Block % only on hits?

Tifi · 03/22/07, 9:45 AM

After reading about the 1 roll system I thought it was not very convincing and decided to do my own tests. After spending some time with the dogs in Netherstorm, I chose the Shadowmoon Chosen to be somewhat better mobs for testing purposes (all lvl 70). Anyway, here's the raw data:

My stats while performing the tests:
Defense 454
Chance to be missed +4.16%
Dodge 12.70%
Parry 14.73%
Block 21.14% (I never used shield block while doing this test)

Combat stats against Shadowmoon Chosen (lvl 70) (code format for monospaced font)

         observed  in %  abs. err.
Swings   4548
Missed    421      9.26%  -0.10%
Dodged    573     12.60%  +0.10%
Parried   688     15.13%  -0.40%
Blocked   873     19.20%  +1.94%
Hit      1993     43.82%

Assuming an unmodified chance to miss of 5% for this mob (9.16% in total), this is pretty strong support for the 1-roll theory. The absolute errors for Miss, Dodge and Parry are -0.10%, 0.10%, and -0.40%, respectively.

The absolute error for Block, however, is 1.94%. Which is still small, but first of all, there is evidence that blocking does not work on the same lines as the other possible swing outcomes. In another thread, someone observed a (crushing) hit with shield block activated, giving him a total block chance of 99.75%. This result would not be possible with a 1-roll system. Second, look at this: The relative error of blocking is 19.20% / 21.14% - 100% = -9.20%. Since the mob missed with 9.26%, that leads me to the hypothesis that the combat system works like this:

In the first step, one roll is performed on a table that looks like this: Miss, Dodge, Parry, Crushing, Crit, Hit. Then, a second roll is performed for Block, but only if the outcome of the first roll was not a Miss. If the result is a Miss, Dodge or Parry, it is taken as the final outcome. Then it checks for a Block. And finally it checks for Crushing, Crit or normal hit.

The numbers support this theory, and it would explain the crushing blow with 99.75% block rate. However, from an implementation point of view, it doesn't seem very plausible, because the second roll would be wasted CPU time if it was a Dodge or Parry.

As a next step, I'm going to test this against a dual wielding mob, because they have a higher miss rate. I assume they also get the -19% to hit modifier as dual wielding PCs, so a level 70 mob should have a miss rate of 28.16%, and I would expect my effective block rate to be (100%-28.16%)*21.14% = 15.19%.

I believe 4500 swings is a good sample size. Unfortunately I'm not good at statistics, so if anyone could enlighten me as to how good it actually is, that would be great.

P.S.: First post, so Hi all! =)

[Edit: Base miss rate for dual wield is 5%+19%]

Kalman · 03/22/07, 9:54 AM

Has anyone ever presented a convincing proof that crushing occurs with <25% block rate, but with >25% total avoidance, prior to shield block, while shield block is active?

• Chicken · 03/22/07, 10:07 AM

Originally Posted by Kalman

Has anyone ever presented a convincing proof that crushing occurs with <25% block rate, but with >25% total avoidance, prior to shield block, while shield block is active?

While this is anecdotal for the most part, I do know that as tanking Paladin I don't seem to get any crushing blows when I'm at a combined dodge/block/parry/miss of over 100%.

I'll remember to gather some proper data on the subject (combat logs, etc.) for analysis.

Tifi · 03/24/07, 6:42 AM

Originally Posted by Kalman

Has anyone ever presented a convincing proof that crushing occurs with <25% block rate, but with >25% total avoidance, prior to shield block, while shield block is active?

Wouldn't it be sufficient to provide a proof for a normal hit to occur under the given circumstances? There's not really a need to only fight 73+ mobs for this kind of test, if we believe that their crushing blows take priority over normal hits.

Anyhow, here's some new data I've gathered yesterday. I had slightly different stats:

Defense 465
Chance to be missed +4.60%, adding up to 5%+19%+4.6% = 28.6% on a dual wielding mob
Dodge 11.92%
Parry 15.17%
Block 21.58% (again, no shield block)

Stats vs Demon Hunter Supplicant (dual wield, lvl 70) (I hope thats the mob's name in the english client)

         observed  in %  abs. err.
Swings	7457		
Miss	2114	28.35%	0.25%
Dodge	880	11.80%	0.12%
Parry	1108	14.86%	0.31%
Block	1326	17.78%	3.80%
Hit	2029	27.21%

So, all in all pretty much the same picture, the values for miss, dodge, and parry are accurate, and we have a somewhat larger error on block. My prediction would have been (100%-28.35%) * 21.58% = 15.46% total block chance, so this data unfortunately does not really confirm my theory. Although it's still a better match than the one-roll theory.

To be honest, I'm running out of ideas of how this could work. Maybe it's more complicated, like there's a maximum of blocks per second/minute or sth. Any suggestions?

• Vulajin · 03/24/07, 10:01 AM

Here's the thing. Does the parser you are using count partial blocks as blocks or as hits? Do you ever experience a blocked crit during these parses?

I have been fixing to test this for quite some time myself, but have never been able to do so because I couldn't find a combat parser that separated the results out with such accuracy. Partial blocks are always attributed as low-damage hits or crits. If you can find (or if anyone can write) a parser that does this properly, then we can get some more accurate data and figure out the answer once and for all.

Nite_Moogle · 03/24/07, 11:34 AM

There was a bugged warrior on test at the time when Naxx was on the PTR that ended up with a few thousand defense and he saw miss, dodge, and parry prioritize over block.

Emeraude · 03/24/07, 3:43 PM

Dug up an old thread/research I did.

I think there is a "priority" on said debuff/buffs that causes the table to change in combat, still roll once mind you, but some things hold more weight than others. Which leads people to come up with the "checks" conclusion. The thing is certain stats hold more weight in combat than others. Miss > Parry > Dodge > Block > Crit > Crush > Hit

We all know, that when you add Parry/Dodge/Block, these things take from the hit table. That is +1% Parry = -1% hit. The thing that makes people say specifically is, "Well it checks for dodge/parry before it checks for block! So if your block is at 100% after it fails it's dodge/parry check block stops it!" is what causes confusion(I should know, I used to think the exact same way).

There is a sense of priority in fact, if you had 50% Parry and 50% Dodge, and you added +1% Parry, would you have 51% parry and 50% dodge? No, actually you'd have 51% Parry and 49% Dodge, Parry holds more weight, this is the reason block doesn't override dodge/parry/miss.

Missing holds more weight than anything, it would be deterimental to a defender if their miss rate was at 95% to allow dodge/parry/block to happen in place of miss. Same with Miss/Dodge/Parry. If you had 40% dodge/40%parry/40% Miss, you would have 40% Miss/40% Parry/20% Dodge. If you had 50% Block/20% Parry/20% Dodge/20% Miss, you rates would be 20%/Miss/20% Parry/20% Dodge/40% Block.

Basically the order is important, even though everything is checked at once, anything below the other on this list is taken up by the thing above it I've found.

Miss - If a mob put +100% miss on you, none of the below would happen(unconfirmed) Example 1
Parry - With 100% Parry the only other thing that can happen is miss(8-piece Wrath) Example 2
Dodge - With 100% dodge, parry/miss still happen(Evasion) Example 3
Block - With 100% Block, dodge/parry/miss still happen(Shield Block) Example 4
Glancing - With 100% Crit, Glancing still happens, but so do miss/parry/dodge/block(Recklessness) Example 5
Crush - Recent testing I did shows that if a mob's Crush is extremely inflated, crit does NOT happen, Crush/Miss/Parry/Dodge/Block do, but Crush takes priority over Crit. Example X
Crit - With 100% crit, miss/parry/dodge/block still happen, normal hit does not(Recklessness again) Example 6
Hit - always happens without inflation, added stats, and is dead last priority, first to be replaced(Zing) in addition 100% hit is not possible, but if it was, it would look like Example 7

Example 1: 100% miss table:

100% miss ;p

Example 2: 100% parry table:

5% miss
95% parry

Example 3: 100% dodge table:
5% miss
5% parry
90% dodge

Example 4: 100% block table
5% miss
5% parry
5% dodge
85% block

Example 5 & 6: 100% crit table against glancing/hit
5% miss
5% parry
5% dodge
5% block
40% glancing
40% crit

Example X: 85% Crushing chance
5% miss
5% parry
5% dodge
5% block
80% crushing

Example 7: 100% hit

5% parry
5% dodge
5% block
5% crit
40% glancing
40% hit

Cathela · 03/24/07, 7:23 PM

Originally Posted by Tifi

To be honest, I'm running out of ideas of how this could work. Maybe it's more complicated, like there's a maximum of blocks per second/minute or sth. Any suggestions?

I tend to doubt it's a strict limit on blocks/sec, becuase when I AoE tank the non-elite packs in the Karazhan ballroom, I can pop Dabiri's Engima and see a very dense string of blocks pop up on SCT. It might be some kind of short-term penalty to your block rate after each successful block or something like that, but that's hard to test.

Very interesting data. I'll try to collect some myself.

Karoshi · 03/24/07, 8:56 PM

Originally Posted by Tifi

After reading about the 1 roll system I thought it was not very convincing and decided to do my own tests.

As I wrote on the other thread, I think there is indeed a very convincing thing about the one-roll-sytem: I'm sure, it's possible for a prot-paly to push crushings off the table, but as Holy Shield (our "Shieldblock") only gives us 30% block we would need 70% base block to avoid crushings. In more than one month tanking Karazhan I didnt get any CBs so we should have only two possibilities:
a) It's a one-roll-system
b) The block of paladins and warriors have two different rolling-systems

In fact a warrior with 26% block and SB up still receives crushings, but I still think thats just a simple lag- & sync-issue. IMO it's really not that complicated, it' just damn bad luck.

Though it doesnt deal with the question if its a one-roll, I'm not quite sure, what to think about Emeraude's theory. It sounds logical in some way, on the other hand the deeper mechanics of WoW seem to be optimized for fast parsing and a something like a sum of 160% block/parry/dodge/miss could still be possible too. So the numbers could just be optimized quite simple - lets say we have miss, block, parry & dodge all at 30% (120% total) it could possibly just shrink to 25% each to have 100% in the end.

• sp00n · 03/25/07, 9:16 AM

Well Emeraud's posting describes the current "standard theory", at least to my knowledge.
Nothing too special about it.

Tifi · 03/25/07, 3:22 PM

Vulajin: I'm using an addon called CombatMonitor (so it's not a WoWCombatLog.txt parser). You can download it from the link below. There is also a screenshot where you can see the output format. It counts total blocks and partial blocks. I don't think it's possible for a blocked hit to crit or be crushing, at least I've never seen it happen.
http://www-fr.curse-gaming.com/files...batmonitor-pl/

Emeraude: While I do believe you're right with the priorities, the data I collected does not suggest that a 1-and-only-1-roll is the answer to everything. The data above shows that the total block rate is lower than what the tooltip says. This cannot be explained by some part of it being pushed of the table, because I just didn't have enough total avoidance. It also doesn't explain (actually it contradicts) the possibility of hits or crushing blows with shield block up, and a block rate of like 97%. I've never seen a CB with SB up myself though.

Cathela: Ok thanks, so we can discard this max # blocks/time idea.

Karoshi: Did you never ever get CBs, or just not when Holy Shield is up? Are you the MT? Could you post your stats for Karazhan?
I just came from my first run, and 9 of 166 hits (257 swings) from the maiden were CBs. (Weird side note: I had 2 full blocks. And normally she hits hard, for like 2.5-3.5k. I even have it in my combat log, so it's not an error of the addon.)

Anyway, thanks for all the input guys, much appreciated!

• Vulajin · 03/25/07, 4:42 PM

Originally Posted by Tifi

Vulajin: I'm using an addon called CombatMonitor (so it's not a WoWCombatLog.txt parser). You can download it from the link below. There is also a screenshot where you can see the output format. It counts total blocks and partial blocks. I don't think it's possible for a blocked hit to crit or be crushing, at least I've never seen it happen.
http://www-fr.curse-gaming.com/files...batmonitor-pl/

I was not referring to separate programs either, but even addons like your CombatMonitor simply parse the messages the game sends to the combat log (because those are always much more descriptive than the barebones event the game fires on a combat event). So they're all "combat log parsers" in a sense. Anyway, that's a bit off-topic.

With respect to blocked crits, those are possible on yellow attacks. I have had blocked Stormstrike and Eviscerate crits on mobs. I have also had a blocked Sinister Strike crit against a player. On white attacks, however, we are fairly certain that block is an independent result from crit.

Blocked crushes, I am pretty sure those are impossible. But not 100%.

Tifi · 03/26/07, 7:54 AM

I'm just reading the code of the addon, and I noticed that it has a notification function if "impossible" things occur. It would play a sound and throw an error if a crit or CB was (partially?) blocked. Never happened to me, just thought you might find this interesting.

I'm not sure if I understand you correctly (English is not my first language). Do you mean a player's special attack being both a crit and blocked by a mob/player? Because I only consider mob attacks on a player. And I don't think mob specials/spells can crit.

Karoshi · 03/26/07, 8:18 AM

Originally Posted by Tifi

Karoshi: Did you never ever get CBs, or just not when Holy Shield is up? Are you the MT? Could you post your stats for Karazhan?
I just came from my first run, and 9 of 166 hits (257 swings) from the maiden were CBs. (Weird side note: I had 2 full blocks. And normally she hits hard, for like 2.5-3.5k. I even have it in my combat log, so it's not an error of the addon.)

Of course with HS up to get ~100% avoidance/partial avoidance. If HS or Redoubt wasnt up, I'd be crushed, thats clear. I dont have any stats to post at the moment as i reset recap, etc. after some days in order to save memory, but i'll try to capture some data in the next runs.

Mob spells can crit as far as i know, but the chance should be very very small.

Hozz · 03/26/07, 9:45 AM

Originally Posted by Chicken

While this is anecdotal for the most part, I do know that as tanking Paladin I don't seem to get any crushing blows when I'm at a combined dodge/block/parry/miss of over 100%.

I'll remember to gather some proper data on the subject (combat logs, etc.) for analysis.

I have had Nightbane crush me on landing with shield block on, and thus dodge/block/parry/miss well over 100%. My block was not over 100%, however.

Drauk · 03/26/07, 9:54 AM

Originally Posted by Karoshi

Mob spells can crit as far as i know, but the chance should be very very small.

No, mobs can't crit with spells at all.

• Chicken · 03/26/07, 10:43 AM

Originally Posted by Hozz

I have had Nightbane crush me on landing with shield block on, and thus dodge/block/parry/miss well over 100%. My block was not over 100%, however.

Have you ever had crushing blows occur outside of when Nightbane was landing with Shield Block up however? Due to the movement involved I can imagine some wackiness going on with regards to Nightbane's actual position and the position he appears to be on your client, that is, it looks like he's in front of you, but he actually isn't, thus making it so you can't block (or dodge or parry).

I haven't had the opportunity to gather any data on the subject at least, I was the best equipped choice for healing out of the choices of third healer we needed for Karazhan this week, so didn't get to tank much outside of the occasional trash, which is too low level to crush.

Karoshi · 03/26/07, 10:51 AM

Originally Posted by Drauk

No, mobs can't crit with spells at all.

Good to know. Wasnt quite sure about that as I never got any crits, but some people meant they did.

magey · 03/26/07, 10:51 AM

Originally Posted by Tifi

(Weird side note: I had 2 full blocks. And normally she hits hard, for like 2.5-3.5k. I even have it in my combat log, so it's not an error of the addon.)

That usually happens to me when I have PW:S on and the mob hits me for an amount that uses up the PW:S completely, but the remaining unabsorbed amount gets blocked fully because it's small enough.

LoL · 03/27/07, 12:35 PM

http://img410.imageshack.us/img410/9333/crushedfo4.jpg

Apate · 03/27/07, 12:51 PM

Originally Posted by LoL

http://img410.imageshack.us/img410/9333/crushedfo4.jpg

Timestamps would probably help a good bit here. If the gain of shield block and the crush were in close succession, it could be lag related, as the next hit was blocked.

• zeidrich · 03/27/07, 12:53 PM

Originally Posted by LoL

http://img410.imageshack.us/img410/9333/crushedfo4.jpg

Sync issues. You can't trust the combat log to give you an accurate account of the order events occur.

For instance, what happened there IMO was something like this.

Server calculates hit on you, shield block is down, it crits. Server sends crush message to you.
You click shield block. Message shows up in log saying you gain shield block immediately.
Sends command to server saying you have activated shield block.
Message about you getting crushed comes to you and gets processed by combat log afterwards.

It's very common. I mean how often do you see a combat log that looks like this:

A mob dies.
You get 300 Experience.
You crit A mob for 987 damage.

Obviously you're not actually killing him after he's dead. Just the way that messages are processed in the combat log are not necessarily in the same order as they're processed on the server.

Kalman · 03/27/07, 1:16 PM

Or the ever so popular:

You die.
X heals you for 9000.
Y hits you for 500.

Hozz · 03/27/07, 3:18 PM

Originally Posted by Chicken

Have you ever had crushing blows occur outside of when Nightbane was landing with Shield Block up however? Due to the movement involved I can imagine some wackiness going on with regards to Nightbane's actual position and the position he appears to be on your client, that is, it looks like he's in front of you, but he actually isn't, thus making it so you can't block (or dodge or parry).

I haven't had the opportunity to gather any data on the subject at least, I was the best equipped choice for healing out of the choices of third healer we needed for Karazhan this week, so didn't get to tank much outside of the occasional trash, which is too low level to crush.

Very possible it was a directional thing, I hadnt considered that.

bobtheorc · 03/27/07, 4:13 PM

I tend to use a block heavy tanking style and lean heavily on my shield block button. Though i didn't take a screenshot, i recal blocking several crushing blows from the Bog Lords in Heroic Underbog as well as on the Romio and Juliet fight in Karazhan. With trinkets going, the combat log looked something like:

Bog Lord hits Bobtheorc: 7500ish Crushing (877 blocked)

This is a pretty common occurrence on the Bog Lords.

I suspect the mechanics would put block more as a damage reduction to the incoming hit after the final result has been calculated than higher up on the priority.