Remove Advertisements

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

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Re: Theck's MATLAB TPS analysis

Postby Yelena » Thu Feb 18, 2010 2:46 pm

I can't upload a combat log from work, but from my observations HV ticks do not proc it. HV applications do proc it though, as I've seen several double procs in the combat log, without a Vindication proc overlapping it.

Code: Select all
15:39:32> [Yelena's] melee swing hits [Heroic Training Dummy] for 674 Physical.
15:39:32> [Heroic Training Dummy] is afflicted by [Yelena's] Holy Vengeance.
15:39:33> [Yelena] gains [Yelena's] Blessing of Light.
15:39:33> [Yelena's] Blessing of Light is refreshed on [Yelena].


Edit: This one was my favorite.

Code: Select all
15:54:14> [Yelena] casts Hammer of the Righteous at [Heroic Training Dummy].
15:54:14> [Yelena's] Hammer of the Righteous hits [Heroic Training Dummy] for 2383 Holy.(265 Resisted)
15:54:15> [Grandmaster's Training Dummy] is afflicted by [Yelena's] Holy Vengeance.
15:54:15> [Master's Training Dummy] is afflicted by [Yelena's] Holy Vengeance.
15:54:15> [Heroic Training Dummy] is afflicted by [Yelena's] Holy Vengeance.
15:54:15> [Yelena] gains [Yelena's] Blessing of Light.
15:54:15> [Yelena's] Hammer of the Righteous hits [Master's Training Dummy] for 1 Holy.(2646 Overkill)
15:54:15> [Yelena's] Hammer of the Righteous hits [Grandmaster's Training Dummy] for 1 Holy.(2647 Overkill)
15:54:15> [Yelena's] Blessing of Light is refreshed on [Yelena].
15:54:15> [Yelena's] Blessing of Light is refreshed on [Yelena].
15:54:15> [Yelena's] Blessing of Light is refreshed on [Yelena].


Seal of Vengeance, no other buffs. Angled such that I couldn't swing my weapon on the target.
Valleri - Frostwolf
Yelena
 
Posts: 308
Joined: Sat Jun 27, 2009 4:54 am
Location: Behind the Eight Ball

Re: Theck's MATLAB TPS analysis

Postby tlitp » Thu Feb 18, 2010 10:46 pm

theckhd wrote:Now for the more statistical tests. Note that I've changed the spec tlitp asked for - he said that the spec should have JotJ, but the 42/7 spec he linked doesn't go deep enough in the tree to pick it up. I'm assuming he meant that you should pick it up (and that the 42/7 was just to guide you away from picking certain prot talents), but I'd hate for someone to have wasted their time doing a bunch of testing with that exact 42/7 spec. :P

*rolls on the floor laughing hysterically*
Listen to Theck, please. Sorry for the confusion.

cypher wrote:The results are in. 500 melee hits on Coprous with a Tomahawk yielded 3 dodges and 2 parries at 22 expertise (5.5%) and none of either at 24 expertise (6% exactly):

Image

Image

While my sample size is admittedly small, I think it's safe (or safer, rather) to assume that the hardcap vs. level 82 mobs is 24 expertise and not 22 as Satrina claimed.

As per Theck's request, posted this on the public forum. There you have it, several theorycrafting/information sources have the wrong numbers (this includes tankspot, wowwiki, spreadsheets). All hail cypher, which nailed the conclusive evidence.
User avatar
tlitp
 
Posts: 554
Joined: Mon Jul 27, 2009 3:25 pm

Re: Theck's MATLAB TPS analysis

Postby cypher » Fri Feb 19, 2010 12:29 am

You beat me to it, tlitp! :)

Latest findings: After thousands of white hits (and many, many Perfect Accurate Huge Citrines), 23.93 parry skill (147 expertise rating +6 expertise skill from talents) still yields parries.

Image

This means that, assuming both 3/3 Combat Expertise and no racial bonuses, 148 expertise rating is the hard-cap vs. level 82 mobs.

Update your trash/heroic sets accordingly.
User avatar
cypher
 
Posts: 14
Joined: Tue Dec 02, 2008 6:38 pm

Re: Theck's MATLAB TPS analysis

Postby theckhd » Fri Feb 19, 2010 6:35 am

So then, presumably the following table is correct (I'll update my Attack Table article appropriately):
Code: Select all
    Target        Miss %         Dodge%   Parry%
    Level     Melee     Spell
    80          5.0      4        5.0      5.0
    81          5.5      5        5.5      5.5
    82          6.0      6        6.0      6.0
    83          8.0     17        6.5     14.0


Note that the dodge/parry values for a level 81 mob have not been experimentally verified yet. I think everything else has been.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Call to Arms 5.x, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7655
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis

Postby theckhd » Fri Feb 19, 2010 8:11 am

Yelena wrote:I can't upload a combat log from work, but from my observations HV ticks do not proc it. HV applications do proc it though, as I've seen several double procs in the combat log, without a Vindication proc overlapping it.

Good, good. This is what I guessed, and exactly the sort of parse data needed to confirm it.

Now we can start formulating some tests to figure out which of the "Unverified" ones do and don't act as triggers. Returning to steadypal's data sets for a moment:
theckhd wrote:Just checking some numbers quick:
SoComm log:
631 melee hits, 111 crits, 255 glances, 997 total swings.
897 SoCom hits, 99 crits, 996 total
224 Vindication applications (and 18 misses)
397 Blessing of Light casts (Last Word buff)

SoV log:
633 melee hits, 114 crits, 250 glances, 997 total swings.
862 SoV hits, 134 crits, 996 total
234 Vindication applications (11 misses)
997 Blood Corruption applications
599 Blood Corruption ticks
1099 Blessing of Light casts


Wowhead lists the Last Word proc as a 37%. I'm not sure if that is post-buff or pre-buff though. Just playing around with some numbers from the first test:

997*0.37 = 369
224*0.37 = 83
(997+224)*0.37 = 452
(997+996)*0.37 = 737
(997+996+224)*0.37 = 820

397/997 = 39.8%
397/(997+224) = 32.5%
397/(997+996) = 19.9%
397/(997+996+224) = 17.9%

for the more statistically inclined:
using binofit
binofit(397,997) = 39.82%, 95% Confidence Interval (CI) goes from 36.76% to 42.93% (contains 37%)
binofit(397,997+224) = 32.51%, 95% CI from 29.89% to 35.22% (does not contain 37%)
binofit(397,997+996) = 19.92%, 95% CI from 18.19% to 21.74% (does not contain 37%)

Conclusions:
  1. So it's feasible that the proc is indeed 37% and was only triggering off of melee swings in the first test. It's also possible (though less likely, under a 5% chance) that it procced off vindication as well.
  2. There's no conclusive evidence here that the proc rate is any higher than 40-43%, which indicates that the 37% value is probably post-buff (i.e. the live proc rate)
  3. It almost certainly does not proc off of SoCom, as we would have expected nearly double the number of procs we saw.
  4. If we believe the proc rate can't be any lower than 30%, this is enough to rule out SoCom procs.
  5. It's not enough to convince me that Vindication isn't a trigger, though the statistical analysis says it's pretty unlikely.

Now some numbers from the second test:

234*0.37 = 87

1099/(997+996) = 55.1%
1099/(997+996+234) = 49.4%
1099/(997+996+997) = 36.8%
1099/(997+996+997+234) = 34.1%

binofit(1099,997+996) = 55.14%, 95% CI goes from 52.93% to 57.34%
binofit(1099,997+996+234) = 49.35%, CI from 47.25% to 51.45%
binofit(1099,997+996+997) = 36.76%, CI from 35.02% to 38.51% (contains 37%)
binofit(1099,997+996+997+234) = 34.09%, CI from 32.45% to 35.75%

Conclusions:
  1. The first test suggested the proc rate was lower than 40%. However, melee attacks and HV applications alone aren't enough to supply the large number of procs in this data set if that's truly the case.
  2. Vindication alone isn't enough to bring the results back in line with the first data set.
  3. Melee + HV applications + SoV procs brings the observed proc rate very close to the expected value.
  4. M+HV+SoV+Vindication still seems a bit low - note the confidence interval for this doesn't contain 37%
  5. This is enough to confirm that SoV procs do trigger Last Word

This is really bizarre, I hadn't expected the two seals to act differently (despite recognizing that as a possibility earlier in the thread).

I'm now reasonably confident that Vindication doesn't proc it either. I'd love to see another 1k-2k swings with no seal up, if steadypal has an hour to AFK at the dummy. That should be enough to get convincing data on Vindication, I think.


Just to get my thoughts together, so far testing has shown that Last Word's proc has the following triggers:

Triggers
Melee swings
ShoR
HV applications
Hammer of the Righteous
Seal of Vengeance procs

Not Triggers
HV ticks (DoT)
Seal of Command procs

Unverified
Vindication (but almost certainly not)
Judgement
Judgement of Light/Wis/Justice applications
Heart of the Crusader

So here's the final few tests that would be required to determine the peculiarities of Judgement:

Test L-5: Completed: Vindication does not trgger LW.
Last Word equipped, hit-capped and expertise-capped for whichever dummy you're using. Vindication in spec, no seal.
  1. Melee the dummy for around 2k swings. No other abilities
This should isolate Vindication and remove it from our list of things to worry about.

Test L-6: Completed: Judgement and JoWis both trigger LW.
Last Word equipped, hit-capped for the dummy. Spec should not contain HotC or RV. Seal of Light.
  1. Judge the dummy from outside of melee range. Repeat ~50-100 times.
We'll look for double-procs in the parse (i.e. cases like Yelena saw where you get two BoL procs from a single Judgement cast). There should be a ~14% chance of this happening on any Judgement, which means we should see 6-7 events on average every 50 Judgements. If we see 0 examples, then Judgement and the debuff can't both independently proc it.

Test L-7: Completed - HotC does not trigger LW, JotJ does.
Last Word equipped, hit-capped for the dummy. Spec should contain HotC. Seal of Light.
  1. Judge the dummy from outside of melee range. Repeat ~50-100 times.
This time we'll look for triple procs. There's about a 5% chance of this per swing if Judgement, JoWis applications, and HotC can all apply the debuff. If the previous test proved that one or the other cannot apply the debuff, then we'll look for double procs instead.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Call to Arms 5.x, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7655
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis

Postby tlitp » Fri Feb 19, 2010 8:21 am

Code: Select all
    Target        Miss %         Dodge%   Parry%   Glancing%
    Level     Melee     Spell
     80        5.0        4        5.0      5.0        6**
     81        5.5*       5*       5.5      5.5       12**
     82        6.0*       6*       6.0      6.0       18**
     83        8.0       17        6.5     14.0       24

*  confirmed by multiple classes
** confirmed by rogues


theckhd wrote:Wowhead lists the Last Word proc as a 37%. I'm not sure if that is post-buff or pre-buff though.

3% (~1 ppm) in the first iteration, 37% (~12 ppm) in the second.
User avatar
tlitp
 
Posts: 554
Joined: Mon Jul 27, 2009 3:25 pm

Re: Theck's MATLAB TPS analysis

Postby theckhd » Fri Feb 19, 2010 8:26 am

tlitp wrote:
theckhd wrote:Wowhead lists the Last Word proc as a 37%. I'm not sure if that is post-buff or pre-buff though.

3% (~1 ppm) in the first iteration, 37% (~12 ppm) in the second.

Yeah, I guessed as much, but I didn't remember the initial proc rate and was too lazy to look it up. It was pretty obvious from steadypal's data set, as well.

Thanks for the glancing numbers, I'll add them to the article.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Call to Arms 5.x, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7655
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis

Postby steadypal » Fri Feb 19, 2010 11:53 am

hit/exp capped, standard prot spec, no seal autoattack, with vindication,,, HMM recount said i had over 1500 autoattacks, wol says 974, wtf, why the huge diff? oh well...

http://www.worldoflogs.com/reports/n5lg ... details/0/


hit capped, seal of light,,, judge of wisdom

http://www.worldoflogs.com/reports/0fzf ... details/0/


now, when i was doing the judgements, i can definately see a delayed second on buff applying, can see when it applies right after judge, and there were times i can see it flicker like it procced twice, as u can see, 100 judgements, and there was 100 buffs applied...


so one more test, same as prior but w/o heart of crusader correct? or do we not want jotj to?
steadypal
 
Posts: 1206
Joined: Sat Mar 15, 2008 12:28 pm

Re: Theck's MATLAB TPS analysis

Postby Gavinas » Fri Feb 19, 2010 12:15 pm

steadypal wrote:hit/exp capped, standard prot spec, no seal autoattack, with vindication,,, HMM recount said i had over 1500 autoattacks, wol says 974, wtf, why the huge diff? oh well...


974+161+381 = 1516

WoLogs lists regular melee hits, melee crits and glancings in different columns. Recount lists them together and gives percentages for each individual type of hit.
Gavinas
 
Posts: 51
Joined: Fri Dec 25, 2009 9:03 pm

Re: Theck's MATLAB TPS analysis

Postby Yelena » Fri Feb 19, 2010 12:22 pm

steadypal wrote:now, when i was doing the judgements, i can definately see a delayed second on buff applying, can see when it applies right after judge, and there were times i can see it flicker like it procced twice, as u can see, 100 judgements, and there was 100 buffs applied...


so one more test, same as prior but w/o heart of crusader correct? or do we not want jotj to?

I've been doing more testing at work, and I'm definitely noticing the delayed additional proc, which coincides with Judgements of the Just being applied.

A few examples:
Code: Select all
12:51:51> [Yelena] casts Judgement of Wisdom at [Heroic Training Dummy].
12:51:51> [Heroic Training Dummy] is afflicted by [Yelena's] Judgement of Wisdom.
12:51:51> [Yelena's] Judgement hits [Heroic Training Dummy] for 1019 Holy.
12:51:51> [Heroic Training Dummy] is afflicted by [Yelena's] Heart of the Crusader.
12:51:51> [Yelena] gains [Yelena's] Blessing of Light.
12:51:51> [Yelena's] Blessing of Light is refreshed on [Yelena].
12:51:52> [Heroic Training Dummy] is afflicted by [Yelena's] Judgements of the Just.
12:51:52> [Yelena's] Blessing of Light is refreshed on [Yelena].


Code: Select all
12:54:57> [Yelena] casts Judgement of Wisdom at [Heroic Training Dummy].
12:54:57> [Heroic Training Dummy] is afflicted by [Yelena's] Judgement of Wisdom.
12:54:57> [Yelena's] Judgement hits [Heroic Training Dummy] for 2038 Holy.(Critical)
12:54:57> [Heroic Training Dummy] is afflicted by [Yelena's] Heart of the Crusader.
12:54:58> [Yelena's] Seal of Light heals [Yelena] for 0.(723 Overhealed)
12:54:58> [Yelena] gains [Yelena's] Blessing of Light.
12:54:58> [Heroic Training Dummy] is afflicted by [Yelena's] Judgements of the Just.
12:54:59> [Yelena's] Blessing of Light is refreshed on [Yelena].


Code: Select all
13:10:37> [Yelena] casts Judgement of Wisdom at [Heroic Training Dummy].
13:10:37> [Heroic Training Dummy] is afflicted by [Yelena's] Judgement of Wisdom.
13:10:37> [Yelena's] Judgement hits [Heroic Training Dummy] for 1019 Holy.
13:10:37> [Heroic Training Dummy] is afflicted by [Yelena's] Heart of the Crusader.
13:10:38> [Yelena] gains [Yelena's] Blessing of Light.
13:10:38> [Yelena's] Blessing of Light is refreshed on [Yelena].
13:10:38> [Heroic Training Dummy] is afflicted by [Yelena's] Judgements of the Just.
13:10:38> [Yelena] gains 88 Mana from [Yelena's] Judgement of Wisdom.
13:10:38> [Yelena's] Blessing of Light is refreshed on [Yelena].
Valleri - Frostwolf
Yelena
 
Posts: 308
Joined: Sat Jun 27, 2009 4:54 am
Location: Behind the Eight Ball

Re: Theck's MATLAB TPS analysis

Postby theckhd » Fri Feb 19, 2010 1:12 pm

steadypal wrote:hit/exp capped, standard prot spec, no seal autoattack, with vindication,,, HMM recount said i had over 1500 autoattacks, wol says 974, wtf, why the huge diff? oh well...

http://www.worldoflogs.com/reports/n5lg ... details/0/


974 hits, 161 crits, and 381 glances, for a total of 1516 melee attacks.
357 Vindication procs
556 Blessing of Light procs

binofit(556,1516) = 36.68%, CI from 34.24% to 39.16%
binofit(556,1516+357) = 29.68%, CI from 27.62% to 31.81%

Since we're now certain that 37% is the current proc rate, I'm going to say that this data conclusively rules out Vindication as a potential proc source.

steadypal wrote:hit capped, seal of light,,, judge of wisdom

http://www.worldoflogs.com/reports/0fzf ... details/0/


now, when i was doing the judgements, i can definately see a delayed second on buff applying, can see when it applies right after judge, and there were times i can see it flicker like it procced twice, as u can see, 100 judgements, and there was 100 buffs applied...


so one more test, same as prior but w/o heart of crusader correct? or do we not want jotj to?

I'd prefer to have neither JotJ or HotC, if possible. A ret spec that skips HotC would be perfect.

But first, let's take a look at your log. Here's the full text after filtering for only Judgement and Blessing of Light:
Code: Select all
[10:32:58.907] Steadypal Judgement Heroic Training Dummy 1111
[10:33:10.145] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:33:19.800] Steadypal Judgement Heroic Training Dummy *2221*
[10:33:29.038] Steadypal Judgement Heroic Training Dummy *2221*
[10:33:29.430] Steadypal gains Blessing of Light from Steadypal
[10:33:29.430] Steadypal's Blessing of Light is refreshed by Steadypal
[10:33:38.305] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:33:38.696] Steadypal's Blessing of Light is refreshed by Steadypal
[10:33:47.515] Steadypal Judgement Heroic Training Dummy *2120* (R: 118)
[10:33:47.923] Steadypal's Blessing of Light is refreshed by Steadypal
[10:33:57.174] Steadypal Judgement Heroic Training Dummy 1178
[10:33:57.584] Steadypal's Blessing of Light is refreshed by Steadypal
[10:33:57.584] Steadypal's Blessing of Light is refreshed by Steadypal
[10:34:06.416] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:34:07.569] Steadypal's Blessing of Light fades
[10:34:15.658] Steadypal Judgement Heroic Training Dummy 888 (R: 222)
[10:34:24.922] Steadypal Judgement Heroic Training Dummy 1111
[10:34:25.318] Steadypal gains Blessing of Light from Steadypal
[10:34:26.127] Steadypal's Blessing of Light is refreshed by Steadypal
[10:34:34.167] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:34:36.167] Steadypal's Blessing of Light fades
[10:34:43.813] Steadypal Judgement Heroic Training Dummy 1110
[10:34:53.500] Steadypal Judgement Heroic Training Dummy 888 (R: 222)
[10:34:53.895] Steadypal gains Blessing of Light from Steadypal
[10:34:54.680] Steadypal's Blessing of Light is refreshed by Steadypal
[10:35:02.351] Steadypal Judgement Heroic Training Dummy 1178
[10:35:02.971] Steadypal's Blessing of Light is refreshed by Steadypal
[10:35:11.937] Steadypal Judgement Heroic Training Dummy *2356*
[10:35:12.332] Steadypal's Blessing of Light is refreshed by Steadypal
[10:35:20.780] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:35:21.184] Steadypal's Blessing of Light is refreshed by Steadypal
[10:35:21.992] Steadypal's Blessing of Light is refreshed by Steadypal
[10:35:29.921] Steadypal Judgement Heroic Training Dummy *2356*
[10:35:30.034] Steadypal's Blessing of Light is refreshed by Steadypal
[10:35:39.303] Steadypal Judgement Heroic Training Dummy 1178
[10:35:40.031] Steadypal's Blessing of Light fades
[10:35:48.516] Steadypal Judgement Heroic Training Dummy 1111
[10:35:48.937] Steadypal gains Blessing of Light from Steadypal
[10:35:57.760] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:35:58.150] Steadypal's Blessing of Light is refreshed by Steadypal
[10:36:07.035] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:36:07.392] Steadypal's Blessing of Light is refreshed by Steadypal
[10:36:07.392] Steadypal's Blessing of Light is refreshed by Steadypal
[10:36:16.646] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:36:17.373] Steadypal's Blessing of Light fades
[10:36:25.516] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:36:25.897] Steadypal gains Blessing of Light from Steadypal
[10:36:34.769] Steadypal Judgement Heroic Training Dummy 1178
[10:36:35.168] Steadypal's Blessing of Light is refreshed by Steadypal
[10:36:44.400] Steadypal Judgement Heroic Training Dummy *1884* (R: 236)
[10:36:44.836] Steadypal's Blessing of Light is refreshed by Steadypal
[10:36:45.603] Steadypal's Blessing of Light is refreshed by Steadypal
[10:36:53.652] Steadypal Judgement Heroic Training Dummy *2120* (R: 118)
[10:36:55.589] Steadypal's Blessing of Light fades
[10:37:03.407] Steadypal Judgement Heroic Training Dummy 1111
[10:37:12.530] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:37:12.952] Steadypal gains Blessing of Light from Steadypal
[10:37:21.781] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:37:22.178] Steadypal's Blessing of Light is refreshed by Steadypal
[10:37:31.424] Steadypal Judgement Heroic Training Dummy 1178
[10:37:31.833] Steadypal's Blessing of Light is refreshed by Steadypal
[10:37:31.833] Steadypal's Blessing of Light is refreshed by Steadypal
[10:37:40.271] Steadypal Judgement Heroic Training Dummy 1178
[10:37:40.666] Steadypal's Blessing of Light is refreshed by Steadypal
[10:37:49.520] Steadypal Judgement Heroic Training Dummy 1178
[10:37:50.647] Steadypal's Blessing of Light fades
[10:37:58.761] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:37:59.171] Steadypal gains Blessing of Light from Steadypal
[10:38:09.150] Steadypal's Blessing of Light fades
[10:38:09.209] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:38:18.454] Steadypal Judgement Heroic Training Dummy *2221*
[10:38:18.863] Steadypal gains Blessing of Light from Steadypal
[10:38:18.863] Steadypal's Blessing of Light is refreshed by Steadypal
[10:38:27.695] Steadypal Judgement Heroic Training Dummy 1178
[10:38:28.854] Steadypal's Blessing of Light fades
[10:38:28.902] Steadypal gains Blessing of Light from Steadypal
[10:38:37.759] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:38:38.146] Steadypal's Blessing of Light is refreshed by Steadypal
[10:38:47.384] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:38:47.802] Steadypal's Blessing of Light is refreshed by Steadypal
[10:38:56.640] Steadypal Judgement Heroic Training Dummy *2120* (R: 118)
[10:38:57.069] Steadypal's Blessing of Light is refreshed by Steadypal
[10:38:57.842] Steadypal's Blessing of Light is refreshed by Steadypal
[10:39:05.882] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:39:06.230] Steadypal's Blessing of Light is refreshed by Steadypal
[10:39:14.870] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:39:15.935] Steadypal's Blessing of Light is refreshed by Steadypal
[10:39:24.810] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:39:25.180] Steadypal's Blessing of Light is refreshed by Steadypal
[10:39:26.007] Steadypal's Blessing of Light is refreshed by Steadypal
[10:39:34.011] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:39:35.963] Steadypal's Blessing of Light fades
[10:39:43.663] Steadypal Judgement Heroic Training Dummy 1111
[10:39:44.073] Steadypal gains Blessing of Light from Steadypal
[10:39:52.957] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:39:53.345] Steadypal's Blessing of Light is refreshed by Steadypal
[10:40:02.339] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:40:03.378] Steadypal's Blessing of Light fades
[10:40:12.010] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:40:12.243] Steadypal gains Blessing of Light from Steadypal
[10:40:21.476] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:40:21.880] Steadypal's Blessing of Light is refreshed by Steadypal
[10:40:22.689] Steadypal's Blessing of Light is refreshed by Steadypal
[10:40:30.724] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:40:31.128] Steadypal's Blessing of Light is refreshed by Steadypal
[10:40:39.967] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:40:41.121] Steadypal's Blessing of Light fades
[10:40:49.626] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:40:58.926] Steadypal Judgement Heroic Training Dummy 1111
[10:41:07.913] Steadypal Judgement Heroic Training Dummy 888 (R: 222)
[10:41:08.105] Steadypal gains Blessing of Light from Steadypal
[10:41:08.908] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:18.173] Steadypal Judgement Heroic Training Dummy 1178
[10:41:18.595] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:27.400] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:41:27.800] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:28.603] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:36.631] Steadypal Judgement Heroic Training Dummy 1178
[10:41:37.036] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:45.886] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:41:46.277] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:46.277] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:47.097] Steadypal's Blessing of Light is refreshed by Steadypal
[10:41:57.070] Steadypal's Blessing of Light fades
[10:41:57.155] Steadypal Judgement Heroic Training Dummy 1178
[10:42:06.253] Steadypal Judgement Heroic Training Dummy 888 (R: 222)
[10:42:06.372] Steadypal gains Blessing of Light from Steadypal
[10:42:15.627] Steadypal Judgement Heroic Training Dummy 1178
[10:42:16.346] Steadypal's Blessing of Light fades
[10:42:24.875] Steadypal Judgement Heroic Training Dummy 888 (R: 222)
[10:42:25.225] Steadypal gains Blessing of Light from Steadypal
[10:42:34.099] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:42:34.500] Steadypal's Blessing of Light is refreshed by Steadypal
[10:42:35.314] Steadypal's Blessing of Light is refreshed by Steadypal
[10:42:44.579] Steadypal Judgement Heroic Training Dummy 1178
[10:42:44.950] Steadypal's Blessing of Light is refreshed by Steadypal
[10:42:53.797] Steadypal Judgement Heroic Training Dummy 1178
[10:42:54.935] Steadypal's Blessing of Light fades
[10:43:03.528] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:43:04.838] Steadypal gains Blessing of Light from Steadypal
[10:43:14.310] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:43:14.646] Steadypal's Blessing of Light fades
[10:43:14.714] Steadypal gains Blessing of Light from Steadypal
[10:43:15.400] Steadypal's Blessing of Light is refreshed by Steadypal
[10:43:23.961] Steadypal Judgement Heroic Training Dummy 1178
[10:43:25.403] Steadypal's Blessing of Light fades
[10:43:33.613] Steadypal Judgement Heroic Training Dummy 1111
[10:43:33.994] Steadypal gains Blessing of Light from Steadypal
[10:43:34.796] Steadypal's Blessing of Light is refreshed by Steadypal
[10:43:42.827] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:43:44.782] Steadypal's Blessing of Light fades
[10:43:52.521] Steadypal Judgement Heroic Training Dummy *1999* (R: 111)
[10:43:52.899] Steadypal gains Blessing of Light from Steadypal
[10:44:02.324] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:44:02.941] Steadypal's Blessing of Light fades
[10:44:03.383] Steadypal gains Blessing of Light from Steadypal
[10:44:11.431] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:44:13.375] Steadypal's Blessing of Light fades
[10:44:20.714] Steadypal Judgement Heroic Training Dummy 999 (R: 111)
[10:44:21.089] Steadypal gains Blessing of Light from Steadypal
[10:44:29.924] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:44:31.075] Steadypal's Blessing of Light fades
[10:44:31.138] Steadypal gains Blessing of Light from Steadypal
[10:44:39.223] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:44:39.574] Steadypal's Blessing of Light is refreshed by Steadypal
[10:44:39.980] Steadypal's Blessing of Light is refreshed by Steadypal
[10:44:48.411] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:44:49.971] Steadypal's Blessing of Light fades
[10:44:57.665] Steadypal Judgement Heroic Training Dummy 888 (R: 222)
[10:44:58.083] Steadypal gains Blessing of Light from Steadypal
[10:45:07.306] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:45:08.065] Steadypal's Blessing of Light fades
[10:45:16.956] Steadypal Judgement Heroic Training Dummy 1111
[10:45:26.199] Steadypal Judgement Heroic Training Dummy 1111
[10:45:35.428] Steadypal Judgement Heroic Training Dummy *1999* (R: 111)
[10:45:35.840] Steadypal gains Blessing of Light from Steadypal
[10:45:44.692] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:45:45.075] Steadypal's Blessing of Light is refreshed by Steadypal
[10:45:45.880] Steadypal's Blessing of Light is refreshed by Steadypal
[10:45:53.919] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:45:54.319] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:03.697] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:46:04.192] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:13.215] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:46:13.607] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:13.607] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:14.409] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:22.870] Steadypal Judgement Heroic Training Dummy 1178
[10:46:24.054] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:32.094] Steadypal Judgement Heroic Training Dummy *2356*
[10:46:32.493] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:40.963] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:46:41.325] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:41.344] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:49.894] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:46:50.982] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:59.213] Steadypal Judgement Heroic Training Dummy 1178
[10:46:59.213] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:08.909] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:47:09.191] Steadypal's Blessing of Light fades
[10:47:18.318] Steadypal Judgement Heroic Training Dummy *1776* (R: 222)
[10:47:18.739] Steadypal gains Blessing of Light from Steadypal
[10:47:18.764] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:19.510] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:27.565] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:47:28.762] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:37.212] Steadypal Judgement Heroic Training Dummy 1178
[10:47:37.607] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:46.501] Steadypal Judgement Heroic Training Dummy 1178
[10:47:46.849] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:46.849] Steadypal's Blessing of Light is refreshed by Steadypal
[10:47:55.694] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:47:56.837] Steadypal's Blessing of Light fades
[10:47:56.923] Steadypal gains Blessing of Light from Steadypal
[10:48:05.142] Steadypal Judgement Heroic Training Dummy 1178
[10:48:05.343] Steadypal's Blessing of Light is refreshed by Steadypal
[10:48:14.583] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:48:15.339] Steadypal's Blessing of Light fades
[10:48:15.810] Steadypal gains Blessing of Light from Steadypal
[10:48:23.872] Steadypal Judgement Heroic Training Dummy 942 (R: 236)
[10:48:24.262] Steadypal's Blessing of Light is refreshed by Steadypal
[10:48:33.105] Steadypal Judgement Heroic Training Dummy 1178
[10:48:33.503] Steadypal's Blessing of Light is refreshed by Steadypal
[10:48:33.503] Steadypal's Blessing of Light is refreshed by Steadypal
[10:48:34.298] Steadypal's Blessing of Light is refreshed by Steadypal


You're Judging every 8-9 seconds at maximum, so there's no way to get more than 1 BoL proc every 8 seconds unless one of the debuff applications causes it. Similarly, there's no way to get N applications per Judgement unless N of the four sources cause it.

There are a two cases like this one where you get 3 procs from one Judgement:
Code: Select all
[10:46:13.215] Steadypal Judgement Heroic Training Dummy 1060 (R: 118)
[10:46:13.607] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:13.607] Steadypal's Blessing of Light is refreshed by Steadypal
[10:46:14.409] Steadypal's Blessing of Light is refreshed by Steadypal


The debuff lines show that all three debuffs were refreshed at this point. The second case occurs at the end of the log, and is basically identical.

So in both cases, you got 3 BoL procs out of 4 sources (Judgement, JoWis debuff, JotJ debuff, HotC debuff). At least three of these sources must proc the buff.

Interestingly, we don't see any cases where we get 4 BoL procs. The probability of this occurring is reasonably low though even if all four sources do proc it (~1.87% chance per Judgement).

It's peculiar that you got exactly 100 BoL procs in 100 Judgements. But I think it must've just been coincidence, because there are many examples where you got 0 procs from a Judgement (and as shown, many where you got multiples from a single Judgement).

I also noticed the same "delayed buff" phenomenon that Yelena posted. There are many double-proc instances that look like this:
Code: Select all
[10:34:24.897] Heroic Training Dummy's Judgement of Wisdom is refreshed by Steadypal
[10:34:24.922] Steadypal Judgement Heroic Training Dummy 1111
[10:34:24.922] Heroic Training Dummy afflicted by Heart of the Crusader from Steadypal
[10:34:25.318] Steadypal gains Blessing of Light from Steadypal
[10:34:25.718] Heroic Training Dummy afflicted by Judgements of the Just from Steadypal
[10:34:26.127] Steadypal's Blessing of Light is refreshed by Steadypal

While we can't rule out combat log latency causing this sort of thing, it happens often enough that it sure seems convincing that HotC and JotJ are both independently causing BoL procs.


Finally, a quick statistical analysis:
binofit(100,300) = 33.33%, CI from 28.02% to 38.98%
binofit(100,400) = 25.00%, CI from 20.83% to 29.54%
So if 3 of the 4 sources proc the weapon, we're within the 95% Confidence Interval. If all 4 proc it, we'd have expected more procs.

"One of these things is not like the other." The trouble now is figuring out which one.

Another parse without JotJ or HotC (i.e., Test L-6) will tell us if Judgement and JoWis can proc it, since we should see a double-proc fairly quickly (~14% chance, so a parse with 50 or so Judgements would probably produce at least a few examples of it, but better to have a full 100-parse).

If that parse shows that we never get a double-proc, then we're done, because we know it's one of those two sources that's not the proc trigger. I'm guessing that this is the case based on the features we're seeing with the HotC and JotJ buff delays.

However, if we do get double-procs, then we perform Test L-7 (i.e. spec HotC) and repeat, though this time we'll want 100 Judgements.

My suggestion would be to clear one spec, put 7 points in Ret for Imp Judgement, and perform L-6. Then spec 3/3 HotC and perform L-7. If you want to spec deeper to get JotW for both tests to make it easier that's fine, as long as you don't pick up HotC along the way.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Call to Arms 5.x, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7655
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis

Postby steadypal » Fri Feb 19, 2010 2:30 pm

k, last test i can do before im out for the weekend


hit and expertise capped

untalented my spec, so in this i am 0/5/7,

http://www.worldoflogs.com/reports/z73s ... details/0/



skimmed thru the logs very quickly and i think i spotted a proc, and refreshed proc before the next judgment..



so whats this mean, we have to figure out if, heart or jotj is the one not proccing it right?,
steadypal
 
Posts: 1206
Joined: Sat Mar 15, 2008 12:28 pm

Re: Theck's MATLAB TPS analysis

Postby theckhd » Fri Feb 19, 2010 4:52 pm

steadypal wrote:skimmed thru the logs very quickly and i think i spotted a proc, and refreshed proc before the next judgment..



so whats this mean, we have to figure out if, heart or jotj is the one not proccing it right?,

Yep. I only skimmed the log briefly, but any double-proc like that means that Judgement and the JoWis debuff are both triggers.

That means that either:
-One of the other two (JotJ or HotC) is not a trigger
-JotJ and HotC are both triggers, but we didn't take enough data to see a 4-proc in our earlier test

I think the next step will be to look for 3-procs with a spec that excludes JotJ (like a ret spec), and then repeat it with a spec that excludes HotC but has JotJ. 3-procs are probable enough that we won't need a huge data set to find them.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Call to Arms 5.x, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7655
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Theck's MATLAB TPS analysis

Postby Yelena » Fri Feb 19, 2010 9:49 pm

I'll apologize in advance for the inconsistent number of Judgements between the logs, 69 and 96 or so respectively, I had a lot of distractions and completely lost count on both parses.

0/53/7 +JotJ / -HotC

There should be at least a few examples of triple procs in that log.

-------------------

0/51/18 -JotJ / +HotC

I didn't notice any triple procs as I was taking the log with this build, further analysis should confirm or deny that.
Valleri - Frostwolf
Yelena
 
Posts: 308
Joined: Sat Jun 27, 2009 4:54 am
Location: Behind the Eight Ball

Re: Theck's MATLAB TPS analysis

Postby Nadir » Mon Feb 22, 2010 1:14 pm

Theck, I was trying to adapt your Mongoose uptime formula to Redoubt for use in Communism's EH/damage taken spreadsheet. Presumably as the bosses' attack speed increases, Redoubt uptime will increase. However you'll start to see uptime depression as the probability of five hits in the 10 second Redoubt duration increases.

Redoubt_uptime = 1-(1-proc_chance)^(proc_duration*boss_hits_per_second)

boss_hits_per_second = (1-player_avoidance)*(boss_attacks_per_second)
proc_duration = 10 seconds
proc_chance = 0.10

What would be the simple way to model that decrease in proc_duration at high boss attack speeds? Multiply proc_duration by (1-(1-player_avoidance)^5)?
Image
Nadir
 
Posts: 266
Joined: Mon May 04, 2009 8:37 am

PreviousNext

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest


Remove Advertisements

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