Curse

Captain Robo · Oct 24, 2007, 05:13 AM // 05:13

I ran some tests earlier today that I thought I'd share.

Essentially, one of my friends said that he regularly runs 16 Axe Mastery on his warriors, claiming that the damage increase was significant enough to warrant a superior rune.

I'd always believed minors were the way to go, so I decided to do some tests and make friends with the Master of Damage

The tests were conducted as such:

Me with a vampiric 15^50 Axe, two different axe hats and the following skillbar:

Eviscerate
Executioner's Strike
Agonizing Chop
Frenzy

I conducted 10 trials for each test; more would have yielded more accurate results, but you'll get the general idea.

The first test involved simply attacking the master of damage until he died; no attack skills or frenzy: just c-spacebar. I stopped attacking the moment he died, then copied his readouts into excel to get a gross DPS estimate.

The second was essentially a repeat of the first test, except I kept up frenzy the entire time. The purpose of this was to test the accuracy of the first (its essentially the same) and to see how dealing more damage in less time has an effect with different runes.

The third was a spike test, in which I built up adrenaline, and then unleashed with the three attack skills while in frenzy.

Here's the DPS comparison charts for 14 (yellow) and 16 (light blue) Axe Mastery

And the spike comparison chart:

Again, most of you know that 14 > 16 in just about every case, but here's some additional data for the records. I do not believe the damage gained from 16 attribute points warrants the 75 health you trade for it.

Riotgear · Oct 24, 2007, 05:21 AM // 05:21

The -75 health is applied after DP, which makes it particularly dangerous for something that is going to be permanent throughout the match. Doing 10% more damage on a spike isn't particularly useful when you do 100% less while dead.

Greedy Gus · Oct 24, 2007, 07:29 AM // 07:29

If you play warrior a lot, there is actually a fairly noticeable difference in how powerful you feel with 16 mastery compared to 14, given the same strength value.

But regardless, the rational argument has pretty much always been the same, and works both ways: It really doesn't make that much difference; if near-lethal spikes are a decent possibility (based on your build, metagame, opponent, etc.), you may as well go with a minor. If near-lethal spikes are more improbable, you may as well go with a superior. In either case, it doesn't really matter whether the damage increase is 5%, 10%, or 15%, it's always been known to be minor but not ignorable, and isn't a main factor in the decision.

chardy · Oct 24, 2007, 11:21 AM // 11:21

also 16 axe crit will trigger spiritbond where as 14 axe crit doesnt

Zabe · Oct 24, 2007, 12:39 PM // 12:39

At the old days it was perhaps more favored due to armor swapping that you can carry the superior helmet with you. When gained DP just swapped to minor.

But in thesedays, I rarely even use majors in PvP, just sometimes to get crucial break points etc.

Lord Natural · Oct 24, 2007, 03:44 PM // 15:44

I still use sup weapon in PvE and arenas. Because of the spike potential in HA, and the fact that VoD is so early in GvG now, a minor is more practical there.

Captain Robo · Oct 24, 2007, 03:46 PM // 15:46

Man I miss armor swapping, that used to be my favorite thing

Quote:

also 16 axe crit will trigger spiritbond where as 14 axe crit doesnt

That's reason alone to not run a sup rune

seekjy · Oct 24, 2007, 04:32 PM // 16:32

Nicely executed test. It may lack statistical rigor, but for all intents and purposes, confirms what many people have long suspected but never supported.

Ec]-[oMaN · Oct 24, 2007, 04:37 PM // 16:37

Quote:

Originally Posted by seekjy

Nicely executed test. It may lack statistical rigor, but for all intents and purposes, confirms what many people have long suspected but never supported.

It's been known+ supported for over a year.

seekjy · Oct 24, 2007, 04:45 PM // 16:45

Quote:

It's been known+ supported for over a year.

I disagree. It's been known, but the support given has been very weak. Every test that has been performed, including the ones posted here and at team-iq have not even come to a margin of statistical significance. I've seen the evidence since Alpha, but it's not terribly strong.

Even the test here is still suspect - the Chi Squared on the Frenzied 14-Axe is 122 on 9 degrees of freedom. The probability of it being this skewed or more is one in a billion. So while there's evidence of an effect, there's even stronger evidence that this test is biased against randomness, like other tests preceding it.

I would suggest removing the Vamp mod to reduce variance. But the formulas for weapon damage have long been known, so people are effectively just testing if observed game mechanics match given formulas.

insightful moth · Oct 24, 2007, 05:09 PM // 17:09

It's good to see threads like this from time to time. While most high end players already know statistics such as these, I'm sure there is a majority of players out there who have never given it much thought.

Armour switching was awesome, matching breakpoints for multiple attributes and scrambling for the right equipment when you found out what the opposition had... whether the difference was substantial or not didn't really matter, there was still that satisfaction from micromanaging it correctly <3

Captain Robo · Oct 24, 2007, 05:34 PM // 17:34

Vamp would not have produced statistical error, as it was used for all tests.

The reason I used vamp is because it produces the most increase in damage of all the weapon mods, and to mimic realistic conditions; warriors should be using vampiric weapons when spiking or damaging any non-warrior target.

Using Sundering (which nobody should ever do) WOULD introduce error, as I'd have to perform many, many more tests just to get around that 20% chance of a sundering proc on a single strike, as opposed to dealing 3 more damage on all strikes. (plus, sundering blows and you should never use it, I don't even need to tell you the math on that.)

I think I'll perform more iterations, possibly up to 30 trials for each test to glean some more accuracy, but even after 10 trials, the error bar for these averages is very small (just eyeballing it, it seems like +-1.5, so take that however you will).

Maybe some without vampiric, just to see if that introduces any error.

Andrew Patrick · Oct 24, 2007, 08:27 PM // 20:27

Quote:

Originally Posted by Riotgear

Doing 10% more damage on a spike isn't particularly useful when you do 100% less while dead.

Haha Well said!

I know that I generally avoid sup runes like the plague in PvP, but it's awesome to have the math to point to when people argue with me.

Captain Robo · Oct 24, 2007, 11:06 PM // 23:06

I went back and did ten more trials.

For Normal Attacks:

The difference between 14/16 becomes even more clear now. 16 Axe Mastery only gives 3 more damage, on average, between normal attacks.

For Frenzy:

Again, the difference closes again; to a damage difference of 5.

Spike Comparison

Here's where things got a little sketchy. I'm going to be redoing this test with completely new data, because I got some GODLY spikes off with 16 Axe Mastery for a good period of time... compared to some truly mediocre spikes at 14. I'll redo this test in a bit and post the results.

erk · Oct 24, 2007, 11:38 PM // 23:38

Good work

Got any stats for a Dev Hammer combo using minor vs superior? Dev hammer is also a very popular spike frontline elite, along with Crip Slash. It would be interesting to see as the hammer has higher base DPS than an Axe, so perhaps the two extra attribute points will have a greater impact overall.

Captain Robo · Oct 25, 2007, 06:36 AM // 06:36

here's some new data for the spike comparison test.

I still kept getting godly spikes with the 16 Axe Mastery... made me wish I was doing this in gvg

, but the damage comparison really approaches a more definite answer

After combining both pools of data (the older tests and the one I just did, 60 trials total), I came to the following results:

Hit Difference: 26.36 ~~ which gets rounded to 26 by Guild Wars

16 Axe Mastery does 1.12 times as much damage as 14 Axe Mastery on spikes.

Captain Robo · Oct 25, 2007, 08:13 PM // 20:13

Quote:

Originally Posted by erk

Good work

Got any stats for a Dev Hammer combo using minor vs superior? Dev hammer is also a very popular spike frontline elite, along with Crip Slash. It would be interesting to see as the hammer has higher base DPS than an Axe, so perhaps the two extra attribute points will have a greater impact overall.

Dev hammer and crip slash aren't true spiking builds in the general sense. Neither sword nor hammer can rip open a three-hit combo that does as much damage, as fast as an axe.

For Sword, you have to go Cripslash-Gash-Final, with your most damaging hit coming from the end of the attack chain; by that time, your target has been protted.

Likewise, with Hammer, the most damage comes from Crushing blow->Mighty Blow, as Devastating Hammer doesn't do any additional damage.

Eviscerate, on the other hand, gets straight to the point: massive damage and you've got a deep wound. Oh wait, more damage on the way with Executioner's and Agonizing.

Sword and Hammer are solid pressure weapons, with swords having consistent DPS, and Hammers having freaking knockdowns, but neither are true spiking weapons, because the majority of their damage comes from the end steps of their spikes; whereas an axe gets down to business quick and early.

I might look into it later, but I'd say toy around with it yourself to get an idea.

Falconer · Oct 25, 2007, 11:18 PM // 23:18

Just a small observation regards the numbers. One item that flat averages like this disregard is the standard deviation or volatility of the damage spread. That's fine if you're doing raw DPS type figures where the nature of the beast is large numbers of individual hits... but a spike hinges around how much potential can it do, can it exceed a certain minimum threshold reliably when used in 2 or 3 hits.

Just for arguments sake, lets say that if the warrior doesn't do at least 200 points on the spike. The spike is guaranteed to fail if I can't meet it. The casters are only adding 400 in spike support. (caster spike support almost always does a fixed amount of damage w/ very little variance)

The 16axe mastery guy, 27 out of 31 rows by my fast count are over 200 total. By comparison only 15 out of 31 rows for the 14 row met this benchmark. That's a pretty big discrepency if it results in successfull kills (87% vs 48%).

Taken from this point of view, the 14 guy is playing it safe, and his kills are 'luck' not skill as is so often bemoaned on these forums.

Phades · Oct 26, 2007, 12:41 AM // 00:41

Quote:

Originally Posted by Falconer

Taken from this point of view, the 14 guy is playing it safe, and his kills are 'luck' not skill as is so often bemoaned on these forums.

This would be assuming the target struck is always at full health, which would be the point of pressure based damage followed up with a spike that warriors are known for. This is ignoring of course physical counter measures, kiting, etc.

DutchSmurf · Oct 26, 2007, 07:20 AM // 07:20

Quote:

Originally Posted by Phades

This would be assuming the target struck is always at full health, which would be the point of pressure based damage followed up with a spike that warriors are known for. This is ignoring of course physical counter measures, kiting, etc.

In the current metagame I think you can count on your target being at full health when you start your spike. Starting a spike after pressuring means you spike after hitting it a couple of times already. Which means it should be well protted and your spike will fail anyway. Unless you are using lots of Shatter Enchantments of course.