search

magey / classic-warrior

109 stars 9 forks source link

Attack table #5

Open magey opened 5 years ago

magey commented 5 years ago

For a summary of all known mechanics and test data see this wiki entry.

For our analysis on aura crit suppression see this wiki entry.

ohganot commented 5 years ago

Current Private Server formula, only the 300-305 wepskill range seems to be inaccurate for Classic Beta.

Skill Difference Multiplier
15 0,65
14 0,7018
13 0,7469
12 0,786
11 0,8203
10 0,85
9 0,8759
8 0,8984
7 0,918
6 0,9351
5 0,95
4 0,9629
3 0,9742
2 0,984
1 0,9926
0 1

Parry chance seems abnormally high vs +3 level mob.

magey commented 5 years ago

We need to test different types of mobs to see if the parry/crit penalties are consistent.

magey commented 5 years ago

Some clues I've found regarding the very high parry chance:

ohganot commented 5 years ago

So they have left in retail boss parry chance probably.. +3 levels means boss attack table basically, so it jumps to 14%

viliustank commented 5 years ago

backstab miss with 8% hit from EJ: 8percentmiss

GuybrushGit commented 5 years ago

The 8% hit cap confirmed by blue post gives more credibility to Beaza's posts on wowwiki back in the day. He came out of nowhere and also claimed a 8% hit cap.

This is his glance reduction formula, which also seems to match the data posted above: unknown http://wowwiki.wikia.com/wiki/Weapon_skill?direction=prev&oldid=362572#Glancing_Blow_damage_reduction

This formula is also very similar to the confirmed WOTLK formula:
http://web.archive.org/web/20130209181158/http://elitistjerks.com/f47/t37804-warrior_fury_compendium/

I think its safe to assume that glancing reduction is then a range between: Low end: 1.3 - 0.05(defense-skill) capped at 0.91 High end: 1.2 - 0.03(defense-skill) capped at 0.99

viliustank commented 5 years ago

agreed

viliustank commented 5 years ago

stuff we still want to test:

  1. dodge: confirm 5%, 5.5%, 6.0%, 6.5% for +0, +1, +2, +3 mobs
  2. figure out what's going on with block if they don't answer my question
  3. confirm there is no benefit to crit suppression from weapon skill
  4. confirm the amount of miss % reduction from weapon skill
  5. confirm parry % for +1 and +2 mobs
  6. confirm whether glance penalty reductions continue to match beaza's formula
  7. confirm we don't miss specials and non-dw whites with 8% hit
  8. confirm that dw miss % penalty is additional 19% or some other number, and is it the same for +0 thru +3 mobs
  9. tell hunters to check for crit suppression
magey commented 5 years ago

I updated the OP with our current task list and moved the individual test breakdown to the wiki entry about attack tables here https://github.com/magey/classic-warrior/wiki/Attack-table

GuybrushGit commented 5 years ago

Came across a comment that could explain the Vem screenshot:

In addition, while our tests have indicated that players are getting the intended 5% miss rate while using a single max-skill weapon against even level enemies (yes, we've tested this exhaustively, and cannot reproduce the 9% miss rate you're reporting for normal swings), we have found the bug causing abilities (ie: heroic strike, sunder armor, etc) to miss around 9-10% of the time under the above conditions instead of the intended 5%. https://blue.mmo-champion.com/topic/69365-18-02-05-kalgans-response-to-warriors/

So at the time there was a bug causing yellow attack abilities to have a higher base miss than expected.

TyrsDev commented 5 years ago
  1. confirm that dw miss % penalty is additional 19% or some other number, and is it the same for +0 thru +3 mobs

Assuming that the 1H chance to miss calculations (before +% Hit gear and talents etc) is (using a +3 Level mob as an example): = Base + (ΔLevel * 1%) = 5% + 3% = 8%

Could it be possible that (with DW as 0 or 1) the full chance to miss calculation is something along the lines of this? = Base + (DW 10%) + (ΔLevel (1% + (DW 3%))) = 5% + 10% + (3 (1% + 3%)) = 15% + 12% = 27%

What this would mean is:

  1. You start with a Base chance to Miss of 5%. 1a. If you Dual Wield, add +10% to this Base chance to Miss.
  2. Increase chance to Miss by 1% per differenc in Level (against higher level enemies). 2a. If Dual Wielding, increase by 4% per Level instead.

If this idea bears any truth, Dual Wielding against an enemy your own level should have a base chance to Miss of 15% (since ΔLevel in the above suggested formula equals to 0).

Edit: Similarly, it could be possible that the chance of being Parried increases by 3% per ΔLevel. 5%, 8%, 11%, 14%.

femdead commented 5 years ago

wonder if LH will change their formulas to match classic beta also, just to clarify - does crit reduction and increased parry chance only apply to player attacking high level mob, or do lower level players/mobs attacking me grant me crit reduction and extra parry?

akylko commented 5 years ago

Please test +2 level targets with 150 and 155 weapon skills. In case +2 level targets will have 14% parry with 150 weapon skill and less than 14% parry with 155 weapon skill, it will means that weapon skill and defence skill difference affects on parry chance but not levels difference. And there will be separate formula for 0-9 skills difference.

magey commented 5 years ago

Added some initial results for +2 and +1 level mobs without any weapon skill bonus. Data provided by dmzor#3092.

magey commented 5 years ago

Added results for +3 level target with +2 weapon skill bonus from dmzor#3092. Although the sample size is low (1200) we can clearly see 27% glance damage penalty at 13 defense/skill delta, as predicted by Beaza's formula.

magey commented 5 years ago

Latest blue post over here finally explains the missed backstab in the 8% hit screenshot; the first 1% of hit from gear is ignored when the difference between your weapon skill and the target's defense skill is bigger than 10. So this means that with no weapon skill bonus, while your chance to miss is 8% your effective hit cap is 9%. I updated the wiki accordingly.

nimeral commented 5 years ago

Hm, so the effective hit cap is 9%, and <=10 weapon-defense difference adds 1% hit.

Also, apparently Beaza's glancing reduction formula is correct.

Do we know if all the other effects of weapon-defense difference are as we expected? I.e. (only) 0.04% difference in parries/dodges/blocks/hits/crits?

GuybrushGit commented 5 years ago

More weight for Baeza's formula, checked 2 of the combat logs: image

The damage variance on glances indicate a range rather than the static reduction private servers have.

Malchar commented 5 years ago

Thank you for all of this great work. Reading through, I am having some confusion about exactly which factors depend on "difference(player level, mob level)" and which factors depend on "difference(weapon skill, defense skill)". The blue posts seem to make no differentiation between the two. However, we know the classic example of glancing blow chance, which depends on difference(player level, mob level).

As I review some of the data, it appears that the "Critical Strike chance is reduced by 1% per each additional level the target has over the player. " only depends on difference(player level, mob level) as well, so perhaps weapon skill does not increase crit chance against higher level mobs? I reference the above-posted data "Target at +3 levels, +5 weapon skill (6.37% crit chance, n=4064)", showing "Crit: 3.62% ±0.59% (147)" which gives a 95% confidence that the crit chance has been reduced by 3% even though the weapon skill difference is only 10.

Either way, it would be very useful to clarify exactly whether we are talking about difference(player level, mob level) or difference(weapon skill, defense skill) in each calculation.

magey commented 5 years ago

Updated results for +3 levels target with +5 weapon skill using new data from Puffymuffins and Towneh.

nimeral commented 5 years ago 
If the target is a mob and defense minus attack rating is 11 or more:
    MissChance = 5% + (TargetLevel*5 - AttackerSkill) * 0.2%

If the target is a mob and defense minus attack rating is 10 or less:
    MissChance = 5% + (TargetLevel*5 - AttackerSkill) * 0.1%

If the target is a player:
    MissChance = 5% + (PlayerDefense - AttackerSkill) * 0.04%

I wish Blizz have confirmed this part, implying 6% raid yellow hit cap on 305 weapon skill. Although it does seem true from the experiments.

Free 2% hit for humans and orcs is, idk, unexpected :D And doesn't it mean a huge change of value of Accuria and DFT..?

Edit: nevermind, I'm dumb, it's effectively the same hit cap as on private servers, no big reason to doubt the tests.

mako-taco commented 5 years ago

The formula we know the most about is dodge, from blue posts:

5% + (level difference) * 0.5%

And from ancient blue posts on weapon skill that state it reduces the chance to X by 0.04% per point, I think it's reasonable to assume this:

5% + (level difference) * 0.5% + (def skill - weapon skill) * 0.04

This is so incredibly simple. When you apply the same constants to parry, you get another simple formula that matches our data for 0-2 level difference:

5% + (level difference) * 0.5% + (def skill - weapon skill) * 0.04

All thats left now is the mechanism for stepping us to 14% when there's a large level gap. Changing the coefficient for level difference based on a large level gap is simple, and fits the data.

5% + (level difference) * (level gap > 2 ? %3 : 0.5%) + (def skill - weapon skill) * 0.04

I guess my point is that we should be trying to find the simplest formulae that fit the data. There's no reason a programmer would go out of their way to make this system so complex... let's take a lesson from physics and strive to find an equation that's "pretty"

GuybrushGit commented 5 years ago

The post above me is nonsense.

magey commented 5 years ago

@mako-taco we know the exact formula for Dodge because Beaza gave it to us; see here. As for Parry, all we can do is guess for now until more extensive testing is done.

magey commented 5 years ago

Updated results for +3 levels target with +5 weapon skill using more data from Towneh and Vilius.

watchyoursixx commented 5 years ago

If you want more samples for ranged I can do so, but otherwise you can remove "tell hunters to check for crit suppression" based on the most recent log I sent you magey

Phibs77 commented 5 years ago

Are there any plans to test with more than +5 weapon skill? (Rogue talent + Human)

GeorgieTragnar commented 5 years ago

Could you confirm Baeza's notes about dualwielding? https://wowwiki.fandom.com/wiki/Miss?oldid=347982

Dual-wielding is a special case. The chance to miss on each weapon is 80% times the chance calculated above, plus 20%. If a miss chance calculated as above comes out to 5% then dual-wielding increases this miss chance to 24% (which is 80% * 5% + 20%) for each weapon. A 10% miss chance with one weapon would become a 28% chance for each dual-wielded weapon, and so on.

which basically means that with 0 bonus weapon skill you should have according to him 26.4% to miss when dualwielding?

cannonpalms commented 5 years ago

@magey Need to separate gear/auras/base stats when reporting attack table tests. A single value for crit is no longer sufficiently descriptive now that this has come to light.

magey commented 5 years ago

I've removed all the test breakdowns from this issue because they were just duplicates of what we have on the wiki page and were often out-of-date.

illutian commented 4 years ago

The Parry formula appears to be:

5+(Target Defense - Weapon Skill)*0.2 for those mobs that are +0, +1 or +2; I used the attack tables and moved the Parry percentages up to the nearest whole number, and both had 'x' equal to 1/5 (or 0.2)

EDIT: Did the math for +0 mobs, any value for 'x' would work, but since '0.2' is the only one that works for 'x' and keeps the rounded value very near what the actual tests show...I figure '0.2' still.

For +3 mobs it appears to be: 5+(target defense - weapon skill)*0.6. This was found by using the 14% parry chance Blizzard confirms is suppose to be there.

Formula used: 5+(target defense - weapon skill)*x = 'rounded up parry percentages from attack tables listed on this project'. Except where +3, which I used Blizzard's stated percentage.

+0 mob Parry: 4.97%; rounded up to 5% +1 mob Parry: 5.75%; rounded up to 6% +2 mob Parry: 6.55%; rounded up to 7% +3 mob Parry: 13.97%; set at 14% (as per Blizzard statement)