Jayde

Hey all, as people who have tried it in the past probably found the Rawr.ProtWarr module was pretty outdated post-3.0, and really didn't have a major look as the original major maintainer of the code was no longer working on it. It was being maintained by HolyMist post-3.0, but he did not have time to do the major overhauls to the base code needed to bring things in line with all the WotLK changes.

HolyMist and I have been working extensively on revamping the module over the last week, and today's launch of the 2.1.9 version of Rawr includes almost all of the updates we have been working on.

With me focusing full-time development on the module at the moment, I'm hoping to get valuable feedback from the Warrior community in regard to what you feel is needed to make Rawr.ProtWarr more accurate, more useful, or generally more robust as a theorycrafting tool. The Mage community has been extremely supportive of Kavan's efforts with Rawr.Mage, so it would be excellent if the same could happen here.

There are some things I am already working on which should be coming soon:
-Full rage-generation module so that the 'Limited Threat' options are more accurate
-Some additional control and methods for ranking avoidance vs. effective health
-Some more research into spell damage reduction, as it appears recently that the resistance formula has changed for WotLK
-Implementation of Glyphs
-A proper way of looking at the impact of Shield Block
-Better implementation of Mongoose and various other Use/Hit items (Essence of Gossamer, for instance)

Some things that I need feedback on:
-Any issues with stats/calculations showing up wrong
-Any suggestions about alternate methods of gear weighting
-Feedback on if Shield Block should be implemented as a toggle state to check off/on, or if it should be averaged based on uptime
-Data on the base damage/speed of the major WotLK bosses (10 and 25-man)

It's been quite a lot of work getting the module to the current state from where it is before and hopefully it can serve as a useful tool going forward. Any feedback or reports are greatly appreciated.

 

Twid

You should remove the Adamantite Weightstone from the Temp. Weapon Enchantments, as they cannot be used on level 80 weapons. I'm playing around with this, and I'll report any other issues I find.

Edit - Is it possible to model expertise as a defensive stat? Currently it only values it as a threat stat, whereas the parry haste reduction is missing.

 

Jayde

I'll see if I can get the weightstones and such removed, good catch.

In regard to Expertise, I would definitely like to get that in there so will see if I can prioritize the addition of code that will allow for that. It's a bit tricky since both the defensive and offensive side of things can increase the speed of the opposing side which means the offensive model can increase the damage taken by the defensive model and visa-versa. However, I'll add it to my list.

 

Xav

I was using the model and noticed something odd with expertise as well. Gems with expertise had 0 defensive value associated, but gear with expertise as a raw stat seemed to definitely be calculated with it as a defensive stat. Pauldrons of Unnatural death for example were rated quite high when I had it only set to display mitigation rankings and not threat. If they were ranked without the expertise in the calculation I assume they would have been nowhere close to T7 in the scale, same with legplates of sovereignty and other items.

 

Suesse

[Pauldrons of Unnatural Death]
[Valorous Dreadnaught Pauldrons]

Same armor, approximately same stam, approximately same avoidance, but T7 has block rating where the other has expertise. I wonder if your average unmitigated hit bar is maxed, I think that will cause Rawr.ProtWar to consider block rating to have a very low value.

Were the "other items" high in block rating as well?

 

Jayde

I think in regard to that particular comparison it's because Block Value % is not very highly impactful when looking at large boss-style hits.

T7.5 Gains:
9 Strength
50 Dodge
29 Block
6 Stamina (with bonus)

Unnatural Death Gains:
11 Defense (with bonus)
53 Parry
30 Expertise

So, realistically, they aren't too far apart. Due to the higher defense, Pauldrons of Unnatural Death actually have a bit more avoidance (0.07%) ignoring the block%, which leaves the block% and slight block value gain from strength to pull out the difference. With unmitigated hits coming in around 9-12k on average, the block% is likely to have a minimal impact on the overall TTL. (It's certainly a factor, but probably not enough to put it significantly ahead.)

I'd imagine once I factor in Expertise to defensive stats as well, the T7.5 will fall behind.

 

Xav

Wont really know till later since Rawr causes my terrible laptop to grind to a halt.

I haven't done an actual in-game comparison of the differences, but I assumed the parry rating on the Unnatural Death pauldrons to be hit pretty hard by diminishing returns, letting the Valorous pull ahead. Particularly if it ignores expertise. I don't recall how large the margin was of Unnatural Death over Valorous.

 

Jayde

In a quick look, Valorous looked roughly 10% better in terms of points with my current stats (without threat included). I suspect on a parry-hastable mob they would be quite similar (if not with Expertise pulling ahead) or in any situation where threat was a major factor.

 

Jayde

I got parry hasting in this evening and, provided it's accurate, it should be providing sufficient survivability increases based on Expertise.

It's not a huge change, as Expertise isn't a massive change to survival... if you consider basically that even with an 8.75% parry rate, you are only increasing the boss auto-attack speed by a little over 5%. (From 2.0s to 1.89s according to my figures.) So, the returns are slow

However, it is still a factor. Back to Unnatural Death vs. T7.5.

Settings: 35k base attack, 2s attack speed, no parry haste, 13100 armor, 0 threat scale, TankPoints ranking.
Pauldrons of Unnatural Death w/ Stam/Expertise gem: 465.22
Valorous Dreadnaught Pauldrons w/ Stam/Expertise gem: 500.22

Settings: 35k base attack, 2s attack speed, parry haste, 13100 armor, 0 threat scale, TankPoints ranking.
Pauldrons of Unnatural Death w/ Stam/Expertise gem: 462.55
Valorous Dreadnaught Pauldrons w/ Stam/Expertise gem: 472.40

Settings: 35k base attack, 2s attack speed, no parry haste, 13100 armor, 0 threat scale, normal ranking (1.0 mitigation).
Pauldrons of Unnatural Death w/ Stam/Expertise gem: 152.92
Valorous Dreadnaught Pauldrons w/ Stam/Expertise gem: 158.70

Settings: 35k base attack, 2s attack speed, parry haste, 13100 armor, 0 threat scale, normal ranking (1.0 mitigation).
Pauldrons of Unnatural Death w/ Stam/Expertise gem: 159.53
Valorous Dreadnaught Pauldrons w/ Stam/Expertise gem: 162.68

So, expertise does help. According to the output, my damage taken with parry haste on is 1920.8 DPS vs. 1814.9 DPS. (This is with 20 expertise, leaving 8.75% parry chance.)

One slot, in particular, I noticed the top-rank item actually does change with parry haste turned on considering my current stats... that being the Sabatons of Endurance vs. Kyzoc's Ground Stompers. Even in TankPoints mode where Kyzoc's has a large lead for me, Sabatons of Endurance make up large ground and take the top spot with parry turned on. Signet of the Accord also makes up large ground compared to other ring itemization.

Another interesting thing is that slower weapons gain a lot of survivability compared to fast weapons with parry hasting turned on.

Anyhow, interesting data at least...

As another note, I will be integrating a 3rd alternate weighting method based on Hypatia's "Burst Time" theory found here. Which, summed up, is basically the average number of hits that a burst could potentially happen to the tank. It can't be easily broken down into survivability/mitigation like the other methods because it looks at the increasing or diminishing returns of both based on the current damage intake and chance of hit streaks vs. health. It's a good way, IMO, to rate gear on a scale of decreasing risk of death statistically.

I should have that added in the next internal build.