waste.


--------

While I like the idea mentioned to have the Bonus Action attack if the creature dies, I feel like GWM and Savage Attacker are things that really should work together, and them having a redundant ability would contradict that.

[El_Jairo]

Here is the final version I want to test tonight:

Once per round when you roll damage for a melee weapon attack, you can re-roll the weapons damage dice and choose either result.


Increase your Strength or Dexterity score by 1, to a maximum of 20.
When re-rolling the damage dice and and if this roll is equal to or lower than the initial roll, you may add half of max. damage dice to the rerolled result and pick that sum. (For example you may add 6 damage on a reroll <= initial result for weapon dice: d12, 2d6 or 3d4; 5 damage for d10 or 2d5, etc.)
On the first critical hit you roll on a round, you may use the re-roll ability for free, not the additional damage part.


I added the part of increasing efficiency of critical hits. Yet I don't want to go overboard on this, because I'm building a crit-fisher with this feature, together with GWM. It might turn out OP if both features reward criticals too much.

[El_Jairo]

So I had my first session with the modified feature.
It turned out pretty well.

First round I score a critical but the reroll netted same roll 15 (on 3d12). On the same turn I used the roll to reroll 8 damage into 7, so adding effectively 5 damage. Which made it feel like scoring a critical (20 dmg) and a small critical (18 dmg).

The rest of the turns, where amongst the same line: being able to boost one attack damage with +5 is pretty nice and it gives a lot more punchy feel to my barbarian.
Yet there is still the possibility to roll worse than the first roll but chances are about 50% lower than with the original SA feature.

[Tanarii]

Savage Attacker ("fixed"):
Before you make a melee attack with a heavy weapon that you are proficient with, you can choose to take a -5 penalty to the attack roll. If the attack hits, you add +10 to the attack's damage.

Then eliminate that line from Great Weapon Master.

"Fixes" two Feats with one stone. Each is now worth a standalone full Feat and properly balanced.

Next up, PAM and SS "fixed".

[Mjolnirbear]

I honestly feel that Savage Attacker must have originally had a crit function that Wizards forgot to replace when they decided to get rid of crit-fishing abilities.

I'd be trying this: keep the original, and add "once per rest, on a melee weapon attack that hits, you can declare that the attack is a critical hit."

Assuming three rests in a day, three crits is pretty awesome, feeling-wise.

However, I cannot math. My gut says it's not that much damage on average, until I consider Smites, Sneak Attacks, Booming Blade, and Assassinate. These make me much less sure.

[Potato_Priest]

Savage Attacker ("fixed"):
Before you make a melee attack with a heavy weapon that you are proficient with, you can choose to take a -5 penalty to the attack roll. If the attack hits, you add +10 to the attack's damage.

Then eliminate that line from Great Weapon Master.

"Fixes" two Feats with one stone. Each is now worth a standalone full Feat and properly balanced.

Next up, PAM and SS "fixed".

I would never take great weapon master if it only gave me a bonus action attack on killing an enemy or critical hitting, but I can at least imagine situations where I would take savage attacker. I do not think this new version of the great weapon master feat is balanced. Perhaps instead of moving things the way you suggested, you should take the second benefit of GWM and move it to Savage attacker, leaving the -5/+10 on GWM.

[CTurbo]

I wouldn't mind getting the math on the way I do it.

When you hit with a weapon attack, you can roll the weapon damage twice and choose which score to take.

I think it's pretty powerful but that's ok

[Tanarii]

I would never take great weapon master if it only gave me a bonus action attack on killing an enemy or critical hitting, but I can at least imagine situations where I would take savage attacker. I do not think this new version of the great weapon master feat is balanced. Depends on class I supposed. I think of GWM as the Barbarian Feat, and Barbarians at least one crit every 1 in 4 rounds after level 5. And tend to drop foes on many rounds too IMX, although that often overlaps with crits of course.

But your alternative suggestion is definitely better than mine. Moving the crit/drop bonus attack to Savage attacker is probably still a balanced full feat. Which says a lot about Savage Attacker.

[Rebonack]

I altered this feat at my table as follows.

When you deal damage roll twice and pick the better result. If you roll minimum damage, the damage is instead maximized.

So basically it is a feat for reducing variance on large dice weapons. It's still worse than GWM by a longshot, but it isn't a trap anymore.

[Zalabim]

In case it wasn't clear yet, the feat already allows you to use it once per TURN, not once per round, and not only on your turn, so it is also usable on reaction attacks that take place on other turns.

So the most boring fix to its power is going to be adding +1 to a stat.

An exciting boost to power, without getting into too weird of calculations, is ending the feat with "If either result is the maximum for its dice, you may use both." For an example, the normal average of 1d12 is 6.5. With the normal reroll and use highest, it's ~8.5. With reroll, use highest, both if either is 12, the average is ~9.65. On the other end, 1d4 averages 2.5, with reroll averages ~3.125, adding both if either is 4 averages 4.125. So it's kinda explode-y dice but still favors using the larger d12s.
reading key
Initial roll -> reroll results
|
v
average result
_____
overall average

1 1 2 3 4 5 6 7 8 9 10 11 13
6.5833
2 2 2 3 4 5 6 7 8 9 10 11 14
6.75
3 3 3 3 4 5 6 7 8 9 10 11 15
7
4 4 4 4 4 5 6 7 8 9 10 11 16
7.333
5 5 5 5 5 5 6 7 8 9 10 11 17
7.75
6 6 6 6 6 6 6 7 8 9 10 11 18
8.25
7 7 7 7 7 7 7 7 8 9 10 11 19
8.8333
8 8 8 8 8 8 8 8 8 9 10 11 20
9.5
9 9 9 9 9 9 9 9 9 9 10 11 21
12.25
10 10 10 10 10 10 10 10 10 10 10 11 22
11.0833
11 11 11 11 11 11 11 11 11 11 11 11 23
12
12 13 14 15 16 17 18 19 20 21 22 23 24
18.5
______
9.65


1 1 2 3 5
2.75
2 2 2 3 6
3.25
3 3 3 3 7
4
4 5 6 7 8
6.5
____
4.125

It's a horrible feat as it is. Even on a moon druid. Reading this,
https://rpg.stackexchange.com/questions/47172/how-much-damage-does-great-weapon-fighting-add-on-average
Not about savage attacker.


You'd think, ok 4 damage, that's... Less than I expected, but lo and behold, it's even less:
https://www.reddit.com/r/DnD/comments/8bl495/when_is_the_savage_attacker_feat_worth_it/?utm_source=reddit-android

So without bothering to do the math in detail: somewhere between 1.5 and 3. Horrible.
Adam-M did the math wrong by not accounting for the option to use either result. For example, the average damage of 2d12 drop lowest is ~8.5, not ~8.

https://www.reddit.com/r/dndnext/comments/530395/how_useful_of_a_feat_is_savage_attacker[/url]

A proper trap feat. I like the version where you add 50% on top of a max roll, kind of a mini crit. I don't know if it is enough to save this option from looking, smelling and feeling like a trap. If the math doesn't add up tag +1 Str or Dex on top.
Once again, Quadratic- does the math wrong, but this time G3nji_17 corrects some of it. Gets ignored. The overall conclusion of it being worse than +2 strength or dex for combat (due to lacking the bonus to accuracy) still holds, but the difference isn't as large as these answers make it out to be. The feat isn't a trap, just something that should only be considered when increasing strength or dexterity is no longer an option. Like Dual Wielder.


Depends on class I supposed. I think of GWM as the Barbarian Feat
Philosophically, I don't generally think of feats as being a barbarian thing, but particularly when it comes to weapon specialization feats I think of those as a fighter thing due to the combination of extra ASI and the theme of weapon specialization. Anyone can be a weapon specialist, that's why they're feats and not class features, but that means making your character more fighter-y, to me. On top of that, the power attack abilities favor hitting with more attacks, which means hitting more, and attacking more both improve the use of the feat. Fighters attack more than anyone else, and while barbarians get advantage to hit more easily, getting advantage to hit at all is not unique to barbarians. The cleave ability favors the one with the better single attacks, and that would be barbarian over fighter as long as availability of the bonus action is equal between them.

Tl:Dr version: I completely disagree with your division of GWM.

[Skylivedk]

An exciting boost to power, without getting into too weird of calculations, is ending the feat with "If either result is the maximum for its dice, you may use both." For an example, the normal average of 1d12 is 6.5. With the normal reroll and use highest, it's ~8.5. With reroll, use highest, both if either is 12, the average is ~9.65. On the other end, 1d4 averages 2.5, with reroll averages ~3.125, adding both if either is 4 averages 4.125. So it's kinda explode-y dice but still favors using the larger d12s.
reading key
Initial roll -> reroll results
|
v
average result
_____
overall average

1 1 2 3 4 5 6 7 8 9 10 11 13
6.5833
2 2 2 3 4 5 6 7 8 9 10 11 14
6.75
3 3 3 3 4 5 6 7 8 9 10 11 15
7
4 4 4 4 4 5 6 7 8 9 10 11 16
7.333
5 5 5 5 5 5 6 7 8 9 10 11 17
7.75
6 6 6 6 6 6 6 7 8 9 10 11 18
8.25
7 7 7 7 7 7 7 7 8 9 10 11 19
8.8333
8 8 8 8 8 8 8 8 8 9 10 11 20
9.5
9 9 9 9 9 9 9 9 9 9 10 11 21
12.25
10 10 10 10 10 10 10 10 10 10 10 11 22
11.0833
11 11 11 11 11 11 11 11 11 11 11 11 23
12
12 13 14 15 16 17 18 19 20 21 22 23 24
18.5
______
9.65


1 1 2 3 5
2.75
2 2 2 3 6
3.25
3 3 3 3 7
4
4 5 6 7 8
6.5
____
4.125

Not about savage attacker.


Adam-M did the math wrong by not accounting for the option to use either result. For example, the average damage of 2d12 drop lowest is ~8.5, not ~8.

Once again, Quadratic- does the math wrong, but this time G3nji_17 corrects some of it. Gets ignored. The overall conclusion of it being worse than +2 strength or dex for combat (due to lacking the bonus to accuracy) still holds, but the difference isn't as large as these answers make it out to be. The feat isn't a trap, just something that should only be considered when increasing strength or dexterity is no longer an option. Like Dual Wielder.


Thank you for doing the extra work and research. Are you referring to the following quote from Quadratic?

"Reckless Attack actually makes it WORSE, because it raises up your accuracy and means that you're much more likely to hit an enemy twice, and there's no extra benefit from that second hit.

Best case scenario and you're rolling that super big crit of 5d8 dice, your average damage from your weapon goes from 22.5 to...25.4. With 20 strength and your rage damage, that's instead 31.5 to 34.4.

In other words, Savage Attacker makes the first critical hit you perform on a turn 9.2% more powerful.

And that's the BEST case scenario. Trust me, it's a terrible feat and you'd be better off with just about anything else."

Well, he also seems to think it's terrible. Then again, if the comparison is to Dual Wielder, then we're also in Trap land. And I'm not talking a late comer to HipHop, but a hidden door underneath your feet with an automatic opener that hides a pit full of spikes smeared in goblin poo. I like your exploding version and the way it favours big weapons. Would you only let it work on a 12 for a greatsword? Because that might just be a great indirect buff for greataxes as well (1/12 chance rather than 1/36).

[Zalabim]

Thank you for doing the extra work and research. Are you referring to the following quote from Quadratic?

"Reckless Attack actually makes it WORSE, because it raises up your accuracy and means that you're much more likely to hit an enemy twice, and there's no extra benefit from that second hit.

Best case scenario and you're rolling that super big crit of 5d8 dice, your average damage from your weapon goes from 22.5 to...25.4. With 20 strength and your rage damage, that's instead 31.5 to 34.4.

In other words, Savage Attacker makes the first critical hit you perform on a turn 9.2% more powerful.

And that's the BEST case scenario. Trust me, it's a terrible feat and you'd be better off with just about anything else."

Well, he also seems to think it's terrible. Then again, if the comparison is to Dual Wielder, then we're also in Trap land. And I'm not talking a late comer to HipHop, but a hidden door underneath your feet with an automatic opener that hides a pit full of spikes smeared in goblin poo. I like your exploding version and the way it favours big weapons. Would you only let it work on a 12 for a greatsword? Because that might just be a great indirect buff for greataxes as well (1/12 chance rather than 1/36).
You can see some corrections to what Quadratic gets wrong in G3nji_17's reply. The feat is bad overall, like Dual Wielder is bad and Tough is bad. It's not bad to have the feat, there's just usually something better you could pick instead.

And yes, I would have it be the maximum value of all the dice. That also somewhat reigns in the likelihood of it triggering on a huge handful of dice from a critical hit too. If you do roll double 12 or triple 12 on a greataxe crit though, it would also likely result in just a ton of overkill. I couldn't imagine how to write the benefit to apply to individual dice within the overall result without getting extremely wordy anyway.

This is harder because the results aren't all equally likely. Good luck.
1 2 3 4 5 6 5 4 3 2 1 /36
2 2 6 12 20 30 42 40 36 30 22 14 /36
1/36 7.055
3 3 6 12 20 30 42 40 36 30 22 15 /36
2/36 7.111
4 4 8 12 20 30 42 40 36 30 22 16 /36
3/36 7.222
5 5 10 15 20 30 42 40 36 30 22 17 /36
4/36 7.41666
6 6 12 18 24 30 42 40 36 30 22 18 /36
5/36 7.7222
7 7 14 21 28 35 42 40 36 30 22 19 /36
6/36 8.1666
8 8 16 24 32 40 48 40 36 30 22 20 /36
5/36 8.777
9 9 18 27 36 45 54 45 36 30 22 21 /36
4/36 9.52777
10 10 20 30 40 50 60 50 40 30 22 22 /36
3/36 10.3888
11 11 22 33 44 55 66 55 44 33 22 23 /36
2/36 11.333
12 14 30 48 68 90 114 100 84 66 46 24 /36
1/36 19
_____
8.75152777

Asking anydice, best of two sets of 2d6 averages 8.37 and of course the normal average of 2d6 is 7.

[qube]

Just ran the math on a t-rex bite (4d12), using https://anydice.com/ to get each damage's probability, then putting up a matrix of all the probabilities (a.k.a. what is the chance your first roll is 20 and your second roll is roll is 30), and each of those values multiplied with either zero (if the new result is lower, as you'd pick the first roll), or the boost in damage (in the example 10); (and then of course divided by 10 000 as the values anydice gives are 0-100 instead of 0-1).

In excel / open Calc terms,

=(C$5*$B6/10000) * (MAX(C$4;$A6)-$A6)
(with the 4 row and A column the damages, the 5 row and B colum the chances, blue calculating the odds, red the damage boost)

summing those numbers only grants + 3.92 average damage

Also, 'normal' attacks give:

2d6: +1.37
1d12: +1.98
1d10: +1.65
1d8: +1.31
1d6: +0.97
1d4: +0.63

though, obviously, that's only on one attack - not all attacks per round.

---------------
looking at those numbers, I'd consider making savage attacker work on all attacks (though obviously still only weapon dice (not sneak attack, holy smite, manouvres, ...)

[El_Jairo]

Here is the final version I want to test tonight:

Once per round when you roll damage for a melee weapon attack, you can re-roll the weapons damage dice and choose either result.


Increase your Strength or Dexterity score by 1, to a maximum of 20.
When re-rolling the damage dice and and if this roll is equal to or lower than the initial roll, you may add half of max. damage dice to the rerolled result and pick that sum. (For example you may add 6 damage on a reroll <= initial result for weapon dice: d12, 2d6 or 3d4; 5 damage for d10 or 2d5, etc.)
On the first critical hit you roll on a round, you may use the re-roll ability for free, not the additional damage part.


I added the part of increasing efficiency of critical hits. Yet I don't want to go overboard on this, because I'm building a crit-fisher with this feature, together with GWM. It might turn out OP if both features reward criticals too much.
I presume that I can quote myself as I started this thread to receive some feedback on my version of the Savage Attacker feature.
So I wonder if I went too far with the buffing.

My arguments for this version are :
1) I wanted to have a part that is 'always on', which is the half ASI
2) more importantly, I want this feature to give you a sizeable boost when you decide to use it. Because it gives you a resource to manage and when you use it, it has a decent chance of providing a damage boost, which is still conditional but can also be applied when you already rolled high for damage. In the original feature the chance was rather low to get a boost when you already rolled low.
3) The boost to the first critical hit is there to give more reliability to the First crit you roll.
As you are rolling more dice, the potential for boosting the roll are significantly lower because of the Gauss distribution.

I would like to know if there is a character build in which you would like to use this feature.

My opinion on the math of damage increasing features is that you have to look at the expected damage and not only at the average damage. Because one die is not spread out like a Guass-curve and the average is only a representative of an infinite number of rolls.

When I am back at the pc where I made the calculations, I can add some graphics to make it more clear.

[MrStabby]

Roll the damage dice twice on any number of attacks. Select from the combined pool. Include all dice rolled in the attack (sneak attack, booming blade, smites, hunter's mark...).

Solid boost to damage and favours more dice and bigger dice.

[Zalabim]

I presume that I can quote myself as I started this thread to receive some feedback on my version of the Savage Attacker feature.
So I wonder if I went too far with the buffing.

My arguments for this version are :
1) I wanted to have a part that is 'always on', which is the half ASI
2) more importantly, I want this feature to give you a sizeable boost when you decide to use it. Because it gives you a resource to manage and when you use it, it has a decent chance of providing a damage boost, which is still conditional but can also be applied when you already rolled high for damage. In the original feature the chance was rather low to get a boost when you already rolled low.
3) The boost to the first critical hit is there to give more reliability to the First crit you roll.
As you are rolling more dice, the potential for boosting the roll are significantly lower because of the Gauss distribution.

I would like to know if there is a character build in which you would like to use this feature.

My opinion on the math of damage increasing features is that you have to look at the expected damage and not only at the average damage. Because one die is not spread out like a Guass-curve and the average is only a representative of an infinite number of rolls.

When I am back at the pc where I made the calculations, I can add some graphics to make it more clear.
Actually, I can do this. I can make a matrix for this type:
1 7 2 3 4 5 6 7 8 9 10 11 12
7
2 7 8 3 4 5 6 7 8 9 10 11 12
7.5
3 7 8 9 4 5 6 7 8 9 10 11 12
8
4 7 8 9 10 5 6 7 8 9 10 11 12
8.5
5 7 8 9 10 11 6 7 8 9 10 11 12
9
6 7 8 9 10 11 12 7 8 9 10 11 12
9.5
7 7 8 9 10 11 12 13 8 9 10 11 12
10
8 8 8 9 10 11 12 13 14 9 10 11 12
10.5833
9 9 9 9 10 11 12 13 14 15 10 11 12
11.25
10 10 10 10 10 11 12 13 14 15 16 11 12
12
11 11 11 11 11 11 12 13 14 15 16 17 12
12.83333
12 12 12 12 12 12 12 13 14 15 16 17 18
13.75
_____
9.9930555

1 3 2 3 4
3
2 3 4 3 4
3.5
3 3 4 5 4
4
4 4 4 5 6
4.75
_______
3.8125
The trend of the average benefit of using the reroll stays similar: Lower results benefit more than higher results, but it drops off a bit slower. The overall average for the d12 becomes ~9.99. For the d4, the average becomes 3.8125, so nothing too weird on the low end either. I only worry about the wording of the ability getting a bit tough for someone to parse.

Allowing rerolling more dice would help it too, but then you're also rerolling more dice. There's a cost in play speed. Rerolling all the hits instead of choosing also makes it more like a passive boost and less like a special ability you use. I think adding the +1 ASI is a good balance, and making the one reroll more powerful, more special, is the better route.

[qube]

For the d4, the average becomes 3.8125
You sure?

max(1,1) = 1
max(1,2) = 2
max(1,3) = 3
max(1,4) = 4
max(2,1) = 2
max(2,2) = 2
max(2,3) = 3
max(2,4) = 4
max(3,1) = 3
max(3,2) = 3
max(3,3) = 3
max(3,4) = 4
max(4,1) = 4
max(4,2) = 4
max(4,3) = 4
max(4,4) = 4

SUM(C1:C16)/COUNT(C1:C16) = 3.125

conform my previous calculation (which was rounded)
4d12: +3.92
2d6: +1.37
1d12: +1.98
1d10: +1.65
1d8: +1.31
1d6: +0.97
1d4: +0.63
(where one can also clearly see how the damage boost of 4d12 isn't twice that of 1d12)

Considering the baseline - primairy stat +2, results in +1 damage per attack, and 5% more hit chance per attack, that feat seems to be good.
weaponwise, the inferiour 1d12 getst the biggest boost - half a point over the 2d6 ... which is nice as it's average damage used to be half a point lower (6.5 vs 7).
the second largest powerboost, 1d10, makes me think "damn, another feat for the GWM/PAM build", thohugh that build also uses a d4, ( attacking twice with a 1d10 & once with a 1d4 only gives a damage boost of 3.95; compared ASI giving 3, and three times a slightly bigger hit chance.)

[El_Jairo]

Actually, I can do this. I can make a matrix for this type:
1] 7 2 3 4 5 6 7 8 9 10 11 12
7
2] 7 8 3 4 5 6 7 8 9 10 11 12
7.5
3] 7 8 9 4 5 6 7 8 9 10 11 12
8
4] 7 8 9 10 5 6 7 8 9 10 11 12
8.5
5] 7 8 9 10 11 6 7 8 9 10 11 12
9
6] 7 8 9 10 11 12 7 8 9 10 11 12
9.5
7] 7 8 9 10 11 12 13 8 9 10 11 12
10
8] 8 8 9 10 11 12 13 14 9 10 11 12
10.5833
9] 9 9 9 10 11 12 13 14 15 10 11 12
11.25
10] 10 10 10 10 11 12 13 14 15 16 11 12
12
11] 11 11 11 11 11 12 13 14 15 16 17 12
12.83333
12] 12 12 12 12 12 12 13 14 15 16 17 18
13.75
_____
9.9930555

1] 3 2 3 4
3
2] 3 4 3 4
3.5
3] 3 4 5 4
4
4] 4 4 5 6
4.75
_______
3.8125
The trend of the average benefit of using the reroll stays similar: Lower results benefit more than higher results, but it drops off a bit slower. The overall average for the d12 becomes ~9.99. For the d4, the average becomes 3.8125, so nothing too weird on the low end either. I only worry about the wording of the ability getting a bit tough for someone to parse.

Allowing rerolling more dice would help it too, but then you're also rerolling more dice. There's a cost in play speed. Rerolling all the hits instead of choosing also makes it more like a passive boost and less like a special ability you use. I think adding the +1 ASI is a good balance, and making the one reroll more powerful, more special, is the better route.
I will try to get the wording better, so everybody will get to understand it right away.

Many thanks for your kind words.


You sure?

max(1,1) = 1
max(1,2) = 2
max(1,3) = 3
max(1,4) = 4
max(2,1) = 2
max(2,2) = 2
max(2,3) = 3
max(2,4) = 4
max(3,1) = 3
max(3,2) = 3
max(3,3) = 3
max(3,4) = 4
max(4,1) = 4
max(4,2) = 4
max(4,3) = 4
max(4,4) = 4

SUM(C1:C16)/COUNT(C1:C16) = 3.125

conform my previous calculation (which was rounded)
4d12: +3.92
2d6: +1.37
1d12: +1.98
1d10: +1.65
1d8: +1.31
1d6: +0.97
1d4: +0.63
(where one can also clearly see how the damage boost of 4d12 isn't twice that of 1d12)

Considering the baseline - primairy stat +2, results in +1 damage per attack, and 5% more hit chance per attack, that feat seems to be good.
weaponwise, the inferiour 1d12 getst the biggest boost - half a point over the 2d6 ... which is nice as it's average damage used to be half a point lower (6.5 vs 7).
the second largest powerboost, 1d10, makes me think "damn, another feat for the GWM/PAM build", thohugh that build also uses a d4, ( attacking twice with a 1d10 & once with a 1d4 only gives a damage boost of 3.95; compared ASI giving 3, and three times a slightly bigger hit chance.)

Zalabim was calculating my variant of the Re-roll pick max. If your reroll <= initial roll, you get to add half of max of damage dice. So with 1d4 you get +2. I have added brackets in Zalabim's spoilers to make it clear that the number before the bracket is the initial roll, not part of the resulting damage.

I'll make another shot at the wording and I'll insert a graphic comparing the Savage Attacker redux to the original and to 2d6 and 2d6 with 2 ASI.

[El_Jairo]

My version of Savage Attacker (redux)


Increase your Strength or Dexterity score by 1, to a maximum of 20.
Once per round when you roll damage for a melee weapon attack, you can re-roll the weapon its damage dice and choose the higher result.
If the rerolled damage <= the initial roll, you may add half the maximum of the weapon damage dice to this result and pick that sum.
Once per round on a critical hit on a melee weapon attack, you may reroll the damage dice (but not add half max damage dice).

How do you calculate the half of maximum weapon damage?
For d12, 2d6 and 3d4 this is 6.
For d10 or 2d5 this is 5 and so on.

So for example you are fighting with a 2d6 weapon and you rolled 8 for damage. You choose to use the Savage Attacker re-roll and you roll a 5. This wille become 11 (=5+6) instead and will give you a +3 bonus damage.

I hope this wording makes it more clear. I have found that once you have used this feature the adding half max dice damage becomes more natural and isn't too complicated.

Here is my analysis in a graphic of this feature for Great-Axe comparing to Great-Sword with +2STR.
https://ibb.co/YX6nYG4 Link to graph (https://ibb.co/YX6nYG4)

As you can see in the graph, the expected damage (which is where about 75% of the curve lies) with Savage Attacker and redux version lies between 7 and 12 damage for 1d12. The main difference in the redux version is that the chance for lower damage is shifted towards a chance of higher damage. The chance for rolling only 1 is even removed completely.
You can also see that I simply increased the chance for 2d6 +2STR by 5% to visually show the increased chance of hitting.

https://ibb.co/wCs1tT8 Link to graph (https://ibb.co/wCs1tT8)
Here the accumulative chance is shown to roll equal or higher. Which is more relevant if you need a certain amount of damage to make sure you finish of an enemy.
Some interesting points are that with SA redux we now have 70% chance of roll 9+ on the damage roll, add in your static bonus and you will know how much HP one hit will remove most of the time. If we take +5 damage bonus, we get to finish 14 HP quite reliably.
Comparing the same point (about 75%) with other damage outputs:

For normal d12 it's only 4
For d12 SA it's 7
For d12 SAredux, it's 8,5
For 2d6 with +2 STR, it's also 7

So here we can see that Savage Attacker didn't really was much of an improvement over choosing +2 ASI.

If we look at the 50/50 point (50% of damage equal or higher) the numbers become:

For d12, 7
For d12 SA, it's 9
For d12 SAredux, it's 10,5
For 2d6 +2STR, it's 8,5



If we look to get lucky around 25% the numbers are:

For d12, 10
For d12 SA, 11,5
For d12 SAredux, 12,5
For 2d6 +2STR, 10,5



So it looks like the Savage Attacker redux feature is granting about +2 damage for one roll more than +2 STR with 2d6 would provide on each hit.

If you would compare it to 1d12 +2STR it provides a boost of about 3,5 to 2,5.

So nothing overly powerful, yet it increases dramatically the reliability of that one attack.
That's why I added in the +1 ASI and a free reroll on the first critical hit.

My reasoning behind this was fluff: a savage attacker will be better physically (+1 ASI) and will be better at hitting but not all the time, as he spends his energy into attacking savagely.
The part of improving one critical hit, is mainly fluff because it wouldn't seem convincing that a Savage Attacker could whiff on a critical hit as often as other warriors would.

I must admit that I have develop this feature with a Half-Orc Barbarian in mind who wants to use a Great-Axe and dip 3 levels into Fighter to got crit-fishing. So he will be taking GWM as well. Where he will have some synergy going on because SA redux will help secure a kill from time to time and generate and extra attack, if he didn't crit already.
My only fear is that the synergy with a already great feature would make this build really powerful, yet the build would only come online at level 11 and I presume that other classes will have gained a lot of power by then also.

I hope that this version of Savage Attacker will also appeal to other character builds, who are looking for a reliable damage output on one hit per round. As you can now also spare the reroll for a reaction, it could combine well with Sentinel and the half ASI part makes it viable if you are sitting on an odd value in STR or DEX.