Hold the Line Maths

[Airowird]

Koji, how did you calculate the value for HtL on Parry?
Because that difference in ratings seems very large.

[Kojiyama]

Well, basically the value attributed to Parry would be equal to Parry's innate value plus the increased uptime of the buff and its impact on the current Mastery situation. With the higher levels of Mastery (and also, keep in mind I was calculating Shield Block uptime into this--which will almost double the value of Critical Block when it's up) this seems to be fairly beneficial.

Obviously, with lower levels of Mastery and Shield Block uptime, the value of the Critical Block gains will drop dramatically and therefore be far less beneficial. So the relative gains of Parry in this scenario definitely depend on your current stat levels.

[Airowird]

Similar to how I did it then, but it's just weird that you get WAY more out of it than Mellvar's example, which isn't that absurdly geared, or is it?

[Kojiyama]

Are you sure I'm getting a value much, much better? I mean, the rating difference does not imply value difference--as both Dodge and Parry are going in the same direction, it will take a longer time to 'catch up' to something with even a diminishing advantage as it has to make up for the cumulative difference.

It's not so much that Parry is substantially better than Dodge, it's just that it takes a long time for Parry to become worse than Dodge. They are two fairly different things.

When I get home I can grab the exact numbers at equalish rating amounts, for instance, but I don't recall them being much greater than the 10% point you mentioned. It just takes a while for the DR curve to make Parry poor enough that Dodge can catch up with the cumulative value of being 10% greater every step along the way. (Considering that Dodge also diminishes going upwards as well.)

[praetoria]

this is awsome.. thanks so much guys. gives a whole new angle to craft from.. I bow to a force far greater then my own.

[MellvarTank]

Thanks to both of you guys for weighing in on this.

[alkihaul]

Great analysis and thanks for the input guys.

So what I am taking away from this is the Parry should be prioritized over Dodge, even at a high differential between the two such as a 10% delta.

[Quinafoi]

Is there any consideration in your math for the fact that the target is three levels higher?

For example, Actual miss chance is 5% - 0.6% = 4.4%. The boss only does 100% of their attacks, they don't attack you 102.4% of the number of swings they have in that time. This applies to all the avoidance factors. Since the actual number of attacks is set to 100%, the avoidance factors have to account for the target level and be shifted to the 100% scale.

So out of 195 swings, and 4.4% of them miss, that's 8.58 misses over the course of the fight (naturally a "partial" miss can't actually occur).

Also since your calculations of how much of each avoidance doesn't include level scaling they are in the wrong porportion. Because the 0.6% decrease has a greater impact on Miss which only had 5% to begin with relative to the same 0.6% decrease to something like Parry that was over 12%.

Total actual avoidance if factors are...

Miss = 5 - 0.6 = 4.4%
Parry = 12.59 - 0.6 = 11.99%
Dodge = 10.46 - 0.6 = 9.86%

Avoidance = 4.4 + 11.99 + 9.86 = 26.25% (you're math resulted in 28.05%)

26.25% of 195 attacks is 51.1875 avoided attacks (again can't "partially" avoid so this would be rounded).

The percentage of attacks avoided that would be missed is...
4.4 / 26.25 = 16.76% (you're math came up with 17.83%)


All of this math needs to be converted to a 100% scale cause there are errors all the way through if you think the boss makes 102.4% of their total number of attacks.

[Quinafoi]

Also there is one factor not included which can cause a variable in avoidance. The Night Elf racial, Quickness. Since that alters the miss chance, it ultimately affects all the math involved, and the end calculated value of Hold the Line would be different for a Night Elf.

[Kojiyama]

Rawr at least takes that into account, although Miss is unlikey to push Parry off the table at any point--so uptime is not affected.

And, yes, all my calculations were done against a level 88 target.

[Petninja]

Also there is one factor not included which can cause a variable in avoidance. The Night Elf racial, Quickness. Since that alters the miss chance, it ultimately affects all the math involved, and the end calculated value of Hold the Line would be different for a Night Elf.

A three inch stick is still a three inch stick no matter how far you slide it up a ruler. The probability of a parry does not change due to increased miss.

[Kojiyama]

Well, Quickness very slightly reduces the value of the Critical Block by pushing Block (and thus Critical Block, since it's a two-roll system) further off the table at higher amounts of Block or with Shield Block up.

However, this is a pretty minor impact.

[Quinafoi]

In the greater scheme of things the numbers are very different, what is good for a Warrior will also be good for a Night Elf Warrior. However in the finer details the math is slightly different and the actual calculated values is slightly different.

As for the other stuff regarding level, I wasn't really directing it at Rawr's calculations, but rather the calculations posted in this topic which do have the flaw of not adjusting avoidance levels down to a 100% scale, since the sum of all events must always be 100% in practical application.

[MellvarTank]

I know what you are getting at Quinafoi, and it makes sense. The reason I have not updated the OP is that Wartotem and Koji have posted (as I'd hoped) better models/tools for calculating this talent. I didn't have the tools to appropriately model avoidance, so had to use linear examples which are, at best, a guess at how the talent would play out.

According to the modified numbers, I would guess that assuming a linear dispersion of avoidance and also assuming that you cannot parry while HtL is up have a bigger effect on the numbers than the .6%.

[mistersix]

so is rawr uptodate?

[Kojiyama]

Should be for the most part, yes. (I thiiiink they managed to squash the HtL bugs.)

Although I will be adding in Shield Block uptime support in the post-PTR patch (next week most likely?) which will boost the value slightly. That should give a full and accurate picture.

[Kojiyama]

New version of Rawr is up, and regardless of if you use it for looking at Stamina vs. Avoidance (old debate, unimportant right now!) it should be highly useful for weighting avoidance vs. avoidance, or specifically the tug-o-war between Parry, Dodge, and Mastery.

I've found it to be quite interesting in playing around with various numbers.

The reality is that with a decent amount of Mastery (say 2k+ rating) Parry is the highest until you have around 2.5% more Parry than Dodge, and at that point Mastery, Parry, and Dodge become very similar. After that, it's a balancing act between increasing Mastery and Parry, which have a bit of a feedback loop between them.

A good way to do this automatically is simply to load up your character and open the optimizer (Tools -> Optimizer) and optimize for Mitigation (rather than Overall.) This should still leave you with plenty of health (is generally only going to regem/reforge, which won't affect your health much) and you can swap health back in as you see fit. But, either way, it's a great way of balancing out your three avoidance stats.

The 'Avoidance' mouseover in the character pane should show you your combat tables both with and without Shield Block up, and you can set your Shield Block uptime in the options now.

[Katzazi]

Ahh thank you Kojiyama. I really waited for this update, because the last versions seems to have some issues left there. I'll try it out directly. :-)

[Kazeyonoma]

Lets just move along, discussion is over, I've cleaned out the thread to prevent spreading of misinformation.

[Iyona]

The reality is that with a decent amount of Mastery (say 2k+ rating) Parry is the highest until you have around 2.5% more Parry than Dodge, and at that point Mastery, Parry, and Dodge become very similar. After that, it's a balancing act between increasing Mastery and Parry, which have a bit of a feedback loop between them.

A good way to do this automatically is simply to load up your character and open the optimizer (Tools -> Optimizer) and optimize for Mitigation (rather than Overall.) This should still leave you with plenty of health (is generally only going to regem/reforge, which won't affect your health much) and you can swap health back in as you see fit. But, either way, it's a great way of balancing out your three avoidance stats.

Rawr does not like me. The downloaded version will not run, saying I don't have .NET 4 installed. The online version crashes when I'm optimizing.

That's not why I write, though. I was wondering, when you say "2.5% more parry than dodge", what do you mean? My toon has more than 4% more parry than dodge, but only roughly 200 more rating points (~16% more). I feel further increasing my parry still should be the best option, but I feel uncertain over when increasing dodge instead would reduce more damage. User Kolmagorov on MMO-Champion drew a graph suggesting a rough break even point at - in the worst case scenario - 30% more parry rating points than dodge rating points from your gear.