Archived

This topic is now archived and is closed to further replies.

bestpike

[Enhancement]: Critical cap analysis (Tow/HotC)

104 posts in this topic

Yes indeed using AEP is just a human convenience. I plead guilty, I use them pretty often. I better not post my math analysis on Hit valiue after crit cap. It was based on AEP.

I will just say some general things about it. When a crit capped person acquires more hit rating (and is not already white attack hit capped of course), and even with that hit rating he is still crit capped, all the newly added white attacks are criticals. So, while you are getting hit, in reality it is crit you are getting, as far as white attacks are concerned. Yellow attacks remain unaffected.

And since when spell hit capped, all that hit rating is offering is more white attacks, when crit capped hit rating should be offering double the amount of damage. But this of course takes for granted that you are constantly and always crit capped, which isn't true. Even with all the raid buffs, Elemental devastation should drop at some point in fight, it doesn't have a 100% uptime. So depending on how much is your average ED uptime (which also varies depending on boss), and how far beyond the crit cap you are, you can calculate the value of Hit rating, which of course is just a convenience as Levva very correctly stated.

Please correct me if I'm wrong somewhere, I wrote all this without giving it great thought.

Share this post


Link to post
Share on other sites

Post it, I'm curious and I don't think EP is such a bad thing provided people are aware of it's limitations.

Share this post


Link to post
Share on other sites
Yes indeed using AEP is just a human convenience. I plead guilty, i use them pretty often. I better not post my math analysis on Hit valiue after crit cap. It was based on AEP.

I will just say some general things about it. When a crit capped person acquires more hit rating (and is not already white attack hit capped of course), and even with that hit rating he is still crit capped, all the newly added white attacks are criticals. So, while you are getting hit, in reality it is crit you are getting, as far as white attacks are concerned. Yellow attacks remain unaffected.

And since when spell hit capped, all that hit rating is offering is more white attacks, when crit capped hit rating should be offering double the amount of damage. But this of course takes for granted that you are constantly and always crit capped, which isn't true. Even with all the raid buffs, Elemental devastation should drop at some point in fight, it doesn't have a 100% uptime. So depending on how much is your average ED uptime (which also varies depending on boss), and how far beyond the crit cap you are, you can calculate the value of Hit rating, which of course is just a convenience as Levva very correctly stated.

Please correct me if I'm wrong somewhere, I wrote all this without giving it great thought.

The thing is you can see very very clearly with the graph feature in Rawr how Hit suddenly rises then falls back around the crit cap. This is similar to using EP values but not for a single value but for multiple values. ie: The program takes the current gear and say ok what's the dps if I add 5 hit rating, then 10, then 15 etc etc.

EP values do the same idea it adds a fixed amount specified in the EP values section of your EnhSim config. However all you end up getting is a SINGLE point on the graph. Now if you happen to have a graph where values are rising and falling due to caps what you are doing with EP values is taking a single set of values at a single point on the graph and trying to make declarations of behaviour at other points on the graph with the invalid assumption that its constant.

If instead you look at the graph as a whole you get a clearer picture. You see the rises and falls, you see where the graph levels off.

Lets take Rouncer's gear from the armoury. He's already optimised it (far too efficient for his own good) so I unoptimised it by changing a 20 hit gem in his gloves to a 20 haste gem. This then gave what he saw before the conversation about crit caps emerged.

4292136269_a865472ec0_o_d.jpg

Now note two things the way that expertise hits the zero line considerably to the left ie: perhaps a little over capped, also note that the hit rating does some funny things it changes angle twice. Once at the point off to the left where he passes the spell crit cap and once where he can overcome the melee crit cap with some extra hit.

I also stuck a red line on the map to show you what you'd get from EP values ie: you'd get hit EP of around 2.3, haste around 2.05 and AP at 1, the rest would be below 1 (note that haste & AP are almost on top of each other ie: haste is only just ahead of AP).

Now what does 2.3 EP hit rating mean?? The idiots out there would have you believe that means "regem for hit" ie: change your gear around and replace your haste gems for hit gems. Is this right? CLEARLY not. We can see that hit rating tails off significantly after Rouncer adds about 10 hit rating. So what does the graph tell us that EP cannot tell us?

The graph shows that he needs about 10 more hit rating before it tails off and that he's about 10 expertise rating too much. Lets test that theory and change his leggings gem from 20 expertise to 10 expertise and 10 hit rating. If you do the calcs in the sim you see that this is a BIG upgrade over any other gem choice for that slot. Once that single gem choice is changed you get the following graph...

4292877750_670172055a_o_d.jpg

Note that the kinks have been dealt with and now haste and AP are neck and neck with hit rating no longer an issue.

So what does this all mean? Well from the graph I was able to see that LOSING some expertise and gaining some hit was what was required. No amount of massaging EP values would EVER have seen that solution. In fact traditional use of EP values would have wrongly suggested that "stacking" hit was a good idea when clearly it wasn't. This is to me damning proof that EP values can be fatally flawed and whilst potentially useful for quick comparisons in game have no place in serious discussions about how we should gear.

In this example the only EP values available are represented by the vertical red lines. As you can see they are a tiny snapshot of what is going on and since the EP values (the red lines) were in the "wrong place" they gave a false indication of how Rouncer should gear. The only way you could have found this out with EP values would have been to run the sim dozens of times with lots of different EP steps ie: effectively producing a graph.

In fact my little tweak using the graph raises his dps from 9098.45 to 9112.04 which is a 14 dps rise from changing a gem. In fact that's the upgrade he'd get from his OPTIMISED gear. The change of a single gem from graph 1 to graph 2 was changing 20 expertise to 10 expertise 10 hit changing dps from 9085.39 to 9112.04 ie: using the graph and changing a single gem gave a 26.65 dps upgrade.

Share this post


Link to post
Share on other sites

I just got the 258 gloves last night so hadn't had a chance to rework my gemming to account for the additional expertise on the gloves.

Levva would it be possible to add a way to zoom in and out on the graph? Maybe give an option to control the size of the Stat Change variable in the X axis? I'd love to be able to zoom into the 20 or 40 stat change area to see what is going on when they are all clumped up like that.

As for the talk of the EP value of hit when just over the crit cap, I would be interested in seeing the math for it's value since that goes into determining the value of adding hit past the spell hit cap when it would also be increasing the value for crit on white strikes as well. Logically it makes sense that hit would then be the best stat to gem for until you got back under the crit cap when Elemental Devastation (as a high uptime proc) was active but seeing the math may help to quantify it better.

Share this post


Link to post
Share on other sites

/script DEFAULT_CHAT_FRAME:AddMessage("Crit:"..((UnitStat("player", 2)+178+52)*.1+178+52)/83.33+3+5+9+GetCritChance())

/script DEFAULT_CHAT_FRAME:AddMessage("Critcap:"..100-24-27+GetCombatRatingBonus(6)+6)

This takes your unbuffed stats and spits out your raidbuffed critchance+critcap, the only thing I'm not sure of is if kings is calculated before or after straight +stat buffs.

Share this post


Link to post
Share on other sites
/script DEFAULT_CHAT_FRAME:AddMessage("Critcap:"..100-24-27-4.8+GetCombatRatingBonus(6)+6)
This line is wrong. You don't have to subtract the 4.8% for the crit depression. If you read through this thread you'll see where Rouncer showed very clearly that this 4.8% crit depression has no effect on the amount of crit required to reach the cap. It should look more like this:
100-(27-(GetCombatRatingBonus(6)+6))-(24)

Share this post


Link to post
Share on other sites

Yeah, that's left over from the old macro and shouldn't be there. I was mostly interested if the first one is correct since I won't raid until sunday.

Share this post


Link to post
Share on other sites
This line is wrong. You don't have to subtract the 4.8% for the crit depression. If you read through this thread you'll see where Rouncer showed very clearly that this 4.8% crit depression has no effect on the amount of crit required to reach the cap.

It should look more like this:

100-(27-(GetCombatRatingBonus(6)+6))-(24)

While I enjoy receiving credit for things in this case that was just me repeating the argument Bestpike made that managed to get my brain to wrap around how it actually worked.

Would one of you mind fixing up that script and posting it in a single [ code ] block so people can have a correct script to use? Pretty sure two of them will be necessary though, since you will need one when there is a Heart of the Crusader, ToW or Master Poisoner debuff on the target and one when it is not present.

Also anyone who has a flawed or incorrect script in their post would you mind going back and editing it out so there is less confusion for people following this thread.

Share this post


Link to post
Share on other sites

I guess I made it known on the shaman part of the forum, but it was actually on the Druid forums that I read this the first time. I forget the name of the poster (Rouncer should have the post link), but thanks to that poster all the same.

On the AEP side of things now, Levva you are right those graphs are much better, they show the truth much more clearer. Although the occasion with Rouncer's gems is slightly different than pure critical cap issues, it presents the stats in a much better way indeed.

Share this post


Link to post
Share on other sites
Apparently the testing that showed that the crit depression couldn't be removed was fubared by an issue with how Recount was interpreting the combat log.

http://elitistjerks.com/f31/t76785-crit_depression_combat_table/p3/#post1548494

Hits can be pushed off the table so everyone has 4.8% more space to deal with before being crit capped.

Interesting finding. So if I understand this correct, you actually need (while expertise and hit capped) 80.8% critical on your character sheet to cap white crits?

Share this post


Link to post
Share on other sites

I opened a ticket at EnhSim so hopefully this will get fixed in the sim by the end of the weekend, EnhSim - View Issue #5963: Crit Cap - Sorry Eralun

Should make the BiS list a bit more interesting at least, since Whispering Fanged Skull and Deathbringer's Will just got a good deal better. Bestpike would you mind updating your spreadsheet? I've found it to be very helpful when looking at how to gem out gear and the crit cap is still going to play a role in our BiS set even with that additional room.

Just to clarify, if you are melee hit capped and dodge capped and hitting a mob from behind the crit cap is indeed 80.8%.

Share this post


Link to post
Share on other sites

Download link of the latest version Crit cap calculator 2.1

Ok I have updated the spreadsheet with the new findings and also some improvements. The changes are the following:

1. Changed the crit cap according to Rouncer's post.

2. Incorporated Expertise rating to the table to make it more accurate.

3. Unified the Draenei and Non-Draenei table.

4. Two new race options. If you are an Orc without any maces equipped, put 1 in the first box and 0 on the second. If you are a Draenei, put 1 to second box and 0 on the first. If you are none, put 0 to both.

As usual, you fill only the RED boxes with info and also while in an unbuffed status. Those are:

1. Char sheet agility,

2. Char sheet melee critical chance,

3. Hit rating,

4. Expertise rating,

5. The race options. Be sure they are always filled with only either 1 or 0.

The green boxes calculate various useful information while the yellow box calculates how much more crit you can have before getting white capped.

Share this post


Link to post
Share on other sites
Ok I have updated the spreadsheet with the new findings and also some improvements. The changes are the following:

1. Changed the crit cap according to Rouncer's post.

2. Incorporated Expertise rating to the table to make it more accurate.

3. Unified the Draenei and Non-Draenei table.

4. Two new race options. If you are an Orc without any maces equipped, put 1 in the first box and 0 on the second. If you are a Draenei, put 1 to second box and 0 on the first. If you are none, put 0 to both.

As usual, you fill only the RED boxes with info and also while in an unbuffed status. Those are:

1. Char sheet agility,

2. Char sheet melee critical chance,

3. Hit rating,

4. Expertise rating,

5. The race options. Be sure they are always filled with only either 1 or 0.

The green boxes calculate various useful information while the yellow box calculates how much more crit you can have before getting white capped.

The new download link is Crit cap calculator v2.0. I will remove the old one. Please report if I have made any mistakes, I wrote the spreadsheet formulas rather fast.

This could be a stupid question, but i do not have Excel. Do i need to download it in order to use this or is there a way around it

Share this post


Link to post
Share on other sites
This could be a stupid question, but i do not have Excel. Do i need to download it in order to use this or is there a way around it

You could just go OpenOffice.org or maybe directly try the spreadsheet over Google Docs, both will do the work.

Share this post


Link to post
Share on other sites

I'm still not sure about the 4.8% value. That may be a rogue specific value. In my crit test some months ago, I had 1.8% critical hits (and 24.0% glancing) in over 10000 attacks vs. the heroic target dummy, while having a constant 4.73% char sheet melee critical chance. 3% seems much more likely to me.

World of Logs - Real Time Raid Analysis

Share this post


Link to post
Share on other sites

I have updated the Crit cap calculator to 2.1. The calculation was wrong when entering an expertise rating high enough to go over the expertise cap.

Download link

Crit cap calculator 2.1

Removing the older link.

Edit: The same calculation error happens when entering a hit rating high enough to go over the white attack hit cap.I won't update to a new version because getting rid of excess hit isn't very hard to do, just don't enter hit values that make your chance to miss a negative number (over the cap) in the spreadsheet.

Share this post


Link to post
Share on other sites
I have updated the Crit cap calculator to 2.1. The calculation was wrong when entering an expertise rating high enough to go over the expertise cap.

Download link

Crit cap calculator 2.1

Removing the older link.

Edit: The same calculation error happens when entering a hit rating high enough to go over the white attack hit cap.I won't update to a new version because getting rid of excess hit isn't very hard to do, just don't enter hit values that make your chance to miss a negative number (over the cap) in the spreadsheet.

Why does your final calculation start with 80.8%? First off, crit depression wasn't real, it was a symptom of people attacking from the front of target dummies and seeing white hits, which were really blocks not hits after all because recount didn't record it properly. Second off if it was real wouldn't you start off @ 71.2%?

To be a little more clear, if you are hit capped and dodge capped attacking a lvl 83 boss from behind you would experience 24% glancing blows which means your total crit rate could be 76% not 80.8%.

Share this post


Link to post
Share on other sites
Why does your final calculation start with 80.8%? First off, crit depression wasn't real, it was a symptom of people attacking from the front of target dummies and seeing white hits, which were really blocks not hits after all because recount didn't record it properly. Second off if it was real wouldn't you start off @ 71.2%?

To be a little more clear, if you are hit capped and dodge capped attacking a lvl 83 boss from behind you would experience 24% glancing blows which means your total crit rate could be 76% not 80.8%.

Crit depression is still a real phenomenon, which means when attacking a boss target 4.8% of your melee crit goes poof. What was wrong was the concept that 4.8% of crit was converted to hits and that those hits couldn't be removed by just adding additional melee crit. So 80.8% is the correct figure for the amount of melee crit required to push hits completely off the table.

It has been suggested that the 4.8% only applies to rogues and that our melee crit reduction is more in the neighborhood of 3% but there has been no definitive testing proving that and due to elemental devastation not having 100% uptime it would be difficult to prove without controlled testing using a build without those talents.

Edit

Decided to confirm/deny the 4.8%. Made a gear set up with no expertise rating except for Orc Racial, no hit rating and no points in Dual Wield Spec. Crit Cap would then be 100 - 27% (miss) - 24% (glancing) - 5.25% (Dodge) = 43.75%

If the crit depression doesn't exist then no hit will be recorded above 43.75% melee crit. If our crit depression is only 3% then no hits will be recorded at 46.75%. If our crit depression is the same as a Rogues then it would require 48.55% melee crit to remove hit from the table.

All testing on the Boss Dummy in TB while attacking from behind to remove parry and block. Stopped series once a single hit was recorded, then added more crit rating (or used an SoE totem) and started the next series. 10k attacks without a hit being enough to prove that hit has been removed from the table.

44.5% melee crit - hit

46.75% melee crit - hit

47.67% melee crit - hit

48.34% melee crit - hit (occurred around 1400 attacks)

48.70% melee crit - no hit so far (took it to 4500 without a hit then got bored but seems enough to prove the point)

With that hit occurring with 48.34% melee crit it seems pretty safe to say that melee crit depression is indeed 4.8% for shaman just as it is for Rogues.

Share this post


Link to post
Share on other sites
Crit depression is still a real phenomenon, which means when attacking a boss target 4.8% of your melee crit goes poof. What was wrong was the concept that 4.8% of crit was converted to hits and that those hits couldn't be removed by just adding additional melee crit. So 80.8% is the correct figure for the amount of melee crit required to push hits completely off the table.

It has been suggested that the 4.8% only applies to rogues and that our melee crit reduction is more in the neighborhood of 3% but there has been no definitive testing proving that and due to elemental devastation not having 100% uptime it would be difficult to prove without controlled testing using a build without those talents.

Edit

Decided to confirm/deny the 4.8%. Made a gear set up with no expertise rating except for Orc Racial, no hit rating and no points in Dual Wield Spec. Crit Cap would then be 100 - 27% (miss) - 24% (glancing) - 5.25% (Dodge) = 43.75%

If the crit depression doesn't exist then no hit will be recorded above 43.75% melee crit. If our crit depression is only 3% then no hits will be recorded at 46.75%. If our crit depression is the same as a Rogues then it would require 48.55% melee crit to remove hit from the table.

All testing on the Boss Dummy in TB while attacking from behind to remove parry and block. Stopped series once a single hit was recorded, then added more crit rating (or used an SoE totem) and started the next series. 10k attacks without a hit being enough to prove that hit has been removed from the table.

44.5% melee crit - hit

46.75% melee crit - hit

47.67% melee crit - hit

48.34% melee crit - hit (occurred around 1400 attacks)

48.70% melee crit - no hit so far (took it to 4500 without a hit then got bored but seems enough to prove the point)

With that hit occurring with 48.34% melee crit it seems pretty safe to say that melee crit depression is indeed 4.8% for shaman just as it is for Rogues.

I thought it was 6.5% dodge not 5.25%?

edit~~ nvm your orc.

Share this post


Link to post
Share on other sites

As it looks like after Rouncer's tests we can come to a conclusion on the topic (and just to make sure I'm understanding all the findings completely) our crit cap can be simplified to

80.8% - miss chance % - Dodge chance % (assumed to be negated by expertise in most cases) = crit cap

OR

80.8% - (27% - Paperdoll hit chance%) - (6.5% - expertise dodge negation%) = crit cap

I know people made a couple scripts to calculate room under crit cap based on your paperdoll hit and crit chances but I'm not sure how outdated those are, can someone better at writing scripts than I make one with these numbers?

Share this post


Link to post
Share on other sites

I know people made a couple scripts to calculate room under crit cap based on your paperdoll hit and crit chances but I'm not sure how outdated those are, can someone better at writing scripts than I make one with these numbers?

I think it would look something like this - modified from the first page, automatically subtracted the glance chance and added the suppressed 4.8:


/script DEFAULT_CHAT_FRAME:AddMessage("Room below crit cap: "..80.8-(27-(GetCombatRatingBonus(6)+6))-(GetCritChance()+3))

Includes the 3% crit debuff from ToW/retbuff#100

Share this post


Link to post
Share on other sites

You would need to run that script while fully raid buffed (Kings, LotP, SoE) and then factor in the 9% of ED on top of that correct?

Share this post


Link to post
Share on other sites
You would need to run that script while fully raid buffed (Kings, LotP, SoE) and then factor in the 9% of ED on top of that correct?

Yes. It checks your current state.

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.

  • Recently Browsing   0 members

    No registered users viewing this page.