16 replies to this topic

[Veev]

Primary link: http://www.veevsvaul...pvppower_ss.xls (.xls file, requires Excel, recommended)
Secondary link: http://www.editgrid....arr_pvppower_ss (does not support drop down functionality, not recommended)

This is a simple spreadsheet I've made to help determine the relative value of PvP power for warriors.  It is also useful for helping you determine how to gem, which socket bonuses to use or ignore, which honor/conquest pieces to purchase, and which stats to reforge.

If you're not familiar with WoW spreadsheets, then I'll do my best to try to give a quick explanation of how it works.  I use a popular tool called SimCraft to generate values for how much one point of every stat gives in DPS.  This is mainly a PvE tool, but I try to configure it to more closely resemble a PvP fight.  For me (personally), this means: setting enemy level to 90, using "helter skelter" combat mode (chaotic and PvP-like movement/uptime), average fight length to 200 seconds, only using the buffs and debuffs that my 3v3 has available (10% attack power, 5% stats, and 4% physical vulnerability), using defensive stance, and removing all food/flask/potion buffs, etc.  SimCraft is responsible for the bulk of the math actually going on here, I am just taking the values that it spits out and incorporating some PvP stats into it.  Please note that if your gear differs from mine, or if your arena team's buffs/debuffs differ from mine, your stat values will be different.  How different?  It's hard to say without fully simulating it.  You can either use my stat values and hope they're relatively close to yours, or you can import your own character into SimCraft and run your own simulation.

The primary goal of this entire spreadsheet is mainly the Wowhead link that it provides at the bottom.  This link is setup to rank gear in ascending order, while showing you which gems would work best in them (note that the recommended helm gem seems buggy, but the rest should be accurate).  And one of the more valuable purposes of these stat weights is to see how PvP power currently compares to your other stats.  Increase your current PvP power and you will notice that its value (or stat weight) will decrease in comparison to others.  (But remember that both strength and PvP power scale off of each other, so as you get future gear upgrades you will need to re-simulate.  I will be updating the spreadsheet with my own values every so often.)  No more wondering, or just assuming, how PvP power stacks up against strength or crit gems!

Another interesting stat is resilience.  Since resilience is a defensive stat, it is difficult to put an exact value on it when compared to offensive stats.  I created a drop down menu for you to choose how much you value resilience (which is going to depend entirely on your comp, play style, and personal preference).  By default I have chosen to make it equal to crit (for a balanced build), but if you change it to be equal to mastery (glass cannon), you will notice that it will advise you to skip over many of the resilience socket bonuses.  If you set it back to be crit, you should notice the Wowhead link telling you to utilize more of the socket bonuses again.

The rest of the spreadsheet is pretty self-explanatory.  I have set expertise to be worth zero since I am currently expertise capped while reforging entirely out of it.  Hit I have set equal to mastery, since those are the two stats that are interchangeable when it comes to reforging and trying to get close to 3.00%.  Don't be afraid to play around with manually adjusting any of these values.  If you don't have Excel, you will probably need to upload the .xls file on your own EditGrid account (to actually be able to edit the cell contents).

Questions/comments/feedback welcome.

[Hyuru]

First, and I love you.

[hirtqt]

Second, and I love you all.

[Braindance]

I don't want to get into specific math here (due the very limited time I have right now), but in order to get a more educated guess on how stats compare to each other, it would be wise to run different combat scenarios ranging from 10% uptime to 90% uptime with different combat times. I am confident that you will get different numbers for each scenario, since, given the probabilistic nature of crit; changing the above variables will result in a different relative crit value, which in turn will change the relative values of all the other stats.

Therefore my suggestion to anyone using the modified SimCraft, is to tun it more than once, by changing stats one at a time, in different time windows, in order to generate a "bulk" of results that will give you a better idea of what stat is superior in which case. This way, you will be able to more effectively gauge what is best for you.

[Veev]

Braindance, on 17 October 2012 - 03:44 PM, said:

I don't want to get into specific math here (due the very limited time I have right now), but in order to get a more educated guess on how stats compare to each other, it would be wise to run different combat scenarios ranging from 10% uptime to 90% uptime with different combat times. I am confident that you will get different numbers for each scenario, since, given the probabilistic nature of crit; changing the above variables will result in a different relative crit value, which in turn will change the relative values of all the other stats.

Therefore my suggestion to anyone using the modified SimCraft, is to tun it more than once, by changing stats one at a time, in different time windows, in order to generate a "bulk" of results that will give you a better idea of what stat is superior in which case. This way, you will be able to more effectively gauge what is best for you.

Rather than just simulating extremes (from 10% uptime to 100%), I think it is best to simply simulate what your average uptime is.  Considering I never have 10% uptime (and it's usually something probably closer to 75% or 80%), those 10% statistics would really carry no significance, and would actually only distort the real stat values.  Now you could argue that you should simulate 75% uptime, 80% uptime, and then combine the two, which may give slightly different results from just simming 77.5%, but that's probably being a little too picky for this project.

Another known "flaw" with these stats is that as game duration changes, stat values will change.  If all of your games are over in 20 second because of lolavatar, then crit would be worth quite a bit less than what is shown here.  But I took the fact that most of my games seem to end around when my second avatar comes up, so using 200 seconds with a built-in 20% variance should be fairly accurate for everything except the 5 second faceroll matches (which I'm not going to build my stats around anyway).

And regardless of all of this, the spreadsheet was designed to allow users to easily submit their own SimCraft values.  So if someone wants to mess around with uptime/game duration/etc., they absolutely should, and they should share their results.

[Hackattack3]

Godx, on 17 October 2012 - 02:12 PM, said:

third, and you just put way to much extra thought into a one shot macro class.

+rep

[Braindance]

Veev, on 17 October 2012 - 09:41 PM, said:

Rather than just simulating extremes (from 10% uptime to 100%), I think it is best to simply simulate what your average uptime is.  Considering I never have 10% uptime (and it's usually something probably closer to 75% or 80%), those 10% statistics would really carry no significance, and would actually only distort the real stat values.  Now you could argue that you should simulate 75% uptime, 80% uptime, and then combine the two, which may give slightly different results from just simming 77.5%, but that's probably being a little too picky for this project.

Another known "flaw" with these stats is that as game duration changes, stat values will change.  If all of your games are over in 20 second because of lolavatar, then crit would be worth quite a bit less than what is shown here.  But I took the fact that most of my games seem to end around when my second avatar comes up, so using 200 seconds with a built-in 20% variance should be fairly accurate for everything except the 5 second faceroll matches (which I'm not going to build my stats around anyway).

And regardless of all of this, the spreadsheet was designed to allow users to easily submit their own SimCraft values.  So if someone wants to mess around with uptime/game duration/etc., they absolutely should, and they should share their results.

My point was referring exactly on extremes like the 20-second win (which from what I hear is not a rare occurrence these days).

PS: I was quoted by Veev - I feel somebody

[Veev]

Also I ran some numbers and apparently the 2200 Malevolent weapons are only about 92% as good as Shin'ka, Execution of Dominion (which is actually a pretty significant, and disappointing, difference).

[Stretchmarks]

is that counting the gem?

[Veev]

Stretchmarks, on 20 October 2012 - 11:45 AM, said:

is that counting the gem?

Yes, it is.  I updated the spreadsheet (near the bottom) to show the expected gains going from T1 PvP -> T2 PvP, and T1 PvP -> Heroic PvE.

[Deew]

Veev, on 21 October 2012 - 10:42 AM, said:

Yes, it is.  I updated the spreadsheet (near the bottom) to show the expected gains going from T1 PvP -> T2 PvP, and T1 PvP -> Heroic PvE.

Does this include making up for the resilience you lose from PvP weapon or is this a glass cannon build?

[Veev]

It just uses whatever value you have specified for resilience.  With the current version of the spreadsheet, that means setting resil to be worth 90% of PvP power (which is more than crit).  If you set resil to zero you will notice the gap widen between the two weapons.

[Deew]

The equations don't work in OO and I can't install excel on on my computer (work) so I can't change anything, that's why I asked.

I really appreciate all the work you've put into this, thank you.

[Wexlike]

i find it stupid, that starshatter (normal mode) has 200 more scorepoints than the 470 ilvl pvp weapon.. -.-

http://www.wowhead.c...0;gb=1#two-hand

[Veev]

Wexlike, on 29 October 2012 - 03:38 PM, said:

i find it stupid, that starshatter (normal mode) has 200 more scorepoints than the 470 ilvl pvp weapon.. -.-

http://www.wowhead.c...0;gb=1#two-hand

I think this is because Wowhead is currently bugged and is displaying those PvP weapons as having only 1 PvP power.  I've asked them to fix this but looks like they still haven't.

You'd have to manually enter the stats for both weapons into the spreadsheet, similar to how I did to compare 2200 PvP against 509 PvE.

[Jækop]

Well done, thanks a lot!

[Veev]

Updated with stat weights for 2200 weapons, also updated to show exact DPS increase going from 2200 PvP to heroic PvE.