AC softcap and low level hardcap discussion

[DeathsSilkyMist]

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

From the haynar post you linked:

"At low levels the softcap is more level based than defense based. I basically doubled transition so at low levels ac means more.

I added a low level raw ac cap of level * 6 + 25."

I understand it as a worn AC softcap but it is ambiguous.




Unless I'm misunderstanding something a softcap would mean that you get less return every AC point but still get something past the softcap, until you bottom out max hits or run out of AC. You get literally no improvement from additional AC past 45. That looks like a hardcap to me. Only way to go beyond is a shield but the 55ac shielded parse doesn't prove there's a hardcap, it shows you bottomed out max hits at 0. It is probable that 55ac is where max hits bottom out though.

Haynar was adamant there is no hardcap so you should always get some return from AC after you reached softcap but that's not what we're seeing.

1. Haynar says "raw ac cap" for the level * 6 + 25 formula, not "softcap" or "hardcap". Since he said both "softcap" and "hardcap" in the same post, I would be suprised if he decided to use "raw ac cap" to mean "softcap" or "hardcap". The specific wording indicates a clamp on worn AC.

2. Haynar also confirms there is a softcap in that post, even at low levels, so we know a softcap exists. He also told us shields increase the softcap, so that should be testable.

3. When Haynar is talking about not having a hardcap, he is referring to an Eashen raid from my understanding. The worn AC clamp is specified for "low levels". This means Haynar is correct that a bunch of level 60s in a raid are not clamped on AC or hardcapped.

4. I'll show some examples of how a softcap works, and how a worn AC clamp will affect it:

A. There is a 45 AC softcap, 20% diminishing returns after the softcap, and no worn AC clamp. The player has 400 worn AC, and no shield. 45 Uncapped AC + 355 softcapped AC * 0.2 diminishing returns = 116 worn AC that is used in combat.

B. There is a 45 AC softcap, 20% diminishing returns after the softcap, and a 55 worn AC clamp. The player has 400 worn AC, and no shield. First the 400 worn AC gets clamped to 55. 45 Uncapped AC + 10 softcapped AC * 0.2 = 47 worn AC that is used in combat.

C. There is a 45 AC softcap, 20% diminishing returns after the softcap, and a 55 worn AC clamp. The player has 400 worn AC, and 12 of that AC is from the shield. First the 400 worn AC gets clamped to 55. Then the softcap gets increased. We don't know how much a shield increases softcap but lets say the softcap goes up by 5. 50 Uncapped AC + 5 softcapped AC * 0.2 = 51 worn AC that is used in combat.

My data reflects these examples, and it is consistent with what Haynar said.

I will start using the wording "worn AC clamp" instead of hardcap. I think this might be the disconnect. A hardcap would imply that the AC is capped after the softcap is applied. A "worn AC clamp" would clamp the worn AC first. There is a bit of nuance there, but I think that is the confusion. This is why he said "raw ac cap" instead of softcap or hardcap. If there was a 55 AC hardcap, you could still get to 55 AC with enough softcapped AC. At 20% returns a person with 95+ AC would hit 55 AC.

[bcbrown]

The shield parse looks about the same to me as the others. I'd want to see another parse replicating it before I'd say there's any impact from a shield here. 400 hits is very noisy and the difference between 1 max-hit and 7 max-hits could easily just be noise.

[DeathsSilkyMist]

Originally Posted by bcbrown [You must be logged in to view images. Log in or Register.]

The shield parse looks about the same to me as the others. I'd want to see another parse replicating it before I'd say there's any impact from a shield here. 400 hits is very noisy and the difference between 1 max-hit and 7 max-hits could easily just be noise.

Two 400 hit parses with a shield, one at 55 AC and one at 190 AC both show an average minimum hits of 179. That is 800 hits total.

Three 400 hit parses without a shield at 55 AC, 61 AC, and 178 AC show an average of 164 minimum hits. That is 1200 hits total.

The datasets are larger than 400 if you agree that I am not getting any additional returns at 55 AC and beyond.

There is a clear pattern where minimum hits increase and other hits decrease as AC increases when looking at all the data. This is also the expected result for the D20 function.

[bcbrown]

Perhaps. There's a hint of something being there, but nothing conclusive yet. I'm still dubious.

[DeathsSilkyMist]

Originally Posted by bcbrown [You must be logged in to view images. Log in or Register.]

Perhaps. There's a hint of something being there, but nothing conclusive yet. I'm still dubious.

I will try to run some larger tests. These smaller tests were to show that the possibility exists, so it warrants further investigation. These tests do take a while to conduct, and mobs can deagro you over time if you aren't occasionally doing something like auto attacking with bare fists (so you don't kill the mob).

[Goregasmic]

Originally Posted by DeathsSilkyMist [You must be logged in to view images. Log in or Register.]

1. Haynar says "raw ac cap" for the level * 6 + 25 formula, not "softcap" or "hardcap". Since he said both "softcap" and "hardcap" in the same post, I would be suprised if he decided to use "raw ac cap" to mean "softcap" or "hardcap". The specific wording indicates a clamp on worn AC.

2. Haynar also confirms there is a softcap in that post, even at low levels, so we know a softcap exists. He also told us shields increase the softcap, so that should be testable.

3. When Haynar is talking about not having a hardcap, he is referring to an Eashen raid from my understanding. The worn AC clamp is specified for "low levels". This means Haynar is correct that a bunch of level 60s in a raid are not clamped on AC or hardcapped.

4. I'll show some examples of how a softcap works, and how a worn AC clamp will affect it:

A. There is a 45 AC softcap, 20% diminishing returns after the softcap, and no worn AC clamp. The player has 400 worn AC, and no shield. 45 Uncapped AC + 355 softcapped AC * 0.2 diminishing returns = 116 worn AC that is used in combat.

B. There is a 45 AC softcap, 20% diminishing returns after the softcap, and a 55 worn AC clamp. The player has 400 worn AC, and no shield. First the 400 worn AC gets clamped to 55. 45 Uncapped AC + 10 softcapped AC * 0.2 = 47 worn AC that is used in combat.

C. There is a 45 AC softcap, 20% diminishing returns after the softcap, and a 55 worn AC clamp. The player has 400 worn AC, and 12 of that AC is from the shield. First the 400 worn AC gets clamped to 55. Then the softcap gets increased. We don't know how much a shield increases softcap but lets say the softcap goes up by 5. 50 Uncapped AC + 5 softcapped AC * 0.2 = 51 worn AC that is used in combat.

My data reflects these examples, and it is consistent with what Haynar said.

I will start using the wording "worn AC clamp" instead of hardcap. I think this might be the disconnect. A hardcap would imply that the AC is capped after the softcap is applied. A "worn AC clamp" would clamp the worn AC first. There is a bit of nuance there, but I think that is the confusion. This is why he said "raw ac cap" instead of softcap or hardcap. If there was a 55 AC hardcap, you could still get to 55 AC with enough softcapped AC. At 20% returns a person with 95+ AC would hit 55 AC.

1. I assume raw AC cap means worn as opposed to displayed. It would be a dick move to introduce a new concept without defining it.

2. Agreed

3. Probable, I came to the same assumption.

4. If you're going to talk about softcaps or hardcaps I fully expect it to be option A with no clamp. That's basically how other stats work (int/wis/cha). In scenario B, C and D worn AC clamp is literally a hardcap in function if you don't use a shield.

I think it is kinda pointless to discuss haynar's hidden meanings though, if you're right the numbers are the numbers. His formula puts the "soft cap" at 259worn @lvl39 and 319worn @lvl49. 259 is basically beyond BIS for tunnel gear and 319 is probably around straight up BIS if even. So no point even bothering. The only thing that matters would be finding clamp points. At clamp points you nearly bottomed out the max hits anyway so I guess that is good enough. A 49 ranger seems to be clamping around 180ac which is a lot more reasonable goal, 51+ players around 200.

If mob level has no bearing on AC parses, like previously thought, that could explain why people claim going from tunnel BiS to straight up BiS feels like it does nothing. Everyone who has 200+ AC are already clamped anyway.

[DeathsSilkyMist]

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

1. I assume raw AC cap means worn as opposed to displayed. It would be a dick move to introduce a new concept without defining it.

2. Agreed

3. Probable, I came to the same assumption.

4. If you're going to talk about softcaps or hardcaps I fully expect it to be option A. That's basically how other stats work (int/wis/cha). In scenario B, C and D worn AC clamp is literally a hardcap in function if you don't use a shield.

I think it is kinda pointless to discuss haynar's hidden meanings though, if you're right the numbers are the numbers. His formula puts the "soft cap" at 259worn @lvl39 and 319worn @lvl49. 259 is basically beyond BIS for tunnel gear and 319 is probably around straight up BIS if even. So no point even bothering. The only thing that matters would be finding clamp points. At clamp points you nearly bottomed out the max hits anyway so I guess that is good enough. A 49 ranger seems to be clamping around 180ac which is a lot more reasonable goal, 51+ players around 200.

If mob level has no bearing on AC parses, like previously thought, that could explain why people claim going from tunnel BiS to straight up BiS feels like it does nothing. Everyone who has 200+ AC are already clamped anyway.

Yes, I agree he is talking about worn AC. That is the better metric to go by, as the displayed AC in the UI is basically useless, other than letting you know if your AC went up or down.

Mob level does have an impact on parses.

Going back to the D20 function:

1. The game has a Roll D20 function with two inputs. I will call these inputs "Wrath" (attacker) and "Mitigation" (defender). The code confusingly calls them "offense" (attacker) and "defense" (defender), but these inputs are more than simply the attacker's offense skill and the defender's defense skill.

2. The D20 roll is weighted based on the ratio of the attacker's Wrath to the defender's Mitigation. An unweighted D20 has an average roll of 10.5.

3. If the attacker's Wrath is 50 and the defender's Mitigation is 100, the D20 has a weighted average roll of ~6.5. This is when you see the cluster of rolls at the minimum damage value.

4. If the attacker's Wrath is 100 and the defender's Mitigation is 50, the D20 has a weighted average of ~14.5. This is when you see the cluster of rolls at the maximum damage value.

5. If the attacker's Wrath is 50 and the defender's Mitigation is 50, the D20 has the unweighted average of ~10.5. This is when you see a roughly equal amount of rolls at the minimum and maximum damage values.

The lower the mob's wrath, the lower their damage is against you.

As an extreme example, let's take an orc pawn attacking a level 60. A level 60 Warrior is going to have at least 300 mitigation, and an Orc Pawn's wrath is probably going to be well below 100.

If 50 Wrath and 100 Mitigation weights the D20's average rolls to ~6.5, a Wrath value of 50 and a mitigation value of 300 would weight the average dice roll even lower. This is why orc pawns basically always hit high level characters for 1 if they aren't sitting.

[DeathsSilkyMist]

I did a longer test. Two sets of 1400 hits with 178 AC. One set with a 12 AC shield, one set without.

==============================
178 AC, 12 of this AC is from a shield
==============================

DV, Count
1, 603
2, 84
3, 92
4, 43
5, 118
6, 77
7, 83
8, 41
9, 87
10, 100
11, 62
12, 10

Total Damage = 5765

===============
178 AC, No shield
===============

DV, Count
1, 580
2, 78
3, 91
4, 34
5, 92
6, 99
7, 101
8, 56
9, 98
10, 102
11, 50
12, 19

Total Damage = 6034

Even at higher sample sizes, the shield is still performing better. Less damage and more minimum hits. The difference is small, so it looks like the softcap increase from a shield isn't 1 AC = +1 Softcap, at least at lower levels. I probably only got 1 or 2 extra AC.

In my opinion, this indicates that the level * 6 + 25 formula is a worn AC clamp for low levels, not a softcap. The falloff past ~45 AC comes from the softcap. Shields do increase the softcap, but it is not a 1 AC to 1 Softcap increase at level 5. It may increase as you level up.

[Jimjam]

So you think there is a cap on shield overcap ac? Maybe it is level based? so at level 5 maybe you can overcap by 5 ac?

[DeathsSilkyMist]

Originally Posted by Jimjam [You must be logged in to view images. Log in or Register.]

So you think there is a cap on shield overcap ac? Maybe it is level based? so at level 5 maybe you can overcap by 5 ac?

Yes, and it makes sense when you think about it. If the softcap increase was 1 AC to 1 softcap, a 45 AC softcap would be completely removed by a 10+ AC shield when a level 5's worn AC clamp is 55.

This would make shields a bit OP at low levels, especially with the abundance of 25 AC Sarnak Battle Shields.

The data right now indicates a 12 AC shield would only increase your softcap by 1 or 2. Maybe it's 10% of the shields AC at level 5, but I haven't compared other shields yet.