Karmon

This data set shows with 5% uncertainity that the real missrate lies in the
interval [4.584% 6.586%]
So the predicted 4.37% missrate based on 27% boss basemiss is too far away.

If we push the uncertainity to 0.1% the interval is larger and at the lower end covers 4.37%
(the factor 1.96 becomes about 3.21 in that case if I remeber correctly)
But IMO 2023 tries is to small for 0.1% confidence interval.

 

Karmon

So from the last tests in that thread we have three relations:

1) X%-5-278/15.769 about 5.58 with no point in Weapons expertise
2) Y%-5-(5*U%))-316/15.769 <=0 with one point in Weapons expertise
3) Z%-5-(10*W%))-308/15.769 <=0 with two points in Weapons expertise

These formulas for miss contain two unknowns ,the boss basemiss chance and the hit/skill faktor.
For all what its worth the boss basemiss (Z,Y above) and the hit/skill (U,W above) may be dependant
on the number of points in Weapons expertise.
With the usual whacking away on a boss we only can prove that a certain pair of
bossbasemiss and hit/skill give 0% miss.
The individual values cannot be determined that way.

For praktical purpose at least we know:
with at least one point in Weapons Expertise, Bossbasemiss 25.5 and hit/skill 0.1 works for all
settings of WEx.
( but note formula 3 is also valid with Z=26.5 and W=0.2!)

At the moment it is unproven if the same effekt is observable with 0 points in WEx but 5/10 skill from
Items.

 

sp00n

Exactly, and this is why we would need to find out how much base miss chance a level 73 mob has.
Once we have this, we can eliminate one unknown, or at least disprove a current assumption.

I severely doubt that the base miss chance is dependent on the actual weapon skill, but rather on the expected weapon skill (= level*5).
Of course every point above/below this expected weapon skill will change the miss rate, but not the base miss rate.

Once we know that value for certain, we can work on a formula for weapon skill, and if it really differs in benefit with increasing numbers.

I doubt there is a difference, but if there is one, one should see if there's also a difference between the talented +hit and hit rating through items.

 

gatz

Based on Olga's post and my experience (never seen a miss with a 2h weapon with 94 hit rating and 359 weapon skill) I guess the effect of WS on hit chance relates to the base miss chance on the target.

 

songster

I think there's a need to look at the 1H formula, which has terms *both* for the level difference *and* for the weaponskill/defence difference. This being the case, why should 2H calculations be different?

 

Karmon

Can you copy the formulas for 1H and 2H (as opposed to dual wield)?
Or provide a link?

 

Olgas

Sp00n, what is the basis for your "severe doubt"?

Base miss rate is always a relative thing... attacker v. defender... why could it not be determined on actual Weapon Skill at the time. I think, it would be just as easy for the combat mechanics to work this way, as compared to any other...


Actually, if the effect of Weapon Skill is indeed non-linear, I think it would be easier for the combat system to work this way.

Let's not be overly influenced by old assumptions. The 1H formula that you refer to is the pre-BC model. And we are seeing a lot of evidence that suggests it doesn't apply anymore. So, if we are throwing it out, I think we probably need to throw all of it out -- meaning we should no longer necessarily believe that base miss is based both on level difference and weapon skill/defense difference.

 

songster

Aye, fair point - though it does indeed seem to apply for 1-handers still if I recall right.

Do we have any boss fight parses kicking around from Mutilate rogues with no +weaponskill gear? That might give a preliminary indication as to whether (as suggested), the first few points of weaponskill count more than the next few.

 

Olgas

Nope. No Moonkin.

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

And, actually, the data you supplied would also be perfectly consistent with the Crezax / Wowwiki theory.

If you have +16 skill, then, when attacking a Level 73, t is as though you are attacking a mob of your same level. (Actually even 1 skill point better.)

Your base miss rate dual-wielding against a level 70 mob is 24%. And you have +18.77% hit from gear and +5% hit from Precision, for a total of 23.77%, which means your observed miss rate should be .23%. If you reduce your miss rate by an extra .1% (because of that 16th point in Weapon Skill), your observed miss rate would be .13%. Which is, essentially, what you are in fact seeing...

 

Karmon

May I kindly refer you to post #136 in this thread and to this post?

That is what started the recent activity here

 

Olgas

Interestingly, both formulas 2 and 3 are also consistent with the Crezax / Wowwiki theory.

Applying the Crezax / Wowwiki theory:
In formula two, the base miss rate would be 25%. And total +hit is 25.03%.
In formula three, the base miss rate would be 24.5%. And total +hit is 24.53%

Admittedly, formula 1 is problematic, as it implies a base miss rate greater than 28%... but that seems to be problematically high for all of the currently contending theories, no?

 

Gruntle

Would just like to point out that relation 1 above fits (within the 95% confidence limits) with a base miss chance vs boss mobs of 27.6 (the standard 8.6x% found for 2h weapons + 19 DW penalty).

I think the non-linear weapon skill theory is probably close to the truth, relation 2 and 3 are almost impossible to reconcile with 1 if this is not the case.

 

sp00n

Hm, if it was based on the actual skill points, we wouldn't need to test against level 73 mobs at all.
Just use a weapon with a low enough weapon skill and see the results there, eg. 300 skill in maces, and fight against 63 mobs.

Of course the problem here would be gaining weapon skill while attacking itself, somewhat limiting the possible testing duration.

On the other hand, one could find out to which level you have gained enough hit through weapon skill to never miss a mob. That'd mean beating low level mobs until you encounter a miss. Maybe even with 0 hit rating equipped.
If you encounter one, proceed to another level below, and so on.

Unfortunately these mobs will die quite fast.

Another idea could be to let the mobs do the testing for you. Of course this would only be viable if mob miss rates follow the same mechanic as player mechanic for white attacks.
A level 70 mob should have a 5% miss chance against 350 defense. A level 67 would be then at 6.5% (if 0.1% per weapon skill / defense is correct).
Warriors, Paladins and Druids could then do a test with 365 defense against a level 70 mob (simulating the 3 level difference).
If the miss rates match, then miss rates is purely based on the weapon skill <-> defense skill ratio.

 

Olgas

Sorry, Sp00n, I didn't understand this explanation. Tried to read through it several times.

Though, it does still sounds to me like you are still assuming a linear relationship between weaponskill/defense differential and miss rate. And this is exactly the assumption we are trying to prove or debunk.

Sorry if I missed your point, but I still don't understand the "severe doubt" -- maybe I misread what it even is that you are doubting.

 

aquasheep

This is the point I tried to make in my post a few pages back. In my experience, the commonly accepted pre-BC theory was never right in the first place. I suspect that even with all of the changes made to weapon skill, the one thing that has stayed consistent is the fact that something is different against mobs of +3 levels or higher. Weapon skill has always, and still does, give you a bigger advantage against "orange" mobs than it does everything else.


I just want to throw out there my opinion, based on my own parses (which I unfortunately no longer have, so anyone is free to take this with a grain of salt; but this isn't just blind speculation), which is that weapon skill does indeed affect your base miss rate on everything. Having a single point of weapon skill (i.e. going from 350 to 351) effectively makes the level difference between you and a boss mob only 2 (and would remain at 2 from 351 to 355, until 356...at which point the difference would only be 1 level and so on). This makes that one theoretical point pretty significant, as I believe it brings your base missrate from something near 26-27% to 25.4%. Additional points above that would give you the consistently quoted 0.1% increase, as you are now (for combat purposes) dpsing a mob between 0-2 levels above you.

I admit, I don't know what the missrate against a lvl73 with 350 weapon skill (emphasis on 350) truly is. Pre-BC, I would have said it was precisely 27.6%, (based on the .2% per level weapon skill used to be), but now I can't say for sure.

 

Dontmindme

I've been looking through some old data, various posts throughout this forum including the old Rogue Discussion. As I found data, I took a few notes and tossed some data in a spreadsheet. I did find a couple old posts with someone experimenting with Latro's without Weapon Expertise, it was in the Old Rogue Discussion post #979. This was the only test in that thread between about posts 850 and 1150 that involved boss fights with less than 360 weapon skill (unless I missed one).

2 Boss fights parsed, 302 Hit Rating, Precision, no WEx, with Latro's (353 Weapon Skill)
Expected hit rate based on the hit rating + 0.1 per weapon skill
(dropping fractional weapon skill rating):
=24.45122% or a 1.04878% expected miss rate.

Result 1 had 11 misses in 567 attacks for 1.94004% miss rate or 0.89126% over expected.
Result 2 had 8 misses in 379 attacks for 2.11082% miss rate or 1.06204% over expected.

Even with such a small sampling, a doubled miss rate seems fairly unlikely, but I'll leave official margins of significance to the resident statistician.

What I found interesting is how this fits in with some of the theories presented.
If one were to take the theory that the boss miss rate (before hit rating and precision) is:
24% + (Def [365] - Weapon Skill) * 0.1 for any difference up to 10
but is
24% + (Def [365] - Weapon Skill) * 0.2 for differences > 10

for 353 weapon skill:
24% + (365 - 353) * 0.2 = 26.4% or a 1.94878% miss rate
Thus Result 1 would now be 0.00874% under expected
and Result 2 is now only 0.16204% over expected which seems clearly within expected margins of error (same qualification as above).

What this implies is that there is still an effect to weapon skill until you actually hit the magic 355. Granted this is a small sampling and I think more testing is warranted, but it is certainly interesting.

Methodology:
For the record I used 10*82/52 as the hit rating conversion rate for these calculations. If one were to actually compare that formula to actual hit rating percentages as shown on the character screen, it agrees exactly with natural rounding. (I have done this with a wide variety of gear changes to check it).

I should also add that I chose to drop the extra fractional skill rating (from 14 weapon skill rating about 0.55 actual weapon skill) because it made the data agree better with the theory. I'm not convinced given the rather small sample size which is necessarily correct. Certainly, much more testing with weapon skill rating is needed before we can determine whether fractions are dropped or not.

 

Karmon

Unfortunately the sampelsize is really too small to rule out the 1.048% missrate purely on statistical grounds.
5% confidence interval for Result 1 is: 0.803%-3.076%
5% confidence interval for Result 2 is: 0.661%-3.559%

In both cases 1. 048% is inside the interval albeit at the lower end.

 

sp00n

I apologize, my posting turned out be somewhat different than I intended (edited it multiple times before actually posting).

Basically, in the course of this editing, my "severe doubts" have turned into a "I don't think so" and was trying to show a way to actually determine which is the case, either a fixed base miss chance based on level difference, or a miss rate based purely on defense to weapon skill ratio.

Moreover, English is not my native language, so I find myself sometimes reading a post after 1 or 2 days and not knowing what I was thinking resp. writing at that time myself.

As I've said, based on my experience so far, I don't believe that the miss chance only depends on the defense to weapon skill ratio, but I could be totally wrong of course, and I would really like to have this tested out.

 

Olgas

Hi Karmon,

I've been thinking about this data point a little further.

According to Wowwiki, base miss rates would be:

v. Level 70: 24%
v. Level 71: 24.5%
v. Level 72: 25%
v. Level 73: 27%

This particular test you did suggests that 27% is probably too low a base miss rate against a Level 73. And, trust me, I hear you on that. So far, it seems to me to be the single most troubling piece of evidence against the "Crezax/Wowwiki theory". (In fact, so far, it seems to be the only data set I've seen on these forums that can't be explained by the Crezax/Wowwiki theory.)

However, as you point out, if 27% is probably too low, than 25.5% is even that much worse. I mean, if this data is troubling to a 27% assertion, it seems almost fatal to a 25.5% assertion.

And if 25.5% is too low, then it's very difficult to believe in the linear ".1% +hit per weapon level" theory, because the ".1% +hit per level" theory is really dependent upon an assumed 25.5% base miss rate. Indeed, the theory really stops working if the base miss rate is any higher than 25.5% -- because it can't be used to explain a lot of the other data sets that we have seen posted here.

So, I guess what I'm saying is, that this data point makes me even more inclined to believe that the impact of Weapon Skill isn't linear...

 

Trazhenko

I can break 324 easily if I swap out my +skill shoulders and weapon for the kara ones and malchazeen, but hitting 324 right on the nose might be hard. I know I've got gear with gobs of extra +hit for testing even higher percentages should we find that you still miss at 25.5 with no +skill.

I probably have 286 on the armory right now, but I can bump that +28 on shoulder (drop the skill shoulders), +15 on mainhand (drop the skill dagger), +18 on legs (T4), +30 on trinket (blood furnace one), and a variable amount on the offhand via a blade of the unrequited that I don't use (has 3x stam gems atm, but it could be a nice tool for fine tuning hit level with gems).

I'll try to remember to muck around with my gear tonight and see if I can hit a couple key numbers: 347 for 27% and 357 for 27.6% (counting precision).

Of course, I'll see a decrease in performance if I actually spec no weap. ex. and gear that way, but maybe I can find some unimportant bosses to stab .

Edit: Come to think of it, both of those numbers are easy.
286 in normal gear, change Shoulder, Pants, MH (+61) = 347.
Further replace WSC with blood furnace trinket for 10 more = 357.

 

Apate

If the data continues to elude the community, it may be worthwhile to test and see if skill from talents (or gear) is bugged and not being applied properly.

See you, auntie.

 

Karmon

Well, time to dust off that spare equipment.

Tonight, various selection of bosses: Maulgar, Gruul, Magtheridon; no Voidreaver (I was forced to wear arcane resist gear :/)
15/41/5, 0Wex, 324 hit, dualwield daggers
426 crits, 661 hits, 0 blocks, 389 glances, 30 misses, 96 dodges, 5 parries,
1607 total attacks
503002 damage, 313.006845 damage per swing
183774 hit damage, 278.024206 hit damage per swing
79752 glance damage, 205.017995 glance damage per swing
26.509% crits
41.133% hits
0.000% blocks
24.207% glancings
1.867% misses <---------------------- should not happen under current theory
5.974% dodges
0.311% parries

5% confidence inteval is 1.204%-2.532%. So basemiss againts lvl73 is something between
26.7 and 28.3%.

So a discontinuty with talent Weapons Expertise is established.

 

Karmon

I dont even think that these values hold up:

Lets see: Level 72 implies under that theory 5 skill from 1 point Wex; now that would give
25%-5% (prec) - 5*0.1% = 19.5 % miss.
Converting to hitrating yields: 19.5*205/13= 307.5
But we alreay know from earlier test that 308 hitrating is needed for a level 71 equivalent mob (10 skill points in Wex). So 25% cannot be right for Level72.
Similar for level 71: This would imply 10 skill from Wex.
24.5%-5%-10*0.1 = 18.5; convert to hitrating: 18.5*205/13 = 291.7
But we already know that 308 is needed to not miss for a level71 equivalent mob (10 skill in wep expertise)

Well all this test were done with +skill exclusive from talent weapons expertise. Maybe
the fact that someone skills weapons expertise gives you better basemissrates.

I still need to get lower city rep to get the +20dagger skill ring. Probably at the weekend earliest.

 

Trazhenko

We are working on Kael tonight, but we will probably do a quick Gruul tomorrow. I can try it out there.

How much do you think I should go for? If I wear too much and get 0 misses it won't tell us anything.

 

Olgas

Sorry, Karmon. You've applied the Wowwiki formula incorrectly. It does hold up with these numbers.

Base miss rates:

Level 70: 24%
Level 71: 24.5%
Level 72: 25%
Level 73: 27%

If you have two points in WepEx, then it is as though you are fighting a level 71 mob, so your base miss rate is 24.5. If you have 308 +hit (19.53%) + 5% from Precision, you have a total of 24.53%, just enough to never miss. So, applying the Wowwiki correctly, yields exactly the expected result: a 308 +hit cap, with 2 points in WepEx.

And, if you have one point in WepEX, then it is as though you are fighting a Level 72 mob, so your base miss rate is 25%. If you have 316 +hit (20.03%) + 5% from Precision, you have a total of 25.03%, just enough to never miss. Which is exactly the outcome you experienced yourself (please refer to your post #142 in this thread): a 316 +hit cap, with 1 point in WepEx.

(For some reason, in your calculations above, you double-counted the effect of Weapon Skill. First, you reduced the miss rate to 24.5%. And then you also added .1% hit per point of weapon skill, too. This was an error. Once you had taken into account the effect of Weapon Skill by reducing the base miss rate, it was a mistake to factor it in again.)

Again, I have yet to see any data points in these forums that cannot be explained by the Wowwiki theory.

The only piece of data that seemed challenging to the Wowwiki theory, was that earlier test you had done which suggested that 27% was too low a base missrate for a Level 73 -- that it had to be higher. However, according to the test you did tonight, even that challenge seems to be weakening -- since your test tonight suggests the base miss rate on a Level 73 is between 26.7% and 28.3%.