Theck's MATLAB TPS analysis - WotLK/3.x

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Finkum » Mon Jan 04, 2010 8:37 pm

Maybe the Libram of Resurgence (+141 spellpower to Consecrate) would be a better choice than the Libram of Reciprocation when you're going to be using SoC?

The proc uptime for the Libram of Reciprocation is estimated on Wowhead to be around 20% in the best case (when judging every 8 seconds), so that's effectively 35 crit rating vs +~5 damage per tick of Consecrate per mob.

Particularly if you end up pulling multiple groups of mobs at a time and thus have 6+ mobs in your Consecrate, I suspect the Libram of Resurgence would be a better choice (perhaps Theck could provide a more definitive answer!)
Finkum
 
Posts: 44
Joined: Tue Dec 22, 2009 9:39 pm

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby lythac » Tue Jan 05, 2010 5:21 am

Theckd wrote:This means that as far as gearing goes, for AoE trash clears you should aim to hit the expertise soft cap, and then the melee hit cap. Keep in mind that these caps are lower for trash - it's 6% melee hit and 6% dodge/parry, which is 197 hit rating and 24 expertise.


Taking hit/expertise cap...

Libram of Reciprocation - Proc chance: 15%

Uptime 9s cast = 1-(0.85)^10/9 = 16.52%
Uptime 8s cast = 1-(0.85)^10/8 = 18.38%

Crit rating 9s cast = 28.58
Crit rating 8s cast = 31.80

Libram of Divine Judgement - Proc chance: 40%

Uptime 9s cast = 1-(0.6)^10/9 = 43.31%
Uptime 8s cast = 1-(0.6)^10/8 = 47.19%

AP 9s cast = 86.62
AP 8s cast = 94.38

theckhd wrote:Next let's look at stats.

Setup:
Rotation: 969
Seals: SoC
Glyphs: none
Talents: 1V+3C
Gear: T10 Progression set
File: calc_stat_tps.m
Number of targets: 1-10

Image


Progression gearset rather than lolheroics, so should undervalue crit and AP librams. Still -

3 targets DPS
Libram of Reciprocation ~21
Libram of Divine Judgement ~60
Libram of Resurgence ~17

10 targets DPS
Libram of Reciprocation ~21
Libram of Divine Judgement ~85
Libram of Resurgence ~56

Very fudged numbers, but meh. WTB libram DPS increase graph!

Edit - wasn't looking at threat as most of the libram talk concerns Heroics and DPS.
Ryshad / Lythac of <Heretic> Nagrand-EU
User avatar
lythac
Moderator
 
Posts: 2685
Joined: Wed Sep 24, 2008 8:10 am

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby tlitp » Tue Jan 05, 2010 6:10 am

Iselian wrote:
theckhd wrote:Test #1: Libram of Defiance
My model that predicts 98% uptime assumes that the buff can trigger off of any HotR cast, regardless of whether it is successful. However, I haven't seen any test showing that the thing procs off of missed HotR's. It should, since the item text says "every time you use your HotR[/b]. But people have reported uptimes of around 85%, which would agree better with a model where unsuccessful HotR's don't proc the buff.

To test this:
1. Strip down to a weapon (preferably one with no hit, a white vendor weapon would be the best) and the libram.
2. Cast HotR on a training dummy until you see the buff proc off of missed/parried/deflected HotR. Stop once you've got 5-10 or so examples of such an event, or if you go 30 minutes without seeing a single buff proc off of an unsuccessful HotR (unlikely).
3. Upload the combat log text file to this thread so that we can confirm the result.

I did this one awhile back, though my combat log halfway borked. I did get some "succesful" deflect/dodge/misses that resulted in evasion. First, at [02:47:09.861], second at [02:47:32.152], last at [02:49:52.056]. Log to be browsed can be found here.

As suspected, LoD's uptime appears to be computed per cast, not per successfully landed trigger.
Iselian wrote:
theckhd wrote:Test #3: Blade Ward - haste normalization
Presumably, Blade Ward works exactly like Berserking and Mongoose - base PPM of 1, and not normalized with haste (i.e. haste will increase the effective uptime). But as far as I know there hasn't been a thorough test of this issue.

To test this:
1. Ideally, you'd need two gear sets - one with lots of haste, and one without. Having windfury available would help as well. The two sets should be roughly the same as far as hit and expertise (don't forget to include the shaman buff). If you can get hit-capped and expertise soft-capped in both sets, and attack from behind, that would be perfect. Alternatively, if you can build a set with high haste but absolutely no hit or expertise, you could do one test naked and one test with the haste set.
2. In the "low-haste" set, melee the dummy for about 30 minutes. Then put on the "high-haste" set and repeat.
3. Upload the results.

First, non-hasted set found here. After about 20 minutes, there were 20 procs. Speed of about 1.6 (used Shiver).

Second, hasted set can be found here. I had roughly 170 haste rating and an improved windfury from an enhancement shaman (horde, no hit buff, didn't need it). Speed of about 1.25 (still used Shiver). Ignore the random hunter in the log, someone else decided to bum the dummy for a bit.

If more testing is needed, let me know. I was a little short on time this round, but could probably find an hour per next week.

In the first test, there is a (successfully landed) trigger rate (discrete, averaged) of :
tr1 = (431+1+93+164)/(20*60+44) = 689/1244
The estimated ppm (assuming haste normalization) :
ppm1 = (1-nthroot(1-ou1,10*tr1))/p, where p=1.6/60 (with Shiver being used) and ou1=0.142 (the observed uptime).
ppm1 ~ 1.02

In the second test, the trigger rate is :
tr2 = (675+128+242)/(21*60+54) = 1040/1314
Assuming haste normalization again :
ppm2 = (1-nthroot(1-ou2,10*tr2))/p, ou2=0.158.
ppm2 ~ 0.8

With the estimated PPMs being disjoint, a FET on the two data sets (first set : 20 procs in 689 attempts, second test : 25 procs in 1040 attempts) gives a two-tailed p-value of ~0.44. Which means that there isn't any (measurable) difference in the proc chance, which means that the initial assumption does not hold. Conclusion : BW is 1 ppm, not haste-normalized.

Thanks for testing, Iselian.
User avatar
tlitp
 
Posts: 556
Joined: Mon Jul 27, 2009 3:25 pm

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Awyndel » Thu Jan 07, 2010 1:37 am

This prolly has been answered before, but I just wanna make sure, especially for 3.3.

In a progression gearset, in ICC, wich enchant would give more threat, and wich one would give more average avoidance. Bladeward or mongoose? Ow and wich one does more dps?

Slightly off topic on maintankadin maybe, but how is this for warriors, I see all of them running around with bladeward nowadays.
User avatar
Awyndel
 
Posts: 672
Joined: Sat Feb 14, 2009 8:49 am
Location: The Netherlands

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Kelaan » Thu Jan 07, 2010 11:28 am

I'm inclined to say mongoose: blade ward won't help threat, whereas mongoose will. Avoidance-wise, I was under the impression that blade ward was still crap? Maybe not, but I believe we have a thread here on it.
User avatar
Kelaan
 
Posts: 4037
Joined: Thu Jan 03, 2008 12:01 pm

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby theckhd » Thu Jan 07, 2010 12:31 pm

I've been meaning to re-run the enchant, glyph, and other simulations, but I just haven't had time. Also, before I run the enchant sim, I'd like to get a confirmation on whether ShoR procs Mongoose and other ppm-based effects. This should be pretty easy to test if someone has 15 minutes to ShoR a dummy with a mongoose-enchanted weapon. If you see even one proc, we know it works.

If someone can run this test for me, I'll get the other simulations posted today or early tomorrow.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7956
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Serv » Thu Jan 07, 2010 1:07 pm

I just tested with /cast ShoR /stopcasting plus positioning sideways so no melee swing would connect. After around 15-20 minutes I got no mongoose procc, but ShoR refreshes Divine Plea. I don't know if this is considered 'contrary'

/edit:

I just tried it again with SealofCleave loaded. It didn't procc. But the strange thing i noticed, was while standing clearly behind the dummy so I only got two damage sources in my logs (ShoR/SoC) one of my seals got parried.
This wasn't the SoC procc from the level 80 dummy however. The seal bounced to the boss dummy and was parried there. At this point I can only assume, that the positioning only counts for the target you directly attack.

Code: Select all
1/7 21:15:11.883  SWING_DAMAGE,0x028000000116415B,"Kharrasch",0x518,0xF1300079AA00180D,"Heroische Trainingsattrappe",0xa28,838,0,1,0,0,0,1,nil,nil
1/7 21:15:12.145  SPELL_CAST_SUCCESS,0x0280000001A7B53F,"Servellia",0x511,0xF1300079A8001810,"Übungsattrappe des Großmeisters",0x10a28,61411,"Schild der Rechtschaffenheit",0x2
1/7 21:15:12.145  SPELL_CAST_SUCCESS,0x0280000001A7B53F,"Servellia",0x511,0xF1300079A8001810,"Übungsattrappe des Großmeisters",0x10a28,20424,"Siegel des Befehls",0x2
1/7 21:15:12.145  SPELL_MISSED,0x0280000001A7B53F,"Servellia",0x511,0xF1300079AA00180D,"Heroische Trainingsattrappe",0xa28,20424,"Siegel des Befehls",0x2,PARRY


<!---- I play with a german client.
Siegel des Befehls = SoC
Heroische Trainingsattrappe = Lvl 83 dummy
Übungsattrappe des Großmeisters = lvl 80 dummy
Image
Thanks to LAVE for the sig

www.rent-a-tank.net ...:::... www.nil-gilde.net
Serv
 
Posts: 228
Joined: Fri Jun 05, 2009 8:50 am

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby theckhd » Thu Jan 07, 2010 1:47 pm

Serv wrote:I just tested with /cast ShoR /stopcasting plus positioning sideways so no melee swing would connect. After around 15-20 minutes I got no mongoose procc, but ShoR refreshes Divine Plea. I don't know if this is considered 'contrary'

For an ability that used to be a ranged attack with melee range, which is now a melee attack that follows most of the conventions of a melee attack but doesn't stack HV or proc enchants, I'm not sure what we'd consider 'contrary' anymore.

In any event, it seems ShoR probably doesn't proc PPM based effects, so I'll start cranking out the code with that in mind.

This wasn't the SoC procc from the level 80 dummy however. The seal bounced to the boss dummy and was parried there. At this point I can only assume, that the positioning only counts for the target you directly attack.

This is consistent with what we know about SoC already - it can cleave to things behind you, so it's not surprising it can cleave anything within a fixed (and apparently quite large) radius despite facing or melee range restrictions.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7956
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Serv » Thu Jan 07, 2010 2:25 pm

I don't know if you and I were talking about differnt things conscerning SoC. I knew it could jump to targets behind me, but I didn't know that it could be parried by the 2nd/3rd target while it definitly couldn't be parried by my main target, which I was attacking from behind
Image
Thanks to LAVE for the sig

www.rent-a-tank.net ...:::... www.nil-gilde.net
Serv
 
Posts: 228
Joined: Fri Jun 05, 2009 8:50 am

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby theckhd » Thu Jan 07, 2010 6:29 pm

Serv wrote:I don't know if you and I were talking about differnt things conscerning SoC. I knew it could jump to targets behind me, but I didn't know that it could be parried by the 2nd/3rd target while it definitly couldn't be parried by my main target, which I was attacking from behind

Oh, I thought you were surprised by the cleave distance. I think it's standard practice that the extra attacks from "cleave" abilities are considered "attacks from the front."
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7956
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby tlitp » Fri Jan 08, 2010 4:15 am

In preparation for the next iterations of the sims, there are some methodology-related questions that need to be answered :
weapon gems*
- module : gear, calc_weaps_speed
- current handling : full-out stamina in gear (no matching), always matching in calc_weaps_speed (red : stam/agi, yellow : stam/def, blue : stam)
- proposal :
  1. full-out stamina, no matching
  2. balanced (exp to 24 in red > hit to 6% physical in yellow > strength in red; mixed colors are preferred), matching when possible (else full stamina)
  3. full-out threat (exp to 24 > hit to 6% physical > strength), no matching
weapon enchants
- module : gear, calc_weaps_speed
- current handling : accuracy by default
- proposal :
  1. "old" default (accuracy)
  2. balanced (mongoose)
  3. full-out threat (berserking)
reference gear
- module : calc_weaps_speed
- current handling : hit and expertise forced to 0
- proposal :
  1. hit/exp forced to 0** (advantage : "fair" reference for regular weapons, disadvantage : proc-based weapons are undervalued)
  2. default hit/exp from gear (advantage : "fair" with respect to proc-based weapons, disadvantage : regular weapons that grant hit/exp can be undervalued)
What do people consider to be the best all-around setup ?


* Matching a socket with respect to "the bigger picture" and/or socket bonus, like a human would do, is definitely tempting but very cumbersome to implement in a generic sim. The keyword here is "generic".
** Do note that setting hit/exp to 0 leads to some (non-necessarily threat-related) subtle effects, like changes in PHR factors, uptimes of dynamic effects, TTL-based metrics. In particular, this restriction will affect rather significantly the uptime of proc-based weapon enchants.
User avatar
tlitp
 
Posts: 556
Joined: Mon Jul 27, 2009 3:25 pm

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Thels » Fri Jan 08, 2010 5:39 am

No idea if it's the best option, but I currently gem Agi/Sta or Hit/Sta, if:

- There is only a single nonblue/nonprismatic socket.
- The socket bonus is 9 or 12 stamina.

In all other cases, I use 30 stamina.

I haven't come to the situation yet where there were no red 9 or 12 stamina sockets (except during the naxx days where I had a set of gear that was completely void of red sockets). If that case would arise, I'd probably do one of the following in order of highest to lowest prio:

- +10 all stats in a yellow 12 sta socket
- agi/sta in a red 6 sta socket
- +10 all stats in a yellow 9 or 6 sta socket
- agi/sta in a red socket that gives +hit
- agi/sta in a red socket that gives avoidance
- agi/sta in any one socket

As for the Hit/Exp settings, would it be possible to make two analizations? One with 0 Hit/Exp and one with both Hit and Exp softcapped? (Having Hit and/or Exp hardcapped seems unrealistic enough). At least making this analization once would allow people to judge the difference at a later time.

I really have no opinion about the weapon enchant. I have 4 weapons, and enchanted them to match the weapon's stats. Shiver has Blood Draining, The ToC10 blade has Potency, the Axe from icc5 has Agi and the Sword from icc5 has Accuracy.
ImageImage
User avatar
Thels
 
Posts: 1295
Joined: Tue May 15, 2007 7:30 am
Location: The Netherlands

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Chicken » Fri Jan 08, 2010 3:21 pm

theckhd wrote:
Serv wrote:I just tested with /cast ShoR /stopcasting plus positioning sideways so no melee swing would connect. After around 15-20 minutes I got no mongoose procc, but ShoR refreshes Divine Plea. I don't know if this is considered 'contrary'

For an ability that used to be a ranged attack with melee range, which is now a melee attack that follows most of the conventions of a melee attack but doesn't stack HV or proc enchants, I'm not sure what we'd consider 'contrary' anymore.
Well it not proccing HV is intended behavior to some extent; it does give the stack-based seal damage still after all. Crusader Strike and Divine Storm function the same way, it's actually meant to be the case that only Hammer of the Righteous is an exception as a melee special that stacks HV. Just check the 3.2.0 patch notes.

Of course it's definitely true that it's hard to tell with Paladin mechanics what is contrary or not, though at least Shield of Righteousness being melee now makes more sense than it being ranged for no discernible reason.
Image
User avatar
Chicken
 
Posts: 1597
Joined: Fri Jun 26, 2009 2:19 pm

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby Nadir » Sat Jan 09, 2010 5:31 am

Theck, it appears that a recent nerf to Bryntroll for Retribution Paladins may have some impact upon Protection Paladin mechanics.

Specifically, refreshing and applying Holy Vengeance no longer acts as an opportunity to proc "on hit" abilities. This includes Drain Life from Bryntroll (the intended target of a recent nerf), as well as PPM weapon enchants such as Berserking and Mongoose (collateral damage).

With respect to Protection Paladins this may impact the value of avoidance value expertise (if refreshing HV isn't an "on hit" attack anymore, can it be parried/dodged?) and I expect that the up time on Mongoose will drop substantially (re: Side Calculation - Mongoose uptime).
Image
Nadir
 
Posts: 266
Joined: Mon May 04, 2009 8:37 am

Re: Theck's MATLAB TPS analysis (A Jonesy derivative work)

Postby tlitp » Sat Jan 09, 2010 6:14 am

theckhd wrote:Also, before I run the enchant sim, I'd like to get a confirmation on whether ShoR procs Mongoose and other ppm-based effects. This should be pretty easy to test if someone has 15 minutes to ShoR a dummy with a mongoose-enchanted weapon. If you see even one proc, we know it works.

To expand on Theck's request, there are a couple of things we'd like to know before running the sims again :
  1. Does ShoR proc any seals if running with no points in Judgements of the Just ? If so, which of them and how many seal procs per ShoR ? Does ShoR proc 2 SoR if running with points in JotJ ?
  2. Does Judgement proc any seals if running with no points in JotJ ? If so, which of them and how many seal procs per Judgement ? Does Judgement still proc 2 SoR if running with points in JotJ ? (tested rather extensively before, but we'd like a confirmation that the known behaviour is still in place in the 3.3 era)
  3. Is the cleave effect of SoComm capable of producing multiple procs of proc-based enchants, as HotR is ?
  4. Are Reckoning procs capable of seal procs ? If so, which seals and how many per Reck proc ? (again, various old experimental data are available - what we want is retesting in 3.3)

Methodology :
  1. Some target dummy smacking, making sure that only ShoR is used.
  2. The same, this time with J.
  3. To test this one you'll need to use the level 60 dummies, making sure that SoC cleaves consistently. Use only autoattacks, upload a combat log of 30+ mins, do not forget to mention the base weapon "speed" of your character.
  4. Request a duel from a rogue, ask him/her to grab a pair of fast daggers (these ones are ideal due to their low output). Let him/her go wild on you. :P

Thank you.


edit : With Nadir's observation in mind, we'd also need insight on this matter. Some testing sessions will be more than welcomed.
User avatar
tlitp
 
Posts: 556
Joined: Mon Jul 27, 2009 3:25 pm

PreviousNext

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest

Who is online

In total there is 1 user online :: 0 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: No registered users and 1 guest