Well you are a hunter, so you might not know.
The 5% comes from the rogue's talent Precision, which increases the chance to hit (or reduces the chance to miss) by 5%.
This is why we are subtracting the 5% from the base miss chance, whether it is 24%, 27.6% or 28%.
Just because we don't need them to fill up with hit rating.

This and well, I believe, rogues are the only class that is dual wielding and can gather enough hit rating to never miss. Hunters do not need as much (melee) hit rating, and I don't think the equipment for Warriors allows them to reach the cap.
Also every rogue even considering going that deep into weapon skill and hit rating discussion will have that talent.

So yes, for a rogue with only one weapon, "5-5? That couldn't be right" is very right.

Heh, referring to my post earlier, I did recover my combatlogs when I was wielding a fist/dagger combination with +10 additional skill for my offhand dagger (sometimes the sloppy temp files handling of windows comes in handy).

For my sword/sword combo and 360 skill each I had a dodge rate of 5.44% over 4502 attacks. With the fist/dagger combo and 360/362.53 skill I had a dodge rate of 5.15% over 9337 attacks.
Of course the different data samples makes this a bit diffuse, but I was only running one weapon with an increased skill, and regarding the 2.6 to 1.3 speed ratio, the dagger was doing quite exactly 2/3rd of the attacks. So presumably, if I would have had 2 daggers, the dodge rate would have been even lower.

One more interesting note: the dodge rate of the special attacks (Sinister Strike for Nightbane and Hemo for the rest) was even lower than on white hits.
The sample size is even smaller here though, only 1671 special attacks with a dodge rate of 4.67%

To sum up:

Sword/sword
360 skill
4502 attacks
5.44% dodge rate

Fist/dagger
360/362.53 skill
9337 attacks
5.15% dodge rate

Fist special attacks
360 skill
1671 attacks
4.67% dodge rate

I would have liked to compare crit rates for special to white attacks for the several fights, but as I was running with fist spec, and my offhand didn't profit from it, it was rather meaningless.


So maybe special attacks have a different formula for weapon skill, or maybe this is just due to the different sample sizes.

Sorry to disappoint you Spoon, but Olgas is right that the -5 in the second formula
is derived from the datapoints you posted.
Otherwise you dont get a linear relationships between 350 and 355 skill rating.

Olgas, sure the 5% base miss and the 19% DW penalty are not disputed.
But if you construct a piecewise linear function all that matters is that the formula
for the two pieces result in the same value at the juncture regardless which branch you use.

So new data for 357 hitrating.

This time Prinz in Karazhan.
Again 15/41/5, 0 WEx, and no +skill items.
397 crits, 640 hits, 0 blocks, 338 glances, 6 misses, 114 dodges, 7 parries
1502 total attacks
386339 damage, 257.216378 damage per swing
146356 hit damage, 228.681250 hit damage per swing
58225 glance damage, 172.263314 glance damage per swing
26.431% crits
42.610% hits
0.000% blocks
22.503% glancings
0.399% misses
7.590% dodges
0.466% parries
5% confidence interval is [0.0803%, 0.718%]
357 hitrating results in 27.6390 hit%.
Adjusting the interval results in [27.719%, 28.357%] with center 28.038%.

Well, I gotta confess I'm not sure exactly what you just said here, lol -- too much math lingo.

But I do understand that both formulas would technically "work" -- in other words, they'd both arrive at the same value. And that's why I concede that the matter needs to be tested.

But I do still think that the formula you guys are proposing is less likely to be true. Why? Because it is a substantial departure from the way miss rate has always generally been calculated -- i.e., as adjustments to a 5% (24%) base.

You've invented a whole additional step -- ignoring that 5% base -- in order to achieve a linearity that just isn't necessary...

So, I'm still not sure why? Is it just because you don't like the thought of the abrupt discontinuity at +11 Def Skill?

I'm sorry I don't get what you mean. I even translated your posting into German to maybe get behind it, but without success. I'm still not entirely sure.



I've always taken into account the -5% through Precision, because a) every rogue wanting to test weapon skill will have it, and b) you would need 79 more hit rating to test it if you didn't have it, which I doubt is reasonable to fulfill with current equipment.

I can see that you posted this
here, maybe this is where the confusion comes from. On both formulae there should be 24 or 24-5, depending on whether you have Precision or not.


Maybe I should post the formulae which I currently assume.

If 365-skill > 10
5 + 19 + 10*0.1 + (365-skill-10)*0.6 = Chance to Miss

If 365-skill <= 10 (and >= 0)
5 + 19 + (365-skill)*0.1 = Chance to Miss


This is
Base Miss Chance
+ DW Penalty
+ 10*0.1 [10 because here the maximum is 10. You're fighting a mob with more than 10 defense above your weapon skill]
+ (Boss Defense Skill - Weapon Skill - 10)*0.6 [Remember the 10 before? They are being accounted for here]


You then have the chance to miss a target. From this chance you subtract any talents (Precision) and multiply by 10*82/52 to see the required hit rating.

Ahhhh.....

Now I see what you're driving at. This is a bit different than the way Karmon expressed it. But I see what you mean, now. And, your formula makes sense to me.

So, we have two potential formulas, both of which I'd say are potentially equally valid. They both seem to fit the data that we have collected so far. The two formulas are:

1. Yours, as stated above.

And

2. The "modified Wowwiki" formula, which I articulated earlier. But which I'm modifying a bit more now to reflect the fact that 27.6% seems to be too low of a miss rate for a Level 73. Let's assume the base miss rate is indeed 28%, so we can see what the formula looks like:

To be honest, at this point, I'm not sure which of the two formulas I think is more likely.

Your formula is definitely simpler and more elegant. And it would be just as easy to program and implement as the modified Wowwiki formula.

However, the Wowwiki formula is closer to being consistent with the Crezax post. Crezax said, against a Level 73 mob, 4 points of Weapon Skill would give you .8% hit (essentially .2% per skill level). And the .6% per skill level in your formula is a lot larger figure than the .2% Crezax indicated. Of course, he seemed to indicate .2% exactly -- and that doesn't look to be just right either, since even under the modified Wowwiki formula the number needs to be a little bit higher than .2%.

I confess, up to this point, I've been somewhat partial to the Crezax / Wowwiki formulation, because it seemed to explain things better than the other theories. And because it was based on a blue post.

But I'm not so sure anymore. As of right now, I'd say I'm evenly split between the formula the way you expressed it and the "modified Wowwiki" formula.

I guess we still need more empirical testing to see which formula is correct.

//EDIT

After writing the above, I read back through this thread and came to post #166. Post #166 describes a test without WepEx, but using Latro's (which adds roughly +3.6 Weapon Skill). I believe the results from that test are more supportive of Sp00n's formula than of the modified WowWiki formula. However, it's impossible to know for sure -- since we really can't test either theory for true accuracy until we definitively know the base miss rate of a Level 73.

Once we know the true base miss rate against a Level 73, we can test to determine how much each incremental point of Weapon Skill is worth. As Sp00n mentioned in an earlier post, testing this with weapons like Latro's will be difficult because those weapons provides something like +3.55 Weapon Skill, instead of an even +3. As Sp00n suggested, maybe we can get a Rogue with Mace Specialization to test it. Or, a Warrior with Weapon Mastery or a Paladin with Weapon Expertise would also do the trick, since those talents add +Weapon Skill in even increments of +2, also.

No. I really talk about basemiss chance. The 5% from precision and the amount of hitrating
is not included here


I take the liberty to reformulate your equation:
If 365-skill > 10
starting with Spoons proposed formula
5 + 19 + 10*0.1 + (365-skill-10)*0.6
5 + 19 +(365-skill)*0.6 - 10*0.6 + 10*0.1
5 + 19 +(365-skill)*0.6 + 10*0.1 - 10*0.6
5 + 19 +(365-skill)*0.6 + 10*(0.1-0.6)
5 + 19 +(365-skill)*0.6 + 10*(-0.5)
5 + 19 +(365-skill)*0.6 - 5
ending with my proposed formula.

Spoon, you just expressed it differently.

Well I would not be surprised if ranged attack and pet attack behave different.

Yes, after a night of sleep I suspected so. All the time I was misinterpreting the 5% as Precision, until I noticed that 24-5+15*0.6 = 28 just as 24+10*0.1+5*0.6 = 28.

As for ranged attacks, the formula would suggest a miss chance of 9%, opposite to the theorized 8.6 or 8.64%.

On the other hand, special attacks for rogues are also suspected to be 8.6% or 8.64%.
May, as seeing in the difference in dodges, special attacks follow a different formula, and ranged attacks the same, or even another one?
Remember, ranged attacks also cannot glance and cannot be dodged.

Yes, I understand that the math works either way.

But the issue I take with your expression of it, Karmon, is that it implies that all 15 points of Weapon Skill are worth .6% each. Which therefore also implies that the base miss rate against a Level 70 is 19%, not the agreed 24%.

By contrast, Sp00n's expression of the formula implies that the first 10 points are worth .1% each, and the next 5 points are worth .6% each. And this preserves the idea that the base miss rate against a Level 70 is 24%.

If you approach this from the perspective of a heavy-duty math person, you'd probably say I'm being silly -- both formulas work -- I'm just splitting hairs.

But from the perspective of the non-mathematician, these types of difference matter a lot -- because they tell a different story. And Sp00n's expression is just a much easier way to understand and conceptualize the situation.

Well so far the data seems to show that e.g. dodge is affected differently for white hits and special attacks, meaning there could be another formula at work for special attacks in regards of weapon skill.
And since ranged attacks do not suffer from dodge and parry (of which at least dodge is affected by weapon skill, parry is unknown), it *might* be the case that ranged attacks utilize the same formula as special attacks.

No it does not imply that that all 15 points of Weapon Skill are worth .6% each.
It only implies that the 5 points between 350 and 355 Wepaon skill are 0.6% worth.
If you look again, my formula for skilldifference between 0 and 10 does not has the -5 term!

No, it also tells the first 5 skillpoints, from 350 to 355 skill, are worth 0.6%. Then the next 10 skill points are worth only 0.1% per skill.

You do seem to ignore the condition that are attached to both parts of the missformula.

If you look again, my formula for skilldifference between 0 and 10 (that imply lvl 70 to 72 mob) also "preserves the idea of 24% miss agains lvl 70"

If you feel better, than use Spoon's formula. But you cannot claim that my formula is invalid.
And it does not tell a "different story".

Urgh. So many different ways of saying the same thing, and tempers rising too. Is there really a need for this?

Conceptually it's quite simple.

We have now established that the base miss rate (i.e. with no weaponskill) against a boss mob is considerably higher than previously thought. That therefore implies that the first 5 points of weaponskill must have more effect than the next 5 points.

The only remaining question is the shape of the graph linking +weaponskill to +hit. On one hand we have Karmon's idea, where each point of weaponskill from 1-5 gives you a flat amount of +hit%, and that subsequent points from 6-15 again give you a flat but lower amount. On the other hand we have Olgas, who thinks that the majority of the discontinuity is loaded into the fifth point of weaponskill, and that going from +4 weaponskill to +5 weaponskill gives you a whole 1.24% to hit from one single point of weapon skill.

I find the latter idea extremely conceptually ugly and unlikely. Not to mention the fact that it jibes oddly with the weaponskill rating system. According to Olgas, we could be in a position where a single point of weaponskill rating (one point in the item budget) could confer more than 1% to hit, depending entirely on what other gear you have equipped.

However, it should be easy to test: all we need is a long log from a sword rogue with Latro's and no points in Weapon Expertise, or logs from rogues with 1, 2, 3 or 4 points in mace spec.

Personally, I suspect both are likely to be wrong, and the function linking +weaponskill to +hit is going to be nonlinear but continuous, without abrupt jumps. The first point will be worth more than the second, which will be worth more than the third, and so on. Furthermore, I strongly suspect that the same holds true for casters, it's just obscured by the fact that skill levels in the various schools of magic vary in discrete steps of +5 as you level up.

However, pinning down the exact shape of the curve would require testing at many many levels of +weaponskill, and is likely to be far more trouble than it's worth.

Well both formulae are basically the same. The only problem with Karmon's variant is that it is not visible on first sight that they are the same.

You are over-reacting. I did not say your formula was invalid. In fact, I said "the math works either way."

However, at a minimum, the "-5" term that you have introduced is just confusing. Heck, it definitely confused me -- and I'm obviously very close to this whole analysis. I mean, look, we've just needed to write about 6 or 7 posts back and forth to explain your "-5". It confused both me and Sp00n.

Sp00n's formulation is easier to understand. It is no more accurate mathematically, but it is a better description of the underlying concept, especially to the uninitiated.

Shouldn't it also be possible to attain a skill >375 with WExp?

Take some of:
Shapeshifter's Signet +5.1 skill
Grips of Deftness +3.8 skill
Belt of One-Hundred Deaths +6.1 skill
and one or two of several daggers that also give +skill

Not that such gear is available to me, sadly, but if the model is correct and symmetrical, there ought to be increasing returns to +skill above 375

Well, I wouldn't say this is the only remaining question. We still haven't definitely determined the base miss rate against a Level 73. Seems like it's close to 28%, yes. But until we know for sure, it'll be hard to test for the shape of that graph.

I agree wholeheartedly that the big jump on one point is extremely conceptually ugly. I don't like it either. The alternative being suggested by you and Karmon is MUCH nicer in that regard. And I appreciate your arguments about the Weapon Skill rating system -- good point there.

But, on the other hand, we have that darn blue post from Crezax... which says against a Level 73, 4 skill points should be worth .8% +hit (which is a lot less than the 2.4% hit that your and Karmon's formulation would indicate). And, why the heck did Crezax use FOUR points as his example anyway? Why didn't he just say that FIVE points would give you +3% hit? That would have been a lot more impressive response to the question he was answering... but that would have implied .6% per point, which is maybe something he specifically wanted to avoid suggesting...

(Look, I don't want to put too much stock in this one blue post. Or even over-analyze it. It could simply be incorrect. And Crezax isn't even with Blizzard any more. But, at one point, he did have insider access to the real truth here, and that is more than you can say for any of us poor slobs.)

Also, the formula on Wowwiki, the way it is written, does imply that abrupt jump at skill point #5. God only knows who wrote that entry on wowwiki. And, yes, that entry could certainly be wrong. But it was written a while ago, ahead of all our thinking, and has really helped advance this whole analysis considerably... so I do give it some limited credence, until it's proven wrong.

As I said earlier, right now, I'm intellectually evenly-split between the two models. Emotionally, I think I actually LIKE your model better -- because it doesn't have that ugly jump.

But the reality is like or dislike has nothing to do with anything. We have to test.


Hmmm.... this is an interesting new twist. But I don't think it can be right. The testing so far on mobs below Level 73 seem to confirm a flat +miss and +hit of .1% per level. And, remember, some of these tests were done with WepEx and some without. So, I'm feeling pretty comfortable that we've established linearity between Levels 70 and 72.

Well, remember Crezax was posting about the change in weapon skill. It's possible he meant it gave 0.2% more than it used to (and we didn't know how much that was...)

The big point here was to tame a pet 3 levels below the target mob. I don't know that much about hunters and pets but I guess the pet has maxxed out weapon skill for its level when it gets tamed? So for a level 54 servant you take a level 51 pet and so on.

I think it's probably a fair assumption that, at the moment you tame the pet, it is maxed for its level. So, upon taming, a level 51 pet would have 255 skill. The problem is that, as soon as you use that tamed level 51 pet in combat, it starts gaining experience towards level 52. And it also starts improving its Weapon Skill too. Unfortunately, the interface doesn't reveal what exactly the pet's Weapon Skill is, so you'd have no way of knowing when its Weapon Skill was increasing from 255 to 256 to 257, etc. And, remember, at the beginning of a new level, you usually see those first couple of points of new Weapon Skill pretty quickly, which I would assume would also occur with the pet. So, I'd guess it'd gain a point or two of new Weapon Skill pretty quickly, but we wouldn't be able to see it and account for it in our calculations.

The solution then would be only to attack with the pet while it's 3 levels below.
As soon as it gains a level, release it and fetch another one.