View Full Version : Shield Block/Crushing Blows

10-05-2007, 05:32 PM
The 64,0000 Copper Question
Does Shield Block push crushing blows off the combat table or not?

Wait, what? Wasn't this settled a long time ago?
Pretty much, yeah. The simple answer is that if you parse all the hits you take while Shield Block is active, you never get a crushing blow and why isn't that good enough for you anyway? There's always been one lingering question, though: What if the client messes up the message when you get a block against a crushing blow? What if, when you block a crushing blow it just says Mob hits you for 1234. (123 blocked) instead of saying Mob hits you for 1234. (123 blocked) (crushing)? Well, we also have the fact that when blocking in PVP we do see Bob crits you for 123. (123 blocked), so we know the CRUSHING/CRITICAL flag in conjunction with a block is handled correctly to begin with. In any case, the question still got asked from time to time. Finally, it came to a head with the usual gnashing of teeth and calling of noobs, and a couple people set out to test it at last.

So, how to test? Shokk posted this methodology for the testing for someone to follow:

- Identify a Test Mob, by selecting a level 63 melee-based boss without a Frenzy or Enrage. Preferably no special attacks such as Mortal Strike either.
- The Testing Tank must have consistant gear and buffs throughout the testing. (ex. 60% damage reduction, 15% parry, 15% dodge, 25% block, 175 block value, etc)
- Determine the Test Mob's Normal Hit damage range vs the Testing Tank by parsing for "Test Mob hits Testing Tank for ####." (ex. 1000-1200)
- Determine the Test Mob's Crushing Blow damage range vs the Testing Tank by parsing for "Test Mob hits Testing Tank for #### (crushing)." (ex. 1500-1800)
- Confirm that Crushing Blow damage range is 150% of Normal Hit damage range.
- Identify all Blocked Attacks by parsing for "Test Mob hits Testing Tank for #### (### blocked)."
- Examine the damage inflicted by each Blocked Attack to the damage range of Normal Hits and to the damage range of Crushing Blows.
- Analysis.

- If "Conventional Wisdom" is correct then every single Blocked Attack will be roughly equivalent to a Normal Hit (ex. "TM hits TT for 930 (175 blocked)." matches a Normal Hit being blocked.)
- If the OP's premise is correct then you will see that many Blocked Attacks inflict damage that is roughly equivalent to a Crushing Blow. This would indicate that that the 'crushing' tag in the combat log is replaced by the '### blocked' tag when a Crushing Blow is blocked, (ex. "TM hits TT for 1430 (175 blocked)." matches a Crushing Blow being blocked)

I suggested Highlord Taelan Fordring in Hearthglen as an easy to find, not too hard hitting level 63. Any priest should be able to keep a warrior up through his attacks for plenty long enough to get a good data set.

Within short order, we had two sets of results:

Xal's Results
OK, so I went and did as Satrina suggested with Fordring. After about 2 hours of sitting there, this is what I came up with.
First, his average crushing blow was 308.807. This is with a standard deviation of 24.25. The high was 346. The minimum was 263. 75% of the datapoints fell between 290 and 330. This was on a sample of 109 crushing blows.

The block value is the actual amount recorded in the combat log. I was also entirely unbuffed for this - didn't want buffs interfering.
The normal hits were out of 154, mean of 203.85. Standard Deviation was 15.46. 75% of the hits fell between 192 and 217. Minimum was 175, Maximum was 231.

My block value was 179. So, what I am looking for is blocks that range from around 84 to 151. There is a bit of fudge in those numbers because block is calculated before AC. However, that effect would push these numbers higer and I can answer the question with just these.
Out of 862 blocked attacks, the mean was 41. The standard deviation is 16.34. The minimum hit was 13. The maximum hit was 70.
Conclusion? In 862 there were 0 crushing blows that were blocked. However, this does not establish error. This is actually slightly more difficult. There were 1932 total attacks. Of that, 209 were misses, 287 were dodged, and 302 were parried. A total of 44.62% of all attacks were blocked. Assuming that 15% of all attacks were supposed to be crushing blows, the probability for an attack to be both a crushing blow and a blocked hit is .4462*.15 - or approximately 6.7%.

Another way of estimating this number is as follows. 15% of hits are supposed to be crushing blows. Only 5.80% were. So that leaves 9.2% to be made up of dodged, parried, or blocked crushing blows (because a miss would be on the same table it doesn't count here). Approximately 59% of these hits were blocked. So 59% of 9.2 leaves approximately 5.4% of the blocks should have been blocked crushing blows.

Those two give us an estimate of an upper and lower bound. Using them, we should get between 46.58 and 57.54 blocked crushing blows. Because this is assuming purely random error, the error in these two numbers can be associated with 1/sqrt(n) where N is the number of samples or .034%. That means we could be off by 29.4 hits either way. Another way of saying that is that 29.4 is the standard deviation on the number of hits that "shoulds" have landed.

That means 0 is between 1.58 and 1.95 standard deviations from what we would expect. While I don't have my tables with me, if I remember correctly this means we can say with 80-95% certainty that you cannot get a crushing blow and a block at the same time.

In my opinion, that pretty well puts an end to the debate on whether or not shield block prevents crushing blows. I think that shows very well that it does.

Scoth's Results
Well, I just spent the last two hours at Hearthglen messing with Taelan Fordring. And my findings pretty much match up exactly with the dwarf. ;)
The first test I did was with my normal tanking gear. My shield block was 231, the lowest crushing he did was around 260. His max melee hit on me was around 218-219. Given 231*.9 is around 208, so I just needed to see a blocked hit for over 12, preferably around 50 dmg.

Well about 30 minutes went by without results. The second test was just with enough gear to get me to 25% block, so I could pop the 100% mark with shield block, and had around 350 defense. Same results, I was crit a few times, but no abnormally high blocks.


So we have it. Actual parsed numbers showing that even by analysing the damage taken, no blocked attack could be hitting hard enough to be a crushing blow. For now, we're safe to say that you cannot take a crushing blow while Shield Block is active. I say "for now", because we all know that Blizzard can and does tinker with mechanics when they feel it's warranted, so someday we may see this changed. Or maybe not.

All of this does leave one question that is still unanswered: Does the Shield Block buff have to be up to push crushing blows off the combat table, or could you do it by naturally raising your dodge+parry+block+miss to be higher than 85% The paladin tanks have answered that conclusively: Get yourself to 102.4% natural combined miss+dodge+parry+block, and you do not take crushing blows.

10-05-2007, 09:38 PM
Empirical testing post-TBC launch.


Test Methodology:
70 warrior: 492 Defense. 21.23% block, 16.86% dodge, 14.34% parry, 10.68% miss (142*0.04% + 5%)
60 Priest for heals.
One small addon that records incoming attack data in separate sets: while Shield Block is active / while Shield Block is not active

Need a mob that ideally:
- Performs crushing blows
- Does not require movement while tanking (see Crits & Crushes (http://www.tankspot.com/forums/evil-empire-guides/33095-crits-crushes.html)).
- Has no melee special attacks to mess with Shield Block charges

In the absence of an equivalent to Taelan Fordring that I used at level 60, I chose Doomwalker as the easiest to get to. He has abilities that knock you around but with a single healer, I died before he uses them anyway while usually being able to get a couple of rounds of shield blocked attacks in before dying. Chain lightning attack doesn't consume shield block so while it would end a round immediately, it did not waste results. The Mark of Death is a pain in the ass, but that's what a good book is for.

For a world boss, we account for decreased avoidance (0.6% per) and so have expected values of:
20.63% block, 16.26% dodge, 13.74% parry, 10.08% miss

Tests were performed by buffing with available stamina buffs (PW:F, Commanding Shout, Elixir of Fortitude). Abilities that affect damage taken and/or health but not type of hit taken were used as available from cooldowns (healing potions, pre-nerf Lifegiving Gem, Last Stand, Shield Wall). A goal of 3000 shield blocked attacks and 3000 non-shield blocked attacks was set for data collection. After collecting 3000 shield blocked attacks, Doomwalker was repeatedly engaged without activating shield block to reach the data set goals

Test Execution:
With Shield Block: Bloodrage then Commanding Shout. Allow Bloodrage cooldown to finish and lost health to regenerate. Approach Doomwalker with Bloodrage, activate Shield Block on each cooldown until death with priest healing, keep Demoralising Shout and Improved Thunderclap up. Run back, resurrect, wait for debuff, repeat.

Without Shield Block: Bloodrage then Commanding Shout. Allow Bloodrage cooldown to finish and lost health to regenerate. Approach Doomwalker until death with priest healing, keep Demoralising Shout and Improved Thunderclap up.. Run back, resurrect, wait for debuff, repeat.

Test Results:

Shield Block Active

Attacks: 3002
Misses: 305 (10.16%)
Dodges: 481 (16.02%)
Parries: 402 (13.39%)
Blocks: 1814 (60.43%)
Critical Hits: 0 (00.00%)
Crushing Blows: 0 (00.00%)

Shield Block Not Active

Attacks: 3000
Misses: 310 (10.33%)
Dodges: 485 (16.17%)
Parries: 414 (13.80%)
Blocks: 625 (20.83%)
Critical Hits: 0 (00.00%)
Crushing Blows: 452 (15.07%)
Normal Hits: 714 (23.80%)

As expected, shield block still removes crushing blows from the table given sufficient supporting avoidance to exceed 102.4% total.

Observe that 25% block is not required to remove crushing blows from the table. (Mere pedantic confirmation as we already knew that paladin tanks are uncrushable without 100% block.)

No critical hits were observed in the testing. In the case of Shield Block active this is consistent with expectation, as removing crushing blows from the table would also remove critical hits from the table, regardless of defense score. In the case of Shield Block not active, this is consistent with 490 defense removing critical hits from the combat table.

However, it must be noted that statistically, if a 0.1% chance of critical hit were indeed always present on the combat table, we could expect to see 3 critical hits over the sample of 3000 non-shield blocked hits. Of course, given the statistical variance in observed results (e.g. 10.33% observed miss rather than 10.08% expected) it is possible that the low expected value simply did not appear in this round of testing. In the case of a 0.01% chance of critical hit remaining on the table, we would expect to not see one in this sample set size in any case.*

* That said, this far into TBC and only seeing certain mobs landing crits against 490+ defense, I still submit that increased critical hit chance is the correct answer, and not "always a small chance". Even at 0.01%, by now we would be seeing many more reported critical hits than we have seen. This is, of course, opinion and nobody outside of Blizzard knows the fact of it.