Ichal

Suppose you're buying a clock, and you're adjusting it. You know that it's 1 o'clock, and had it been summer time, it would have been 2 o'clock. Now, you can't be bothered to remember when the summer time switch is. So you do the clever thing: You set the clock to 1:30. That way your watch is less wrong than it could be, but the downside is that now it's always wrong.

That is how you're suggesting to handle the value of hit rating. And it's very silly, since it's so plainly easy to handle correctly: If you're evaluating a piece of gear with 20 hit rating, and you're 10 from the cap, then the effective amount of hit you will get from that piece is 10. The end.

And this brings me to why your second example was wrong: You're considering replacing a piece of hit gear that will send you under the cap, so the hit has value (it's *additional* hit rating that has no value). The only possible way to enter an infinite loop is if you use the wrong scale when calculating the value of the hit gear you're considering to replace.

 

galzohar

My entire point was to say that this is completely wrong. The extra 10 hit you got, in quite a few situations, may then allow you to make another gear swap to lose those 10 hit to gain more str or other useful stats. Even if not now, then possibly in the future. You could compare items like that if you're assuming the rest of your gear is set-in-stone, but this isn't the case in reality. Gaining hit above the cap has a value that is greater than 0 simply because you're wearing more than 1 item and have more than 1 choice for each slot.

All I really wanted to say is that "hit is X until capped, and 0 after" is just a pretty bad way to evaluate it in any situation where you're anywhere near the hit cap. The best thing would be to manually pick gear combinations and see what's best, but even estimating the value of hit as something between 0 and full is better than "hit is X until capped, and 0 after".

 

Arthaal

Problem with a non zero value is it won't ever represent the relative worth of that extra hit realistically for anyone. It really depends on what other pieces you have available for swap and deciding what its worth should really come down to a judgment call on your end. In such situations, it's best to resort to more holistic gear evaluators, such as Rawr.

EDIT: yes I was agreeing with you.

Percent modifiers R'US

 

galzohar

That is pretty much exactly what I said.

 

Helot

Hmm.. Perhaps Pawn needs to have a slight modification then, with a Max Value setting, so you can adjust what the value of a stat is worth past X. So past 295 Hit would be zero, or close to that, and before 295 it'd be w/e number it is now.

We could always ask the Author if he would be willing to include such a feature. I'm sure it would be valuable to every class that has a cap, not just us.

Chaith logs on
<zyl> Actually, I do like my paladin. He's fun to play, but don't tell Chaith.
<chaith> Looks like i logged in at the right time
<zyl> ....
<zyl> I pressed enter half a second after you logged on.

 

Ichal

Of course, but the made up value for hitrating wont help you make that swap in any way.

Suppose you end up these 10hit rating above the cap, and you can replace exactly that with 1str elsewhere. Doing this would obviously be an upgrade, but your mod wont tell you this because your settings make it value those 10 hit rating significantly.

If instead you set up your mod to give you two values: With and without hit raiting valued, then you have proper information to help you choose if a piece of gear is a direct upgrade, or an upgrade if you juggle around gear/gems elsewhere.

 

gezmodean

This brings up an interesting point. We know that Pawn is useful... but there are things that pawn could implement to be MORE useful for us.

Aside of a threshold system where you can change the value of something based on how much of that stat you have... (particularly for hit and expertise), what other features would make Pawn more powerful for us?

(and I apologize for the absence from the thread recently... glad to see this still going)

 

Bronzehoof

Aside from expertise and hit, the threshold thing could be useful to take into account diminishing returns for various stats.

 

• frmorrison

Aside from reaching the hit cap or expertise cap, there are no diminishing returns for dps stats.

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Malleus

Haste has diminishing returns, does it not?

 

• frmorrison

For a caster, you could hit a point where your GCD is at 1 second, so adding more haste past that is really low increase.

For melee, since our GCD is hard set at 1.5 seconds except for the three paladin spells, adding more haste is just as good as the previous haste (not much gain, but it is constant).

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Caronome

My understanding is that the absolute numbers don't matter, just the relative ratios.

1) However, the two methods seem to be very different. Frmorrison would say 120 spellpower scores the same as 225 intellect while the wowhead would score 120 spellpower the same as 48 intellect. Is that correct? 48 versus 225 is a pretty big difference.

2) wowhead seems to score Stamina and Intellect the same. Is this just a typo; wouldn't Intellect be considerably more valuable in PvE raids?

As a Macintosh user without Excel, rawr and spreadsheets are not options so I would especially like some sort of rule of thumb.

edit: #1 math is wrong - see below

 

• frmorrison

The ratios and numbers change, but they don't change much.

You are misreading something. If an item had 225 Int, I would need to see an item with 300 spell power to be equal.

Giving a score to stamina in PvE is not a good practice, since you get more stamina as item level goes up.

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Caronome

My math was obviously wrong. The Frmorrison algoritm would score 150 SP as 150*1.2 = 180 and score 80 int as 80 * 2.25 = 180. I.e., Int is 2.25/1.2 (1.875) more valuable than SP or it takes 1.875 SP to be equivalent to 1.0 int.

However, I still have two concerns with the wowhead numbers.

a) In the Frmorrison algorithm, int is more valuable (1.875) than SP, as I would expect. But in the wowhead numbers, int is less valuable with a int/SP ratio of 40/100 or .4. I could see people proposing 2 or 1.75 but do not understand the 0.4 in wowhead.

b) Similarly, in wowhead the ratio of int to stam is 1.0 which i interpret to mean that would score 1 int the same as 1 stam. I could see stam being 0% or 10% of int, but not 100%

 

Endoscient

Don't put any value in the numbers on Wowhead, they are extremely wrong. The current ones still values Intellect as the worst stat, and mp5 being better then it (per item point).

 

pamela

I have to say, I agree with Ichal in this situation. Worth X until capped, and then 0 after is the correct way to model the values of pieces of gear. The paradox that you have mentioned is one that you are causing yourself by using an incorrect algorithm. The correct comparison is to compare the delta between wearing nothing and the hit gear vs. the delta between wearing nothing and the str gear. There is no "changing" the value of the hit item, because you have to decide whether the rating on that item is going to apply to you before you decide what value it has in the first place.

As to your other point, if YOU want to value hit above the cap, then that's perfectly fine, but you have to understand it is a completely subjective decision, roughly analogous to giving some value to sta, since it keeps you from dying and dead pallies do no damage.

edit: for clarity

 

ajohnson

What are the new scales for the latest Redcap edition posted on 12/14/08?
Also, are Red Sockets suppose to have this value: StrengthValue*16? That is if we use +16 strength in red sockets?

 

galzohar

Yes, like you should be. Heck str is so much better than other stats, you will often give up socket bonuses and put 16 str in yellow/blue sockets as well.

 

• frmorrison

I just updated the Ret string according to the 12/14 spreadsheet.

Also I added a Ret PvP string.

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Radus

I saw in the holy string Mp5 have a 2.5 value and HasteRating is 1.
What about if I need to focus on haste?
Isn't 2.5 for Mp5 value to bigger compare to haste?

 

Omniro

It actually isn't such a big number. If you take a look at the current gear you will find mp5 values around 15-20 . When you look at haste however, you will more likely find ratings around 50-100.
So a regular piece with 50haste gives you 50pts compared to a regular mp5 piece giving 30pts.

 

Ankha

Ok I might just be lost in the discussion but I failed to find generic stat weights for holy pally gear that made sense. I was looking for something that looks like this...

Int 2.25
Crit 1.5
Haste 1
Spell Power 1.6
Mp5 2.2


These values dont really make sense due to all the talk about MP5 being pretty useless and Int being huge for holyP. Can anyone out there spit out a good and useful PAWN stat weight for holy P's who are trying to gear up with heroics and 10 man naxx? Thanks.

 

• frmorrison

While Int and Mp5 are nearly worth the same amount, look at the item budget. I can buy 10 int gems or 4 mp5 gems for the same item budget, that is why people don't like mp5 (it costs to much for what it gives).

 

gezmodean

Anyone have any thoughts on the 3.1 changes?

 

• frmorrison

Once 3.1 is live and Rawr is updated (likely not until 3.1), then I can update my Pawn values. We don't know how Holy will change until they are done changing around the talent trees.

None of the two Ret spreadsheets are updated for 3.1 either.

DK - Ashbane Failure is the condiment that gives success its flavor.