Remove Advertisements

ShoR testing on the PTR

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

ShoR testing on the PTR

Postby jere » Sat Jul 11, 2009 10:36 am

I wanted to keep this in the Adv. Training forums.

Here is some data I gathered on ShoR on the PTR:

Code: Select all
                  Damage
BV             MIN     MAX
---------------------------
2400      2920   2920
2413      2932   2933
2439      2956   2956
2454      2969   2970
2475      2987   2988
2521      3026   3026
2627      3103   3104
2673      3133   3134
2825      3211   3211
2978      3260   3260
3022      3269   3269
3025      3269   3270
3041      3272   3272
3087      3278   3278
3100      3278   3279
3117      3280   3280
3130      3280   3280
3155      3281   3281
3282      3280   3280


Note the oddity at 3155. I double checked both it and 3282, and those results were repeatable.

I then extracted out the block value past 2400 and also back calculated the "effective" block value of the damage done:

Code: Select all
              Eff Block Value
Extra BV       MIN     MAX
-----------------------------
0        0   0
13       12   13
39       36   36
54       49   50
75       67   68
121      106   106
227      183   184
273      213   214
425      291   291
578      340   340
622      349   349
625      349   350
641      352   352
687      358   358
700      358   359
717      360   360
730      360   360
755      361   361
882      360   360

Again, the line associated with 3155 has an oddity.

I tried using the interpolation algorithm and came up with three equations:
Ignoring last two lines: y = -0.0006x^2 + 0.9510x - 0.2712 ===> R^2 = 1
Ignoring only last line: y = -0.0006x^2 + 0.9507x - 0.2511 ===> R^2 = 1
Using all lines: y = -0.0006x^2 + 0.9360x + 0.6595 ===> R^2 = 0.9999999

y ==> effective block value
x ==> actual block value

So there are some interesting oddities.

As far as spec, I went up into protection for redoubt since I didn't have enough block value gear to get above 2400 without it, but I avoided any damage increasing talents like 1HWS or Crusade

NOTE: Some of the recorded damage values could be incomplete. I hit them for a while each time, but it could be just a really low probability of hitting the other min/max value. So there is some error there.


EDIT: Here is the chart for the "ignore last data point" fitting. Series 1 is the min values, and series 2 is the max values. I fit the min values for those equations.
Image
Image
User avatar
jere
 
Posts: 2951
Joined: Fri May 11, 2007 5:12 pm

Re: ShoR testing on the PTR

Postby toothdecaykills » Sat Jul 11, 2009 10:41 am

Similar findings were posted (albeit without the actual DATA) on the wow tanking forums.

With this, is there a way to see what the ACTUAL cap for BV is? Its unlikely any of us will hit it before we gain another 10 levels. I apologize if this somehow seems like childs play, but I'm likely to never understand how to do any of this math. I'm a security officer for a reason, the only thing I can do properly is look imposing ;)
toothdecaykills
 
Posts: 138
Joined: Sat Jul 28, 2007 10:50 pm

Re: ShoR testing on the PTR

Postby jere » Sat Jul 11, 2009 10:51 am

It's somewhere between 3100 and 3117. I didn't have enough granularity to find the spot where it hits 360 first.
Image
User avatar
jere
 
Posts: 2951
Joined: Fri May 11, 2007 5:12 pm

Re: ShoR testing on the PTR

Postby Sabindeus » Sat Jul 11, 2009 4:54 pm

So you stopped seeing extra damage between 3100 and 3117, and in the patch notes they said the maximum contribution would be at 34.5*lvl, which is 34.5*80 = 2760... which I guess is referring to damage, not BV.
3280-2760 = 520. So that works out.

Thanks for doing these tests Jere.
Image
Turn In, an NPC interaction automator - http://wow.curse.com/downloads/wow-addo ... rn-in.aspx
User avatar
Sabindeus
Moderator
 
Posts: 10455
Joined: Mon May 14, 2007 9:24 am

Re: ShoR testing on the PTR

Postby theckhd » Sun Jul 12, 2009 7:24 am

Just to confirm your results, here's what I got with the curve fitting tool in MATLAB for your data set, excluding the last data point:
Code: Select all
Linear model Poly2:
       f(x) = p1*x^2 + p2*x + p3
Coefficients (with 95% confidence bounds):
       p1 =  -0.0006259  (-0.00063, -0.0006217)
       p2 =      0.9507  (0.9475, 0.9539)
       p3 =     -0.2511  (-0.641, 0.1387)

Goodness of fit:
  SSE: 1.906
  R-square: 1
  Adjusted R-square: 1
  RMSE: 0.3564


and excluding the last two data points:
Code: Select all
Linear model Poly2:
       f(x) = p1*x^2 + p2*x + p3
Coefficients (with 95% confidence bounds):
       p1 =  -0.0006264  (-0.000631, -0.0006218)
       p2 =       0.951  (0.9476, 0.9545)
       p3 =     -0.2713  (-0.674, 0.1314)

Goodness of fit:
  SSE: 1.834
  R-square: 1
  Adjusted R-square: 1
  RMSE: 0.362


These agree with your model very well. From this we can infer that the intercept (p3) is zero, which makes sense intuitively as well. If we force that to zero and just fit p1 & p2, we get:

Code: Select all
Linear model Poly2:
       f(x) = p1*x^2 + p2*x + p3
Coefficients (with 95% confidence bounds):
       p1 =  -0.0006245  (-0.0006282, -0.0006208)
       p2 =      0.9494  (0.947, 0.9518)
       p3 =           0  (fixed at bound)

Goodness of fit:
  SSE: 2.108
  R-square: 1
  Adjusted R-square: 1
  RMSE: 0.3749


I'm going to make a guess here that the exact values are p1 = -6.25e-4 and p2 = 0.95. Fixing all three values as such, we get:

Code: Select all
Linear model Poly2:
       f(x) = p1*x^2 + p2*x + p3
Coefficients (with 95% confidence bounds):
       p1 =   -0.000625  (fixed at bound)
       p2 =        0.95  (fixed at bound)
       p3 =           0  (fixed at bound)

Goodness of fit:
  SSE: 2.527
  R-square: 1
  Adjusted R-square: 1
  RMSE: 0.3855


With these coefficients, the point when you reach the maximum would be x=-p2./(2*p1), or 760, for a total of 3160 BV from gear. The amount of added damage you get at this point is 361.

So this suggests the appropriate formula is:
Code: Select all
BV<2400           520 + block_value
3160>BV>2400      520 + 2400 + 0.95*(block_value-2400) - 6.25e-4*(block_value-2400)^2
BV>3160           520 + 2400 + 361 = 3281


<edit> added ^2 that I missed
Last edited by theckhd on Sun Jul 12, 2009 4:30 pm, edited 1 time in total.
"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: ShoR testing on the PTR

Postby Dread » Sun Jul 12, 2009 10:34 am

Yay math!
Dread
 
Posts: 89
Joined: Sat Nov 01, 2008 5:36 pm

Re: ShoR testing on the PTR

Postby majiben » Sun Jul 12, 2009 11:39 am

Dread wrote:Yay math!

Image
Amirya wrote:some bizarre lovechild of Hawking, Einstein, and Theck
User avatar
majiben
Moderator
 
Posts: 6999
Joined: Fri Aug 22, 2008 4:37 pm
Location: Retired

Re: ShoR testing on the PTR

Postby jere » Sun Jul 12, 2009 1:04 pm

You can always count on Theckhd for awesome stuff!
Image
User avatar
jere
 
Posts: 2951
Joined: Fri May 11, 2007 5:12 pm

Re: ShoR testing on the PTR

Postby theckhd » Sun Jul 12, 2009 2:04 pm

jere wrote:You can always count on Theckhd for awesome stuff!

Don't shortchange yourself, you took all the data. I just spent 10 minutes playing with cftool.

Also, I have a vested interest in knowing the formula, because I need to update the TPS simulations. I plan on re-running the stat sim and including block value as one of the independent variables, so we can see how it drops off compared to our other options.
"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: ShoR testing on the PTR

Postby jere » Sun Jul 12, 2009 3:08 pm

Oh Theck, I think there should be a "^2" somewhere.

I.E.

Code: Select all
BV<2400           520 + block_value
3160>BV>2400      520 + 2400 + 0.95*(block_value-2400) - 6.25e-4*(block_value-2400)
BV>3160           520 + 2400 + 361 = 3281


Should be
Code: Select all
BV<2400           520 + block_value
3160>BV>2400      520 + 2400 + 0.95*(block_value-2400) - 6.25e-4*(block_value-2400)^2
BV>3160           520 + 2400 + 361 = 3281
Image
User avatar
jere
 
Posts: 2951
Joined: Fri May 11, 2007 5:12 pm

Re: ShoR testing on the PTR

Postby theckhd » Sun Jul 12, 2009 4:29 pm

jere wrote:Oh Theck, I think there should be a "^2" somewhere.

Yup, there should be. Thanks.
"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: ShoR testing on the PTR

Postby Widdox » Mon Jul 13, 2009 7:50 am

Thanks Jere and Heck.

You both rock!
Image
Widdox
 
Posts: 350
Joined: Thu Nov 01, 2007 8:33 am

Re: ShoR testing on the PTR

Postby Garath.Gorefiend » Mon Jul 13, 2009 8:33 am

A couple of things.

First off, the BV Cap includes BV from both items and Str.

It also included BV from procs (librams and 4pce T8)

As a result, reaching the cap with a fairly standard prot build (which includes redoubt) is trivial.

Effectively, this has nerfed paladin threat fairly hard. On live, it was trivial to get 5k+ hits / 10k crits with raid buffs. On PTR, you will cap out at 3828 k / 7656k (roughly). The only buffs this will benefit from will be any % damage increasing buffs.

I was easily able to attain the cap with only self buffs on PTR (in full block gear, with self buffs and procs I reach 4.1k BV).

Effectively this means that paladin threat will only scale on about 75% of our threat abilities with raid buffs.

Redoubt now has much less value as a talent as its benefit drops as we get more block gear switching from a combined threat/mitigation talent to strictly a mitigation talent.

The change on threat from SoV does not make up the difference.

Addendum:
I was also able to demonstrate some extraneous results fooling around below the cap...occassionally, as long as my BV is slightly below the cap (2680 BV with procs) I would get unexpected results. The odd hit for 3947/crit for 7893 which is above the expected cap. The majority of hits would be as expected (3694 hit/7388 crit).
Still, no where near the results I can get on live.
5091 hit/10180 crit in full block gear (3026 BV with procs and cooldowns)
3946 hit/7892 crit in what I consider to be my normal tank gear 2346 BV with procs)

Note that either set on live gives me more threat from ShoR than I can get on PTR
Last edited by Garath.Gorefiend on Mon Jul 13, 2009 9:38 am, edited 1 time in total.
Garath.Gorefiend
 
Posts: 406
Joined: Mon Aug 25, 2008 2:07 pm

Re: ShoR testing on the PTR

Postby Minarva » Mon Jul 13, 2009 9:17 am

Then dont wear block value gear like everyone else isn't. I sit at around 1.2k unbuffed BV, and I'm nearly in total BiS hardmode gear from 25 man, the only gear that has BV on for me are my tier gloves.
This is not a nerf for me YET because I simply don't hit that high, however it is an extremly retarded change. This makes it along with shield slam the only damaging ability to have dimishing returns - and yet we are not damage classes (it makes me wonder why dimishing returns was not added to DKs Frost strike when it was destroying the arena, or any other PvP destroying ability). It seems counter intuitive to buff block value to the point it was -ok- to have on your gear, and then nerf it so that in any real amounts it is only a mitigation stat - and an extremly poor one at that.
Another point to consider is how this will change in future content. I would imagine even if you TRIED to not get BV in T9 you would end up with a crapton of it, and heaven forbid level 90. There needs to be some sort of scaling based on itemlevel or something...don't know but it does just make the whole stat suck even more than it does now (to be honest).

Personally I hope this is just a PTR thing, because otherwise it shows how little foresight Blizzard seem to have with their own changes, and frankly seems very unproffesional (particularly if you track the changes to shield slam). Band aids to fix a band aid...no thanks.
Minarva
 
Posts: 123
Joined: Tue Jul 08, 2008 12:58 pm

Re: ShoR testing on the PTR

Postby Zalaria » Mon Jul 13, 2009 9:35 am

Minarva wrote:...
Another point to consider is how this will change in future content. I would imagine even if you TRIED to not get BV in T9 you would end up with a crapton of it, and heaven forbid level 90. There needs to be some sort of scaling based on itemlevel or something...don't know but it does just make the whole stat suck even more than it does now (to be honest).

Personally I hope this is just a PTR thing, because otherwise it shows how little foresight Blizzard seem to have with their own changes, and frankly seems very unproffesional (particularly if you track the changes to shield slam). Band aids to fix a band aid...no thanks.


Don't forget that they have some plan to clean it up that just isn't ready for 3.2. This is just a band-aid on a band-aid, while there's a miracle drug coming out that's supposed to cure it completely.
Whether it does or not... we'll see.
Dovie'andi se tovya sagain - It's time to roll the dice
User avatar
Zalaria
Maintankadonor
 
Posts: 641
Joined: Tue Sep 11, 2007 7:27 am

Next

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