Expertise: Max or 1200 - Debate has ended

[Smashbrother]

props to OP and others who posted actual data (live test or simulation) along with experimental methodology! Good discussion!

 

TANGENT: Has anyone answered the healing expertise question? I didn't notice a definitive answer.

 

The two possibilities are:

1. expertise of healing target determines expertise healing buff
   
2. expertise of healing caster determines the expertise healing buff
   

 

I don't know which is correct but I would expect #2 because healing is just negative damage and for damage we use the caster's (aka attacker's) expertise. If #1 is the case then the only thing expertise does for a pure healer is damage reduction. Meh.

 

It's B.

[cashogy_reborn]

it's b.

 

2 = b?

 

Edited February 4, 2013 by cashogy_reborn

[funkiestj]

2 = b?

 

 

I edited my original post with a quote from James Ohlen that essentially confirms #2 but the edit never showed up. Oh well. If you google for the James Ohlen PvP 1.2 interview he makes it clear.

[Artlu]

i have tried running 1200,1250 and full expertise on OP healer the most current is of 1250, imo for surviving full expertise has been much better, healing is harder to track game to game so i wont bother to say anything about that. I think it comes down to add power inc healing/damage or expertise inc damage, damage reduction and healing. As a healer id would rather everyone be max expertise but that is an opinion.

 

I’ve tested this at 1214, 1255, 1296, 1396. The OP is correct 1214 exp. Vs 1396 exp. the difference on a normal non-crit hit on say “blaster whip” is 3-4% in the advantage of 1396. Same goes for heals received. Where I found the DM really takes effect on expertise is 1255 and up. The % of advantage for higher exp. drops off to less than 3% 1255-1396 and then less than 1% 1296-1396.

Funny thing I noticed was the opposite for crits. They are about even 1214 vs 1396 and went in the advantage the more exp. I dropped for Power, Surge and Cunning.

 

As a healer I would seriously have to dissect my combat logs to tell who is running with 1214 and who has max expertise.

 

As a burst DPS running with 1255 exp. I rarely notice the difference in my damage output so I can't tell if the person I'm hitting has more, less or the same exp than I do. With a 2 pc. PVE bonus of 15% to my “backblast” damage, if I get back to back crits on “shoot first” and “backblast” throw in a “blaster whip” I just took half your health while you were face first in the dirt. I still have 2 CC’s and my bursts stacked up x2.

 

So I’m all for 2 PVE armorings and 1 crystal.

Edited February 4, 2013 by Artlu
math error

[Xerain]

now maybe more of you won't bother buying pve armorings and crystals huh?

[cashogy_reborn]

now maybe more of you won't bother buying pve armorings and crystals huh?

 

good job reading any of this thread

[Mharz]

... suggests that 1314 is the optimal DPS PvP build, as the damage increase is greater than the loss in damage mitigation

 

/thread

 

Nice followup Cash. +1

[Xerain]

good job reading any of this thread

 

It's 14 pages now! I'm not reading all that. I did OP, good enough.

[tekhiun]

May i suggest that you do 3 samples

 

1- both with 1300+ expertise

2- both with less than 1300 expertise

3- the one you did it now

 

but do it 300 times on each.

 

I know its a pain but the more data the better, 100 might seem like a lot but its still not really enough to rule random variance

[cashogy_reborn]

so i wrote a simulation model using microsoft excel. here is a link if anyone wants to fiddle with it themselves, and look at different builds/values: https://www.dropbox.com/s/xoqnuvb3ykpqtre/swtor_expertise_sim.xlsx

 

i used Flame Burst as the test attack, to maintain some consistency with this test. it also removes the variabilty of player Armor Mitigation from the equation, since internal/elemental damage mitigation is 10% for everyone. i also based this on a 7/3/31 build for Powertech, which includes a 6% increase to fire effects (which i did account for)

 

here are the two builds i have tested

1396: http://swtor.askmrrobot.com/character/4569cc73-0c64-498e-bdaf-b0cec17c179a

1214: http://swtor.askmrrobot.com/character/b47e385d-8d69-45e1-b054-c356c425069c

 

i ran the simulation out to 10,000 randomly generated data points. here are the results:

 

1214 vs 1396

Min: 1145.14

Max: 1204.3

Mean: 1174.93

 

1396 vs 1396

Min: 1129.5

Max: 1189.8

Mean: 1159.52

 

 

the difference

Min: 1.3%

Max: 1.2%

Mean: 1.3%

 

 

 

so there ya go. 1214 expertise does 1.3% more damage than 1396 expertise vs the same 1396 target

 

for kicks, here is a 1314 build: http://swtor.askmrrobot.com/character/99aaf9c6-dc2c-474f-be59-d39ff3610e83

 

1314 vs 1396

Min: 1140.61

Max: 1200.41

Mean: 1170.35

 

the difference

Min: .97%

Max: .88%

Mean: .93%

 

1314 does just south of 1% more damage than 1396 vs the same 1396 target, with a ~1% defensive mitigation loss.

 

if you are a DPS class, PvP healing bonus does not matter. it applies to heals you cast, not heals you receive.

 

and you are sacrificing 1.9% damage mitigation for 1.3% more damage output w/ a 1214 build. as a DPS player, where sacrificing a bit of defense for a bit of damage is a regular occurrence, this seems like one of those smart decisions.

 

and the actual defense difference between 1396 and 1214 is 1.9%

1000*(1+.253)*(1-.1864) = 1019

1-(1000/1019) = .019 -> 1.9%

 

for anyone wondering, here is the defense loss for 1396 v 1314

1000*(1+.253)*(1-.1952) = 1008

1-(1000/1019) = .008 -> 0.8%

which actually suggests that 1314 is the optimal DPS PvP build, as the damage increase is greater than the loss in damage mitigation

 

recap for those that missed it

Edited February 5, 2013 by cashogy_reborn

[veyl]

[quote name='cashogy_reborn']recap for those that missed it[/QUOTE] So a model that you created in a simulation did more damage then someone (the OP) tested in a real game environment and you think you are correct? Did you ever stop and think that something in your math/simulation is wrong since it's clearly conflicting with REAL TIME game numbers? Just curious. Edited February 5, 2013 by veyl

[cashogy_reborn]

So a model that you created in a simulation did more damage then someone (the OP) tested in a real game environment and you think you are correct? Did you ever stop and think that something in your math/simulation is wrong since it's clearly conflicting with REAL TIME game numbers?

 

Just curious.

 

lol. the simulation i made is not wrong. all game formulas are known. i know i am correct

 

and i ran the exact same test as the OP to test, and produced the same results.

 

1396v1214

Min: 1151.36

Max: 1212.82

Mean: 1182.42

 

difference between the 1396v1214 and 1214v1396

mean differences of .63%.

 

which is similar to the OP's result of .7, but differs b/c i ran the simulation 100 times longer to further reduce the effects of RNG

Edited February 5, 2013 by cashogy_reborn

[jjmartin]

Can someone summarize the differences between how Healing and DPS are affected by this?

 

/good info here, btw

[Philelectric]

A good way to test it out again would be simple imo.

 

Lets say I take my assassin. Have 2 power crystal and 2 expertise crystal. find a friend. Put the power crystals in my weapon and my offhand. Hit the targer till I get 100 normal hits then switch to the expertise crystals and do the same thing. The target would have the same expertise but mine would be 82 higher. I could also use my augBlack hole bracer(or whatever is the name, its the 146 rating) instead of my augEWH when I test the lower expertise side to get a more noticeable difference. The other stats are almost the same between EWH and Black hole bracer. (I thing its 1 willpower in favor of the pve.

[Skolops]

So a model that you created in a simulation did more damage then someone (the OP) tested in a real game environment and you think you are correct? Did you ever stop and think that something in your math/simulation is wrong since it's clearly conflicting with REAL TIME game numbers?

 

Just curious.

 

The OP didn't test exactly the same thing. He just figured out - in a way far less mathematically sound as running formulas in a spreadsheet, by the way - who would do more damage to the other between a person of lower expertise and one of higher. What he did not test, and what these calculations do, is who would do more damage to someone wearing maximum expertise. This is what people have really wanted to know in order to most effectively evaluate their gearing options.

[Smashbrother]

The OP didn't test exactly the same thing. He just figured out - in a way far less mathematically sound as running formulas in a spreadsheet, by the way - who would do more damage to the other between a person of lower expertise and one of higher. What he did not test, and what these calculations do, is who would do more damage to someone wearing maximum expertise. This is what people have really wanted to know in order to most effectively evaluate their gearing options.

 

Already settled in previous posts in this thread.

[Loki_]

I will always take two pve 61/63 armourings for bonus and drop my expertise on classes like scoundrel dps +15% crit on back blast, sage healer -1.5sec on healing trance, guardain tank increased duration on blade turning and warding call, commando dps +15% crit on grav round etc.

 

What you also dont account for is the person with two pve armourings might do 0.63 more dmg but he will have alot more than 0.63 health boost so it works out the same in the long run in a 1v1 situation.

[Yeochins]

The OP didn't test exactly the same thing. He just figured out - in a way far less mathematically sound as running formulas in a spreadsheet, by the way - who would do more damage to the other between a person of lower expertise and one of higher. What he did not test, and what these calculations do, is who would do more damage to someone wearing maximum expertise. This is what people have really wanted to know in order to most effectively evaluate their gearing options.

 

Its funny how people like to use mathematical soundness to convince themselves of stuff. There is a reason the "Engineering" profession exists. If those "Mathies" or people who focus purely on the theoretical, were the ones building bridges, you would see many more of them collapse than you do today.

 

Theory is one thing, practice often tells you another. Like for instance, all the number-crunching people didn't think of integer-to-floating point decimal conversions or the error inherent in a 32-bit decimal representation.

Edited February 5, 2013 by Yeochins

[Xerain]

 

Theory is one thing, practice often tells you another.

 

Finally someone gets it. I keep saying this over and over to people but they don't seem to get the concept. Just because it's one way on paper doesn't always mean it is going to be in practice.

[Skolops]

Its funny how people like to use mathematical soundness to convince themselves of stuff. There is a reason the "Engineering" profession exists. If those "Mathies" or people who focus purely on the theoretical, were the ones building bridges, you would see many more of them collapse than you do today.

 

Theory is one thing, practice often tells you another. Like for instance, all the number-crunching people didn't think of integer-to-floating point decimal conversions or the error inherent in a 32-bit decimal representation.

 

When you're dealing with real world issues, this might be a valid argument. The problem is that we're not... we're dealing with a computer game which is using pure math to determine what happens "in reality." The stuff you plug into a spreadsheet is exactly and without variance what the game is doing. This is quite a bit different from designing bridges, where a whole plethora of random variations in nature impact how a design will work out in practice. With things like SWtOR's damage interactions it's all just fixed numbers running through fixed equations in a consistent arithmetic environment.

 

On top of that, the OP and those few who support his conclusions have yet to acknowledge - or perhaps simply to recognize - that the item people are curious about is not who will win a one on one between two players with a 200 or so gap in expertise. This is what his "test" actually looks at. Rather, the question is who will do more damage to some third party target - that is, who does more damage as a whole.

 

Most people already know from experience that the player trading expertise in for power will do more damage, and the math just confirms it.

Edited February 5, 2013 by Skolops

[Yeochins]

When you're dealing with real world issues, this might be a valid argument. The problem is that we're not... we're dealing with a computer game which is using pure math to determine what happens "in reality." The stuff you plug into a spreadsheet is exactly and without variance what the game is doing. This is quite a bit different from designing bridges, where a whole plethora of random variations in nature impact how a design will work out in practice. With things like SWtOR's damage interactions it's all just fixed numbers running through fixed equations in a consistent arithmetic environment.

 

On top of that, the OP and those few who support his conclusions have yet to acknowledge - or perhaps simply to recognize - that the item people are curious about is not who will win a one on one between two players with a 200 or so gap in expertise. This is what his "test" actually looks at. Rather, the question is who will do more damage to some third party target - that is, who does more damage as a whole.

 

Most people already know from experience that the player trading expertise in for power will do more damage, and the math just confirms it.

 

Wrong. If you knew anything about Software Engineering you'd know Computers are far from pure math.

 

If you knew anything about game development you would know they don't even have the mathematical model you're trying to simulate. The developers actually run the entire game to simulate numbers because the spaghetti code makes it nearly impossible to extract a mathematical model.

 

Factor in network latency into the model and then you'll find all your mathematical models incorrect. There is a reason being a programmer requires more than a high-school degree now.

Edited February 5, 2013 by Yeochins

[Skolops]

Wrong. If you knew anything about Software Engineering you'd know Computers are far from pure math.

 

If you knew anything about game development you would know they don't even have the mathematical model you're trying to simulate. The developers actually run the entire game to simulate numbers because the spaghetti code makes it nearly impossible to extract a mathematical model.

 

When they're working correctly for basic human applications they may as well be. This isn't a Pentium from 1993 that is going to miscalculate some digit of Pi 1,000,000 places in. It's basic arithmetic concerned with 7 or 8 place values in very defined formulas which, like it or not, are what the game uses.

 

You're trying to make an argument by sounding like you have something profound to say, but in reality you're talking out of your butt.

[cashogy_reborn]

Its funny how people like to use mathematical soundness to convince themselves of stuff. There is a reason the "Engineering" profession exists. If those "Mathies" or people who focus purely on the theoretical, were the ones building bridges, you would see many more of them collapse than you do today.

 

Theory is one thing, practice often tells you another. Like for instance, all the number-crunching people didn't think of integer-to-floating point decimal conversions or the error inherent in a 32-bit decimal representation.

 

hi, im an engineer. if you really want to get technical about it

 

and guess what? engineers use computer simulations ALL THE TIME. whether you are dealing with physics or with a computer game, there are certain governing laws and equations which control the environment you are working with

 

that guy building a bridge? yeah, hes been using a computer model to determine that what he is building is going to be able to hold up against the different stresses applied to it. do you know why those bridges DONT collapse? b/c they have run the simulations, done the calculations, built scale models, etc.

 

you make it sound like a guy goes out and just inherently knows how to build a bridge b/c his job title is "engineer" and not "math professor". engineering is nothing more than the practical, real life application of mathematics; specifically physics.

 

 

the model i did is correct. just b/c a test is done in "real-life" does not mean that a mathematical model cannot be created to duplicate that experiment on a much larger scale and be just as accurate. and really? integer-to-floating point decimal conversions and 32 bit? youre really going to pull that out? you would never even notice a loss of accuracy in this kind of situation.

 

people wanted to know what 1214v1396 and 1396v1396 looked like. i showed them

 

Factor in network latency into the model and then you'll find all your mathematical models incorrect. There is a reason being a programmer requires more than a high-school degree now.

 

what. please, tell me you are joking. Network latency has NOTHING to do with this. this is not a debate of who wins a 1v1, or even a simulation of an entire fight. its a model that predicts (and accurately at that) damage output for a single attack. and its not like these equations are complicated. they are very simple, and all readily available if you know where to look.

Edited February 5, 2013 by cashogy_reborn

[Smashbrother]

When you're dealing with real world issues, this might be a valid argument. The problem is that we're not... we're dealing with a computer game which is using pure math to determine what happens "in reality." The stuff you plug into a spreadsheet is exactly and without variance what the game is doing. This is quite a bit different from designing bridges, where a whole plethora of random variations in nature impact how a design will work out in practice. With things like SWtOR's damage interactions it's all just fixed numbers running through fixed equations in a consistent arithmetic environment.

 

On top of that, the OP and those few who support his conclusions have yet to acknowledge - or perhaps simply to recognize - that the item people are curious about is not who will win a one on one between two players with a 200 or so gap in expertise. This is what his "test" actually looks at. Rather, the question is who will do more damage to some third party target - that is, who does more damage as a whole.

 

Most people already know from experience that the player trading expertise in for power will do more damage, and the math just confirms it.

 

If you looked at my previous post, you'll see that a sage with 1200 exp does 17 more damage on his TKT to a max expertise target, than a max exp sage against a max exp target. It's negligible to the point of worthlessness. So if people want to spend millions upon millions of credits to do nothing, go ahead.

Edited February 5, 2013 by Smashbrother

[Skolops]

If you looked at my previous post, you'll see that a sage with 1200 exp does 17 more damage on his TKT to a max expertise target, than a max exp sage against a max exp target. It's negligible to the point of worthlessness. So if you want to spend millions upon millions of credits to do nothing, go ahead.

 

It isn't necessary to spend millions of credits. All it takes is swapping in 2 power crystals. In any case, people don't use PvE armorings for extra bonus damage - at least, they shouldn't. They do it to allow them to balance crit better while taking more power mods in their other slots.