Before you invest in Combat Agility/Lightning Reflexes.... [Archive] - Everquest

Tudamorf

05-13-2003, 11:03 PM

...you might want to see these results.

Why waste 27 AA on these skills in theory? While HP used to be the benchmark for survivability, PoP has shifted the focus to damage avoidance with the insanely high max damage amounts. Avoiding even a few hits for 500-800 would be a lot more important than getting a small HP upgrade. I was browsing the enchanter forum and saw several rave reviews about this skill, claiming it transforms casters into much better tanks in PoP. Of course, I don't take anything I read there as gospel, so I ran a test myself.

Starting with Combat Agility 2 on my enchanter, I tested each level against the level 60 mist panther in Wakening Lands. I kept AC/HP/AGI constant for each test. I then logged the total number of attacks (kept around 500 for each test), % normal hits, % normal misses, % dodges (normal hits and kicks dodged), % kicks, and % missed kicks. While the division between normal and special attacks makes it a bit confusing, with percentages not adding up to 100 where you'd expect, I purposely wanted to keep normal and special attacks separate so that the hit % would be directly comparable in each test.

Here are the results:

<table cellspacing=2 cellpadding=2 border=2><tr bgcolor=yellow><td>Skill Level</td><td>n= (samples)</td><td>Avg Hit</td><td>% Hits</td><td>% Misses</td><td>% Dodges</td><td>% Kicks</td><td>% Missed Kicks</td></tr><tr><td>Combat Agility 2</td><td>483</td><td>85.6</td><td>74.8%</td><td>25.2%</td><td>1.7%</td><td>4.3%</td><td>40.0%</td></tr><tr><td>Combat Agility 3</td><td>481</td><td>84.8</td><td>71.7%</td><td>28.3%</td><td>1.5%</td><td>6.7%</td><td>31.9%</td></tr><tr><td>Lightning Reflexes 1</td><td>539</td><td>81.5</td><td>73.3%</td><td>26.7%</td><td>1.7%</td><td>9.5%</td><td>37.0%</td></tr><tr><td>Lightning Reflexes 2</td><td>463</td><td>84.6</td><td>69.9%</td><td>30.1%</td><td>2.4%</td><td>5.8%</td><td>46.0%</td></tr><tr><td>Lightning Reflexes 3</td><td>500</td><td>82.1</td><td>73.0%</td><td>27.0%</td><td>2.6%</td><td>9.4%</td><td>31.9%</td></tr><tr><td>Lightning Reflexes 4</td><td>488</td><td>83.6</td><td>72.2%</td><td>27.8%</td><td>1.6%</td><td>9.6%</td><td>36.5%</td></tr><tr><td>Lightning Reflexes 5
(initial test)</td><td>534</td><td>82.5</td><td>62.7%</td><td>37.3%</td><td>1.3%</td><td>6.4%</td><td>38.2%</td></tr><tr><td>Lightning Reflexes 5
(new test)</td><td>1003</td><td>82.4</td><td>69.8%</td><td>30.2%</td><td>1.7%</td><td>9.1%</td><td>33.6%</td></tr></table>

Normally I'd run a test of significance at this point, but the results up to LR4 are obvious. The skill had absolutely no impact on the hit percentage! With LR5, I saw a small drop in the hit percentage, but the statistical significance is questionable. I would have to run the test again with a larger sample to make sure.

[Edit] I ran a new test on LR5 with double the sample the next day, and posted the results in the table. I also corrected an error in the way the miss % was being shown in the table (the hit % figure was not affected). The low hit percentage on my initial test was likely the result of a streaky RNG during that particular time period, as the new test gives us a percentage close to, if not totally within, the margin of error (which I have yet to calculate).

Overall, if the mist panther data is representative of a typical fight against a PoP-level NPC, then LR5 is a waste of 27 AAXP. While I used my enchanter for the test and not my druid (who is busy on other AAs), I definitely don't see myself buying this skill for the druid, since druid and caster tanking ability are pretty close to one another according to past testing I've seen.

I also wonder about all of the other defensive AAs, since it's entirely possible that you could spend over 100 AAXP on skills that have absolutely no effect on your character. I certainly haven't seen testing on any forum which really shows that they have an effect on a caster or priest. In the meantime, I'd definitely consider steering clear of these skills unless you have nothing better to do with your AA.

Oldoaktree

05-13-2003, 11:08 PM

You just saved me a lot of headache...

Heh I have been saving my 27 to do this on my druid and see if it really mattered, with the exact same goal in mind. Not very far on my saving though, I have found the big number has demotivated me, AA wise.

My gut said that it wouldn't, but others had suggested that as a leather wearer, druids mitigate more like monks (ie it might matter more).

I will mull it over and keep saving for a while.

I was even planning on using the same test mob lol.

Tiane

05-13-2003, 11:15 PM

Just to be clear... the term most melees use when describing the above effect is Damage Avoidance.

The term Damage Mitigation is used to describe the effect that high AC or aa's like combat stability have to reduce the amount of damage you take when you get hit.

Druids have nowhere near the damage avoidance of monks, the skill caps arent comparable, nor do we have block. Same for all casters, which is why any of these melee-centric AA's are going to have only a small effect at best, and as demonstrated by Tuda, are really mostly just a waste of AA's.

Tia

Oldoaktree

05-14-2003, 12:05 AM

The arguments I have seen have basically stated that leather armor mitigates differently than plate, and is more about the avoidance piece.

I will be honest. I find the distinctions that are drawn between the different dmg related issues very hard to pin down. They seem to all operate in the same arena, and it often sounds arbitrary to me to allocate one benefit to one box versus another.

Combat stability for instance, I have been told, should not really be thought of as an adjustment to AC. But it has exactly the same impact that an adjustment to AC would.

Sounds like the same thing again with Combat Agility/LR versus dmg avoidance.

Does this then mean that melee dmg taken is calculated based on at least 4 factors?

* A random chance of avoiding dmg which is unrelated to AC or armor tables (this is where LR would come in)
* An AC related factor that has the same impact (allows you to avoid some of the dmg)
* An AC related factor that lets you reduce the dmg
* A random chance of reduced dmg that is unrelated to AC

Plus perhaps some other effects?

I will say, however, that I found the following quote very telling:

Patch message April 03:

We have evaluated the mitigation abilities of cloth and leather classes, and have made changes as appropriate. For the most part, this change is a reasonable improvement to the ability of monks to absorb damage. At some point in the future we may re-examine the mitigation abilities of chain and plate classes. At this time, and with this final round of changes to cloth and leather classes, we believe that something very close to the appropriate balance is in place.

I think it is important to note that leather classes are treated as a group.

I know we are on the priest dmg table (whatever that really means)...but this quote suggests to me that we really do mitigate differently than clerics or shaman do.

Tiane

05-14-2003, 01:40 AM

Each class has a mitigation value that is independant of the type of armour they wear. This was most clearly demonstrated after the Monk Nerf, when Beastlords remained unaffected, despite having available nearly identical gear and no weight limit.

If you want an actual formula, I think they've worked out a lot of it over on Steel Warrior. Last I looked they had some pretty decent sample sets. But remember, originally (I'm told) AC was divided into 2 values, 1 for avoidance and 1 for mitigation/absorption, and they were combined before release for some reason. There's no real way to seperate them, and displayed AC really isnt a meaningful number when it comes down to the avoidance/mitigation calculations. Instead you have to deal with things like raw AC and skill caps and class etc. Also do remember it is a CLASS modifier, not an armour-type modifier, nor is it related to the piece of armour. It's only by class, which for some reason the PoP devs chose to ignore when they itemized PoP....

Most of that is meaningless for casters, as our class modifiers for mitigation are extremely low. Our avoidance is better, comparitively, but still not great. If you're going to put AA into one or the other, avoidance is the way to go, especially in pop. But as Tuda's samples show, its not a very big bang for the buck.

This is all confused by the fact that, technically, both damage absorption and avoidance can be considered "mitigation", but when people who normally talk about such things talk about them, mitigation = the ability to reduce the damage taken when hit, avoidance = the ability to not be hit at all, and displayed AC = a relatively meaningless number that combines both values and a number of other effects.

Tia

Bismuth

05-14-2003, 07:50 AM

Tuda's test actually lends some support to both sides of the ND vs. CA/LR discussion. The somewhat significant spike in misses between LR4 and LR5 shows that the skill may directly check against the mob's ATK rating to determine whether or not successful misses can be granted (as opposed to seeing more marginal results with each increase in the skill).

This could give credibility to CA/LR in PoP because the perceived lower ATK ratings by the planar trash mobs would yield earlier success, where clear results will be visible at a lower level of the skill. As far as I've seen there aren't any extensive tests of druid defense versus PoP mobs, but I'd love to conduct some. It's my assumption, though, that if druid defensive behavior mimics other classes ("proportionally," with our fragility in mind), more significant results from the damage avoidance AAs would be found in PoP.

Prior to the PoP expansion, I would say most druidic folk favored ND3 over CA, and Tuda's test shows that it was probably rightly so. ND3 may remain tempting since even LR5 is pretty feeble against mobs with mid to high ATK levels, but I wouldn't write off the combat avoidance skills at all.

Remi

05-14-2003, 10:05 AM

Bleh! I'm working on LD3 now, thinking it was better to avoid a 500+ hit than to mitigate 2 500+ hits. But, based on this testing, it looks like the odds of my avoiding that second 500+ hit are so low as to not be worth the AA. How disappointing.

Oldoaktree

05-14-2003, 10:05 AM

It isn't really possible to parse this for PoP.

Joe Average druid can only take a few hits from a typical PoP mob. Collecting data on the order that Tuda did (500ish swings per level) would be virtually impossible to achieve.

Better geared druids who could survive more hits also tend to have 200+ aa's and so will also tend to already have purchased the skill.

Tudamorf

05-14-2003, 01:12 PM

Bismuth says: The somewhat significant spike in misses between LR4 and LR5 shows that the skill may directly check against the mob's ATK rating to determine whether or not successful misses can be granted (as opposed to seeing more marginal results with each increase in the skill)

You may want to check my updated data, which places the hit percentage very close to, if not totally within, the margin of error for the test (which I have yet to calculate). The low hit percentage on my initial test was most likely due to a very streaky RNG during the time I ran the test.

This is an interesting hypothesis, however, and could probably be tested by taking two similarly equipped characters, one with no CA/LR and one with LR5, and pitting them against the same slowed PoP NPC.

Oldoaktree says: Joe Average druid can only take a few hits from a typical PoP mob. Collecting data on the order that Tuda did (500ish swings per level) would be virtually impossible to achieve.

Not really, it would just take a lot longer since you'd have to slow the NPC. Against a slowed NPC, I think a druid could survive for the half an hour it would take to collect about 500 samples. And you could simplify the test by just comparing the extremes -- no skill versus LR5 -- on two druids that are similarly equipped.

L1ndara

05-14-2003, 05:46 PM

I noticed a dramatic increase in getting my ass handed to me when I started going up from CS1/CA1. I could reliably count on getting missed about half the time by PoP mobs with CS1/CA1. By the time I got to LR3 I was literally getting hit ALL THE @#%$ TIME. It was very common for PoP mobs to hit me more than 10+ times in a row. I really think something is broken and broken badly. I never got around to it but I'm 99.99% sure that if I took my naked cleric with 0 AAs and parsed her against my druid with over 40 AAs in combat skills with full equipment, my cleric would get hit less.

By the by, the mist panther is a very special case mob. While it is level 60, it has very unusual stats and I'd be pretty careful useing it to test with. Give it a try with a HoH gaurd or a pusling in CoD.

Tudamorf

05-14-2003, 07:14 PM

L1ndara says: By the by, the mist panther is a very special case mob. While it is level 60, it has very unusual stats

What's unusual about it, other than the fact that it has a very low max hit? It seems to connect on the hits about as often as anything else the same level.

Palarran

05-15-2003, 02:03 AM

I wouldn't read too much into these results...it's pretty clear that, unfortunately, 500 swings isn't enough to give a reliable average miss rate (just look at the difference between the two LR5 tests...)

More than anything I think these results suggest a genuine streakiness to the RNG, as opposed to the perceived streakiness of a sequence of truly random numbers.

Oldoaktree

05-15-2003, 07:47 AM

500 should be an ample sample size for this.

And if you need a larger sample set than 500 swings to perceive a difference, it does argue that that the ability is indeed useless on low mitigation classes.

For me, on a typical raid day say in VT, I will get agro through healing somewhere between 3 and 5 times over the course of 4 hours. Each of those times, the mob will swing at me about 4 times before I die (or if it is slowed maybe 8 times and I will live).

So over the course of a 4 hour raid, we are talking about me taking perhaps as many as 35-40 melee hits. If it takes more than 500 for me to be able to perceive any difference, then the skill is virtually useless.

Maody

05-15-2003, 10:53 AM

Well, i have LR 4 now.

I am not into number crunching in EQ (have done enough of that in RL), but i have the *feeling* that i am getting hit much less than without CA/LR.

In BoT i often have several combat rounds after drawing agro, with only 1 or 2 hits. After beeing summond i often dont even need to heal myself now.

I addition i think green yard trash is hitting me MUCH less than ever. Unfortunately this is a side effect i dont like that much because i am often relying on damage shield when hunting greenies and light blues.

I was not able to solo tank The Spire Lord in Mons Letalis 3 month ago. Now its no problem with selfbuffs and selfheal.

Today i survived an Emperor Ssrra summon: miss, miss, miss, hit 900, hit 400, miss, miss, miss. Maybe its only imagination, but its ok for me. :)

Oldoaktree

05-15-2003, 11:41 AM

Going back to my earlier question...

If all mitigation hangs from AC (which in its hidden depths is based on two elements, avoidance and dmg reduction), then:

* CA/LR may be more or less like a saving throw, or an additional check for a miss, and so be fairly constant in value across classes.
OR
* CA/LR may be a % modifier on the avoidance component of your overall AC.

A third option is that CA/LR affect avoidance entirely independently of AC, in which case it is just a flat chance to be missed, and in theory stable across all classes.

I personally lean toward it being tied in some way to AC, rather than an independent calculation. I also think it is more like the second option I listed rather than the first.

The impact of this is that the higher your AC, the better your gain from CA/LR would be. This would naturally make it a greater benefit to melee, and a lesser one to priests, and the least benefit to int casters, at least most of the time.

Now, on types of armor, if in fact, as I have seen people say before, the type of armor you use creates your overall mitigation using different mixes of avoidance and dmg reduction, then someone wearing cloth or leather (more avoidance oriented) will have a greater than normal benefit than someone who wears dmg reduction oriented armor (plate or chain).

I would add that the last time this came up someone had posted some parse results for a warrior who had seen about a 10% increase in misses between CA3 and LR5, iirc (which indeed I may not...I can't seem to find that thread).

At some point I will pore through the threads on melee sites and see if I can edumacate myself better on the latest theories of mitigation calculations. However, I still feel I can reach one conclusion.

The upshot of all this is that I don't really know how projectable the value of this skill really is, especially across classes. Even within a class it may be that the value of the skill is uneven, and may give greater returns if your AC is high (if it is tied to your AC number on a % basis) or even better returns if your AC is relatively low (if it is a universal % saving throw it would be a relatively larger benefit for someone whose AC gives them fewer misses).

I still lean towards repeating the parse that Tuda did on my druid. I still have a billion points to save, but from where I am now even a modest gain would help me. I just die too fast, and my AC is rising only slowly.

Tiane

05-15-2003, 12:35 PM

Oldoak:

<blockquote>Quote:<hr>So over the course of a 4 hour raid, we are talking about me taking perhaps as many as 35-40 melee hits. If it takes more than 500 for me to be able to perceive any difference, then the skill is virtually useless.[/quote]

Is so totally on the money, and you can apply it to nearly any piece of loot, ability or skill in the game.

Tia

Tudamorf

05-15-2003, 01:19 PM

Maody says: I am not into number crunching in EQ (have done enough of that in RL), but i have the *feeling* that i am getting hit much less than without CA/LR.

Two words: Placebo Effect. You believe that the skill is doing something, and therefore you psychologically place more emphasis on the random streaks of misses. This is why scientific trials that rely on subjective perceptions are double-blind, i.e., neither the tester nor the subject knows whether they're in the control group.

In reality, unless you recently installed a CPU and memory into your brain, there is no possible way you can casually perceive a difference of about 10% in the averages between two large data sets. The difference would have to be quite dramatic before you would just casually notice it, like 75-100%, maybe even more. If you doubt this, I will gladly post two strings of 500 numbers, one with an average 10% higher than the other, and see if you can reliably detect which one has the higher average.

Oldoaktree says: I would add that the last time this came up someone had posted some parse results for a warrior who had seen about a 10% increase in misses between CA3 and LR5, iirc (which indeed I may not...I can't seem to find that thread).

That's what I'm wondering. Has ANYONE done any scientifically valid test showing that this skill works for any class? Not "oh I think I get missed more it rox" or "omg I just got missed 5 times", but a good sized data set against a control sample?

The impact of this is that the higher your AC, the better your gain from CA/LR would be. This would naturally make it a greater benefit to melee, and a lesser one to priests, and the least benefit to int casters, at least most of the time.

Not necessarily. The last test I read on this on the enchanter forum (about 1.5 years ago) showed that once you got AC to about 900, cleric and enchanter mitigation was nearly identical against the test subject, which I think was an ilis knight. Below that, at about 800, there was either no difference or a minute one. I can't seem to bring up the thread on a search, though.

My druid already has CA3 from before, perhaps I can pump it up to LR5 and try the same test I ran. Or, if another druid here already has LR5, they can test it against the same panther and report the results. As far as I can tell, he's a static NPC that is always up and spawns at the same level.

Oldoaktree

05-15-2003, 01:27 PM

Part of the quest for my crappy neck item...and you just do a trade with him so he never dies.

Oldoaktree

05-15-2003, 01:31 PM

There was a thread about it here, with a link to Steel Warrior.

With the changed forums for Steel Warrior (I went looking for the thread there) I am not certain if the original SW thread survived, and you can be sure any hyperlink in the thread here would not work.

There was a nice chart for that parse too, at each level.

And I asked a lot of the same questions then that I do now heh.

Weoden

05-15-2003, 01:50 PM

steelwarrior.xwarzone.com/ (http://steelwarrior.xwarzone.com/)

steelwarrior.xwarzone.com...hreadid=13 (http://steelwarrior.xwarzone.com/showthread.php?s=&threadid=13)

top one is steel warriors link, bottom one is an uber mob tanking thread. I'll try to clear up some questions asked based on what warriors accept as correct understanding about ac/mitigation and avoidance.

Oldoaktree

05-15-2003, 02:02 PM

It is one of the most useful ones..

However, I remember very vividly a post that looked a great deal like Tuda's but done for a warrior.

And it was talked about (and possibly the charts for it were reproduced) here.

That is the one I can't find.

Prior to Tuda's post it was the only parse I had seen, and I remember I didn't trust that what was true for a warrior would be true for a druid, in terms of benefit.

Quelm

05-16-2003, 12:47 AM

15 minutes in Wakening Lands

1029 ac, 175 agi, LR5 ID5, no debuffs on the mist panther

eqcompanion provided the following data, apologies if the table messes with browser widths:

<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=3 bgcolor=#17386A><a href=http://eq.castersrealm.com/players/default.asp?SearchString=Quelm&Action=QuickSearch>Quelm</a></td><td colspan=18 bgcolor=#17386A align=right>Table by <a href=http://home.inreach.com/kai/EQCompanion>EQ Companion 0.6 (build 651) </a></td></td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td><td>Dmg Taken</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>slash</td><td>5:29:03</td><td>5:29:03</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>1</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>100%</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>non-melee</td><td>5:29:03</td><td>5:43:24</td><td>0:14:21</td><td>4,823</td><td>100.0%</td><td>5.60</td><td>371</td><td>_</td><td>100.0%</td><td>0.43</td><td>13 (371)</td><td>13 (371)</td><td>13</td><td>67%</td><td>30%</td><td>3%</td><td>_</td><td>_</td><td>32</td><td>_</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>Total:</td><td>5:29:03</td><td>5:43:24</td><td>0:14:21</td><td>4,823</td><td>100.0%</td><td>5.60</td><td>371</td><td>1</td><td>99.7%</td><td>0.43</td><td>13</td><td>13</td><td>13</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>29,914</td></tr></table>
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=21 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Mist+panther'>Mist panther</a></td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td><td>Dmg Taken</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>kick</td><td>5:29:03</td><td>5:43:24</td><td>0:14:21</td><td>5,267</td><td>17.6%</td><td>6.12</td><td>51</td><td>48</td><td>51.5%</td><td>0.11</td><td>108 (2)</td><td>103 (47)</td><td>103</td><td>100%</td><td>_</td><td>_</td><td>_</td><td>_</td><td>87</td><td>_</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>claw</td><td>5:29:03</td><td>5:43:23</td><td>0:14:20</td><td>24,647</td><td>82.4%</td><td>28.66</td><td>320</td><td>222</td><td>59.0%</td><td>0.63</td><td>115 (10)</td><td>58 (97)</td><td>77</td><td>38%</td><td>57%</td><td>5%</td><td>_</td><td>_</td><td>27</td><td>_</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>Total:</td><td>5:29:03</td><td>5:43:24</td><td>0:14:21</td><td>29,914</td><td>100.0%</td><td>34.74</td><td>371</td><td>270</td><td>57.9%</td><td>0.74</td><td>115</td><td>58</td><td>81</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>4,823</td></tr></table>

59% hit rate for normal attacks

Tudamorf

05-16-2003, 12:59 AM

Nice chart. To compare stats, I was working with only a couple of buffs, AC 783 and AGI 114. Now all we need is the druid test with similar stats and no defensive skills to make the comparison. I'll try to get as close as possible to 1029 AC/175 AGI on my druid and run the test, this will be with CA3 only.

I also find it pretty interesting that a druid with 1029 AC, 175 AGI, and ID5 took average damage only 6% less than an enchanter with 783 ACI, 114 AGI, and no damage mitigation (CS/ID) skills. I know damage mitigation is similar with priests and casters, but it really makes you wonder whether ID5 does anything at all.

AmonraSet

05-16-2003, 05:07 AM

My feeling is that the sample size is nothing like large enough to detect what are likely to be fairly small reductions in the hit percentage. The very fact that the test was run twice for LR5 and produced results which were 7% different supports this conclusion, when the overall change was maybe 12%.

I think the tests do show that the effect isn’t huge, but I wasn’t really expecting it to be. The description for CA3 states that it is a 10% increase in avoidance (or something like that). Assuming the enchanter without CA/LR avoids 30% of attacks, it would mean with CA3 him now avoiding 33%. If LR1-5 give say 2% each, you would expect to see a 36% avoidance by the end.

As for those who say that any increase is worthless if it takes 500+ hits before you can notice a difference, then I would say the same applies to almost all equipment/AA’s. If I lose one piece of equipment and so lose 50-100 hp then I almost certainly won’t notice a difference on my survivability on raids. If I lose all of my equipment then I certainly will notice the difference. It’s just a matter of many insignificant effects adding up to produce a significant overall effect.

For interest, I ran a quick test myself, generating a 500 random numbers between 0 and 1 and seeing how many were greater than 0.5 (for reference, none came up exactly 0.5). To simulate a small change which you might expect from CA and LR, I did the same test a few more times (with new random numbers), but checking which were greater than 0.48, 0.46, 0.45, 0.44, 0.43, 0.42:

0.50 > 52.4%
0.48 > 51.2%
0.46 > 54.8%
0.45 > 55.0%
0.44 > 55.4%
0.43 > 55.2%
0.42 > 57.0%

Overall there is an increase but its not as big as it should be, and random fluctuations tend to drown out most of the underlying changes.

Romidar

05-16-2003, 06:53 AM

I did not do a controlled study, fighting one mob repeatedly. However, I have been logging and parsing my fights for almost as long as I've been playing.

I'm a paladin, 60th level. You can check my magelo, though my equipment was a bit different when I did these parses (thought it was almost identical between the two parses).

I have two data sets from the same camp (55th to 57th level mobs in Chardok). Here's the avoidance statistics, from Caladel's EQParser.

With no combat skills (~1285 buffed AC):

<blockquote>Quote:<hr>Parse Results -- miss ratios(damage avoidance)
-----------------------------
NPC Total Avoidance: 1480 total misses out of 2938, 50.37%
&nbsp &nbsp &nbsp &nbsp You missing NPC: 1070, 72.30% of total misses, 36.42% of total swings
&nbsp &nbsp &nbsp &nbsp NPC dodging your swings: 128, 8.65% of total misses, 4.36% of total swings
&nbsp &nbsp &nbsp &nbsp NPC Parring your swings: 161, 10.88% of total misses, 5.48% of total swings
&nbsp &nbsp &nbsp &nbsp NPC Riposting your swings: 121, 8.18% of total misses, 4.12% of total swings
Your Total Avoidance: 2306 total misses out of 4790, 48.14%
&nbsp &nbsp &nbsp &nbsp NPC Missing you: 1657, 71.86% of total misses, 34.59% of total swings
&nbsp &nbsp &nbsp &nbsp You Dodging NPCs: 208, 9.02% of total misses, 4.34% of total swings
&nbsp &nbsp &nbsp &nbsp You Parring NPCs: 250, 10.84% of total misses, 5.22% of total swings
&nbsp &nbsp &nbsp &nbsp You Riposting NPCs: 191, 8.28% of total misses, 3.99% of total swings
-----------------------------[/quote]

With Combat Agility 3 (no lightning reflexes):
<blockquote>Quote:<hr>Parse Results -- miss ratios(damage avoidance)
-----------------------------
You total misses: 1347 total misses out of 2668, 50.49%
&nbsp &nbsp &nbsp &nbsp You missing NPC: 1016, 75.43% of total misses, 38.08% of total swings
&nbsp &nbsp &nbsp &nbsp NPC dodging your swings: 94, 6.98% of total misses, 3.52% of total swings
&nbsp &nbsp &nbsp &nbsp NPC Parring your swings: 138, 10.24% of total misses, 5.17% of total swings
&nbsp &nbsp &nbsp &nbsp NPC Riposting your swings: 99, 7.35% of total misses, 3.71% of total swings
NPC total misses(your damage avoidance): 1616 total misses out of 3054, 52.91%
&nbsp &nbsp &nbsp &nbsp NPC Missing you: 1203, 74.44% of total misses, 39.39% of total swings
&nbsp &nbsp &nbsp &nbsp You Dodging NPCs: 131, 8.11% of total misses, 4.29% of total swings
&nbsp &nbsp &nbsp &nbsp You Parring NPCs: 163, 10.09% of total misses, 5.34% of total swings
&nbsp &nbsp &nbsp &nbsp You Riposting NPCs: 119, 7.36% of total misses, 3.90% of total swings
-----------------------------[/quote]

Compare the NPC's total avoidance in the two parses: 50.37% vs. 50.49%. Quite close!

Compare my total avoidance: 48.14% vs. 52.91% (this includes dodges, parries and ripostes)

Now look at the breakdown of avoidance:

Miss: 34.59% vs. 39.39%
Dodge: 4.34% vs. 4.29%
Parry: 5.22% vs. 5.34%
Riposte: 3.9% 3.99%

I should point out, in case it isn't widely known here that defensive skills are checked BEFORE the game determines hit or miss, so you wouldn't expect to see a reduction in dodges, parries and ripostes (which is good, actually - about 4% of my DPS comes from riposting the enemy; if misses were checked first, I'd lose DPS by getting CA3).

What should be pretty obvious is that other than misses, the data are extremely close across the two samples (one of around 4000 swings against me, the other of around 3000 swings against me). I am surprised to see them so close since it was NOT a controlled experiment, although about 80% of the mobs were exactly 55th level.

These results are consistent with what people have posted many times on the paladin boards - combat agility, at least, shows a dramatic improvement in damage avoidance (despite the fact that it is this "large," however, it's still not something you can perceive without parsing).

From what we've been able to tell, the ability works on your BASE chance of avoidance. If a naked 60th level paladin and a naked 60th level druid get hit the same proportion of attempts, then the skill would have the same affect. If, however, the base chance differs, then the total gain from CA would vary as well.

At least outside of PoP, we've concluded that CA is significantly (might even say dramatically) better than CS for a couple of reasons (the AC soft cap is "easily" reached against most non-PoP mobs and most non-PoP mobs - particularly Luclin mobs - have such a small Damage Interval that mitigating from maximum to minimum hit makes a small difference).

I haven't seen any good parses (on the paladin board) of lightning reflexes, but no one has reported an INCREASE in how often they are hit (and there is a pretty large segment of us who parse).

I will say that though I can't SEE that I'm getting hit less, I do know that I need to be healed less. Against these mobs I used to require at least 1 complete heal every fight. After I got CA3, though, I could get by with one complete heal every other fight (that is, my own healing could keep up) much of the time. (Getting hit less means getting stunned less which means more DPS, faster fights, etc.)

Quelm

05-16-2003, 07:38 AM

Round 2:

same stats, 1029 ac 175 agi

<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=3 bgcolor=#17386A><a href=http://eq.castersrealm.com/players/default.asp?SearchString=Quelm&Action=QuickSearch>Quelm</a></td><td colspan=18 bgcolor=#17386A align=right>Table by <a href=http://home.inreach.com/kai/EQCompanion>EQ Companion 0.6 (build 651) </a></td></td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td><td>Dmg Taken</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>slash</td><td>11:45:29</td><td>11:45:29</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>1</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>100%</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>non-melee</td><td>11:45:31</td><td>11:58:36</td><td>0:13:05</td><td>4,238</td><td>100.0%</td><td>5.40</td><td>326</td><td>_</td><td>100.0%</td><td>0.42</td><td>13 (326)</td><td>13 (326)</td><td>13</td><td>72%</td><td>25%</td><td>3%</td><td>_</td><td>_</td><td>31</td><td>_</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>Total:</td><td>11:45:29</td><td>11:58:36</td><td>0:13:07</td><td>4,238</td><td>100.0%</td><td>5.39</td><td>326</td><td>1</td><td>99.7%</td><td>0.42</td><td>13</td><td>13</td><td>13</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>26,225</td></tr></table>
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=21 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Mist+panther'>Mist panther</a></td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td><td>Dmg Taken</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>kick</td><td>11:45:29</td><td>11:58:34</td><td>0:13:05</td><td>4,635</td><td>17.7%</td><td>5.90</td><td>45</td><td>48</td><td>48.4%</td><td>0.12</td><td>103 (45)</td><td>103 (45)</td><td>103</td><td>100%</td><td>_</td><td>_</td><td>_</td><td>_</td><td>84</td><td>_</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>claw</td><td>11:45:29</td><td>11:58:36</td><td>0:13:07</td><td>21,590</td><td>82.3%</td><td>27.43</td><td>281</td><td>240</td><td>53.9%</td><td>0.66</td><td>115 (14)</td><td>58 (88)</td><td>77</td><td>32%</td><td>62%</td><td>6%</td><td>_</td><td>_</td><td>26</td><td>_</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>Total:</td><td>11:45:29</td><td>11:58:36</td><td>0:13:07</td><td>26,225</td><td>100.0%</td><td>33.32</td><td>326</td><td>288</td><td>53.1%</td><td>0.78</td><td>115</td><td>58</td><td>80</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>4,238</td></tr></table>

Surprisingly enough, the second trial resulted in an even lower hit percentage for the mist panther. 53.9% of normal attacks connected.

I also ran a test using Hand of Ro and Illumination. For the test, I rooted the mist panther, applied Hand and Illumination before engaging, and kept the debuffs on for the duration of the fight.

<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=3 bgcolor=#17386A><a href=http://eq.castersrealm.com/players/default.asp?SearchString=Quelm&Action=QuickSearch>Quelm</a></td><td colspan=18 bgcolor=#17386A align=right>Table by <a href=http://home.inreach.com/kai/EQCompanion>EQ Companion 0.6 (build 651) </a></td></td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td><td>Dmg Taken</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>non-melee</td><td>12:13:19</td><td>12:26:32</td><td>0:13:13</td><td>4,355</td><td>100.0%</td><td>5.49</td><td>335</td><td>_</td><td>100.0%</td><td>0.42</td><td>13 (335)</td><td>13 (335)</td><td>13</td><td>73%</td><td>22%</td><td>5%</td><td>_</td><td>_</td><td>31</td><td>_</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>Total:</td><td>12:13:19</td><td>12:26:32</td><td>0:13:13</td><td>4,355</td><td>100.0%</td><td>5.49</td><td>335</td><td>_</td><td>100.0%</td><td>0.42</td><td>13</td><td>13</td><td>13</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>24,671</td></tr></table>
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=21 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Mist+panther'>Mist panther</a></td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td><td>Dmg Taken</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>kick</td><td>12:13:27</td><td>12:26:33</td><td>0:13:06</td><td>5,562</td><td>22.5%</td><td>7.08</td><td>54</td><td>39</td><td>58.1%</td><td>0.12</td><td>103 (54)</td><td>103 (54)</td><td>103</td><td>100%</td><td>_</td><td>_</td><td>_</td><td>_</td><td>85</td><td>_</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>claw</td><td>12:13:19</td><td>12:26:37</td><td>0:13:18</td><td>19,109</td><td>77.5%</td><td>23.95</td><td>281</td><td>234</td><td>54.6%</td><td>0.65</td><td>109 (1)</td><td>58 (152)</td><td>68</td><td>40%</td><td>50%</td><td>10%</td><td>_</td><td>_</td><td>26</td><td>_</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>Total:</td><td>12:13:19</td><td>12:26:37</td><td>0:13:18</td><td>24,671</td><td>100.0%</td><td>30.92</td><td>335</td><td>273</td><td>55.1%</td><td>0.76</td><td>109</td><td>58</td><td>74</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>_</td><td>4,355</td></tr></table>

The hit percentage is higher than the results from round 2, but the average hit is lower. Interestingly enough, the mist panther never clawed for the true maximum value (115). Instead it hit for 109 once in the 608 swings of the third test.

Oldoaktree

05-16-2003, 07:43 AM

Over a very large data set, it is true that this will help.

However, a druids best survival strategy is and has always been to not get agro to begin with.

At it's simplest, these skills will not be the difference between life and death 90-99% of the time for a druid getting agro. Will missing one hit out of 10 save your life if 4 hits kills you? Remember, that doesn't mean the mob won't just swing a 5th time, since you will by then also have low health.

Romidar's parse does suggest that the results are linked to AC, since, in simple rankings, a 785 ac chanter saw less effect than a 1000ac druid, who saw less effect than a hmm 1200ac paladin.

Tuda's original point seems very valid. These skills really should not be a high priority for a druid. I am sorry but I just don't see that perhaps missing one hit out of every 10 when I might get hit 40 times a day is going to be urgent. The poster above who said that all these gains are incremental and only important in aggregate is correct, but some skills give bigger increments than others.

This is about prioritzation, and unless this skill dramatically shifted your propensity to be hit over a small data set, it is unlikely to yield a measureable benefit for a druid. Compare that to something like say Dire Charm (which I also don't have yet) that lets you add tremendously to your DPS in a group, or AHA3 which increases each and every heal you cast (more important for HPS than HPM in my opinion).

I am still torn about doing the test myself so I can measure the results on a druid. But if the streakiness of the game's random number generator is greater than the measurable affect of the skill over small data sets, logically this can't be that great for us.

I would give up this ability in a heartbeat to have Spell Casting Subtlety instead, though.

On the points about Ro's above, this makes perfect sense to me. The melees have derived that DB+xDI formula for melee dmg, where X is a function of ATK. Debuffing ATK shifts the dmg down, and makes it less likely for a mob to land its max hit, and would not affect misses.

Romidar

05-16-2003, 08:16 AM

<blockquote>Quote:<hr>Tuda's original point seems very valid. These skills really should not be a high priority for a druid. I am sorry but I just don't see that perhaps missing one hit out of every 10 when I might get hit 40 times a day is going to be urgent. [/quote]

I am in complete agreement that there's very little bang for the buck from these skills for someone who isn't getting hit a LOT. They are good for helping longterm effeciency, but it would have to be a HUGE long term for someone who isn't getting hit hundreds of times in an encounter. For a tank, completely avoiding an extra 2-3 blows out of 100 is a great thing - rarely life saving, but it doesn't take many hours for me to take on the order of thousands of blows; at that rate, it doesn't take long to save an entire health bar.

My druid probably won't ever get these skills - it would be VERY unusual for them to save his life (well, maybe he might be able to get off an evac on a rare occasion when he wouldn't have without the skill), and I'm more concerned with overall power and effeciency. Of course the other problem is that druids just do not seem to have many really appealing AA skills to choose from (coming from a hybrid who has a huge pool to choose from but must get many more AA points to maximize effeciency - not that that is really bad!).

iegil

05-16-2003, 09:14 AM

What I've noticed in POP since buying LR5.

I don't know how the rest of the druids out there play their toons but I routinely push the edge of the envelope with regard to aggro.

What this means is that I routinely get summoned when the mobs are under 2-3 percent health. At this point the mob will usually have 4 to 8 swings at me still. Thats all I have to live through. Pre-LR5 I typically died to this action. Since I bought LR5 the mobs tend to miss or not hit for max nearly as much.

I would propose 2 things to everyone with regard to this.

1) The utility of this skill is more of a bell curve with mobs at the appropriate attack values missing more than the ones that are much higher or lower.

2) That it does offer superior mitigation without outright avoidance also on this curve. IE You mitigate more versus Fennin Ro or Blazzax the Omnifiend than you do against an orc pawn in Wakening Lands.

Why I think this? Before I had LR5, Blazzax, Fennin and mobs at that level routinely hit me for 2-3 max damage swings and I was dead. Now Blazzax takes 4 swings and Fennin takes 6+

Of course, my sample size is small. But go figure when the mob hits for 3540 unmitigated how many samples you get.

Either way, I still get summoned 4-5 times a raid, and need 1-2 rezz's now instead of 4-5.

Quelm

05-16-2003, 09:57 AM

<blockquote>Quote:<hr>At it's simplest, these skills will not be the difference between life and death 90-99% of the time.[/quote]

Mana Preservation doesn't help with 90-99% of the mana cost of spells cast, yet it is probably the most important focus effect for a caster because of the 1-10% of mana that *is* saved with each cast. A 1-10% difference helps, especially when compounded with other, similar benefits.

I've chosen to pick up gear with extra ac/hps, gone for ND3, LR5 and ID5, and it takes a lot more to kill me than it used to. Giants in storms, yellow and red cons in tactics, and angry pets have all taken turns trying to kill me, and more often than not, there's time to run away or land a root or channel a heal, catch a patch heal, or at least exodus.

I would like to see a test with a PoP mob, but the sample size would be smaller without a dedicated healer or slower.

Oldoaktree

05-16-2003, 10:16 AM

Is that druids cast as their primary activity at a raid, whether it be healing or nuking. Saving 10% on your casts all the time is a very dramatic benefit, because it is always being used.

Batou062671

05-16-2003, 12:43 PM

<blockquote>Quote:<hr>Mana Preservation doesn't help with 90-99% of the mana cost of spells cast, yet it is probably the most important focus effect for a caster because of the 1-10% of mana that *is* saved with each cast.[/quote]
Yes, exactly why it's so much better than LR5. EVERY cast gets a benfit. LR5 seems to slightly reduce the chance that you might get hit. It's results are not consistent and even with large datasets, it's hard to pick out if it has a deterministic effect. Since MP3 gives mana savings on every cast, it's effects are readily apparent; LR5 on the other hand, does not allow you to see an effect every fight, and even with parsing you're not sure if it's really doing anything.

Weoden

05-16-2003, 12:58 PM

Hi, lost the thread yesterday but found it today. Sorry for taking as long as I have to post. Anyway, I probably should cover some basics that everyone knows.

ac = reduces/increases average damage taken.
attack= reduces/increases average damage done.
agi = increases ac.
dodge/defense/parry = all based on your skill rating. Defense raises ac.

k, damage taken/done is broken down into 20 distinct numbers where there is a minimum possible number and maxium possible number and there is an increment between each number.

example: min damage is 10 and max is 100 but the increment is 1. you might see 50, 51... 69.

If your ac went up you might see 40, 41... 59.

If your ac went up a lot then you might see 11, 11, 11 11, 12,13 ... 36.

or if your were nakid then 80, 81, 82... 100

the point is that ac will shift where the distribution of these 20 possibly numbers are. Each number occurs randomly but there are only 20 possible numbers.

Attack:
You can raise and lower where this distribution occurs by raising or lower your/the monsters attack. The effect can be quite dramatic over long fights.

Miscellaneous: defense seems to raise ac so that can be an easy method of raising your ac.

On the point of misses. Level seems to be directly related to a PC's max damage but im not sure how misses are calculated exactly. I would guess with dodge skill and that is capped pretty low for a druid or your level relative to the creatures.

CA/LR On my warrior, I notice a difinite difference when I had cs3 and lr5 but I am taking sustained damage. I just notice clerics needing fewer heals to heal me during a fight.

Now a suggestion. Get hit points and ND3. Hit points items work with ND. The same is true with sta. Both are modified by nd and if you read the war board they say go for nd first however wars are getting ch'ed.

I hope this answers some questions.

PS When I was lv 65 war in pov tanking golems, It would take 2 ch's but once I got nd3 + ca/lr/cs/(forgot the other one) I can tank mobs unslowed or require one heal per battle and thats it.

Tudamorf

05-16-2003, 12:59 PM

To compare to Quelm's data, I tested with my druid with CA3, and got my stats fairly close to his -- 997 AC, 170 AGI. These are my results:

<table cellspacing=2 cellpadding=2 border=2><tr bgcolor=yellow><td>Skill Level</td><td>n= (samples)</td><td>Avg Hit</td><td>% Hits</td><td>% Misses</td><td>% Dodges</td><td>% Kicks</td><td>% Missed Kicks</td></tr><tr><td>Combat Agility 3
(DRU)</td><td>688</td><td>81.6</td><td>63.2%</td><td>36.8%</td><td>1.6%</td><td>8.3%</td><td>44.7%</td></tr></table>

Amonraset says: Overall there is an increase but its not as big as it should be, and random fluctuations tend to drown out most of the underlying changes.

Well, you are comparing apples and oranges, since you were probably testing a random distribution with nearly independent trials and I was testing a binomial distribution without independent results.

In any event, you are highlighting the problem. If the data doesn't rise above the random noise, then we can't say that the skill had any effect whatsoever, and there's no point in buying it.

Oldoaktree says: Will missing one hit out of 10 save your life if 4 hits kills you?

The real issue is, if you can spend AA on any defensive skill (after you've finished your core casting skills), what's the best one?

If defensive skill X causes you to miss 1 more hit out of 10, then getting skill X is as about attractive as getting 10% more hit points in terms of survivability. However, if skill X causes you to miss less than 1 more hit out of 100, then you're better off eating a Halas 10lb meat pie.

As mentioned here, for tanks, the equation is different -- they care more about the average damage reduction over time than survivability in an attack, so even a few percentage points can be worthwhile. This is not important for a druid, however.

From the data I see here, my best recommendations for druid defensive skills are ND3 (12 AA, gives you a few hundred HP) and +STA. CA and LR don't even seem to be relevant (thus far).

That brings us to the AA description. The poor druid reading it would never realize how he's being short changed. Combat Agility says it "increases melee damage mitigation by 2, 5, and 10 percent," for 2/4/6 AAXP. That sure sounds like I'll get missed 10% more if I spend 12 AAXP. Lightning reflexes says it "further increases your chance of completely avoiding incoming melee damage," and costs 15 AAXP for 5 levels. Extrapolating from Combat Agility, I'd expect it to get me missed about 22.5% more total.

Getting missed 22.5% more for 27 AAXP sounds fair. Getting missed for exactly the same amount (or for a lesser amount that's not even detectable with extensive testing) for 27 AAXP isn't fair at all, and makes the 12 AAXP ND3 far superior. They should either lower the AAXP cost for casters and priests, or just remove the skill entirely and replace it with something worthwhile.

Oldoaktree

05-16-2003, 03:12 PM

is that I am looking to the AA skills and trying to determine if there are any that materially help me survive the agro that I regularly incur due to our high agro spells.

In a clear to one of the Zek twins, I find myself chain casting my pitiful CB since NI and TR seem like instant death. I reserve those spells for the ocassions when like a Pally I might sacrifice my own life for the good of the raid.

I think my decision is pretty much made and I will continue to ignore all the melee related AA for another 100 aaxp or so whlie I max out the various casting skills.

Since my only real agro control will continue to be 'cast less frequently' with or without the AA, there is no pressing need to pursue it for the foreseaable future.

I appreciate everyone sharing the data. I think I will spend my small bank of points tonight, perhaps to buy up my mana pool since I feel the most regret about wasted mana pool potential at the moment, rather than the next increments on my heal or nuke abilities.

Quelm

05-16-2003, 09:45 PM

Combining my two first tests, we see 601 normal hits and 462 misses for 1063 total, 56.5% hit rate, 43.5% miss rate.

Compared to Tuda's druid, that's a fairly substantial increase in misses. (43.5-36.8)/36.8 = 18.2% more misses or, if you prefer (56.5-63.2)/63.2 = 10.6% fewer hits, simply by adding 32 ac, 5 agi, and LR5. That, IMO is worth 15 AA. Are the results statistically significant?

Oldoak mentions trips to zek twins. Recently I tagged along on a TZ raid. I used NI quite a bit. At one point against a flayer, we lost a few tanks, and I ended up spam healing a ranger. The direct heals eventually drew aggro, and I tanked the flayer for a few rounds. After 6 hits with only 1 for the max of 770, and 5 misses, I lived and the flayer didn't. I've picked up aggro on the elites, sometimes unslowed, and haven't died yet.

Another pleasant surprise came one night while visiting a friend in PoD. I joined his group in the left castle courtyard for some groupsay and fun nuking. Their warrior left, but the enchanter, cleric and monk were still up for a few more kills. I volunteered to tank for a bit. The beetles hit like sebilis frogs. Even with harmtouch and a resisted slow, the cleric had plenty of time for complete heals. The former puller, a 62 monk with 1200+ ac was in shock, and said so repeatedly. My friend the enchanter, laughed as he tried to out-taunt my debuffs and nukes (yay horses).

I'm not recommending LR/ID over SCM, obviously. I was one of the early "SCM > all" folks, picking SCM3 up with my 18th point. The comparison to Mana Pres. items was simply an illustration of how noticeable a 1-10% improvement could be. I don't think the skills are must-haves for everyone, but I'm not convinced that the skills are useless.

-Q

Oldoaktree

05-16-2003, 11:35 PM

LR5 has no impact on max hits.

Only on misses. Those are very streaky. I have been missed 5x out of 6 on mobs without even CA1.

I am not saying never, but I feel very safe moving this back down the list.

I spent the points I had banked tonight.

I will still be better off just trying to avoid getting agro as I have always done.

Tudamorf

05-17-2003, 01:29 AM

Quelm says: Are the results statistically significant?

Possibly, but I can't say because there's another unknown variable out there affecting the results -- the time-dependent streakiness of EQ's "random" numbers. It's going to be tough to analyze a sample that was captured during one small period. I think several shorter sessions at different times of the day might disperse the effect somewhat and give us a more accurate picture.

Also, remember that our two data sets were on different druids with slightly different stats, equipment, and buffs, which could also skew the results. Ideally the tests should be run on the same character.

The former puller, a 62 monk with 1200+ ac was in shock, and said so repeatedly.

Almost everyone is in shock when they realize the "druids/casters can't tank" propaganda isn't true. Tanking is primarily a function of your level versus the target level.

As for your other examples, they're really not too helpful at understanding the skill. With the actual data so muddy, it's impossible to just draw a conclusion based on your subjective mood during a particular fight.

Then again, I bet some people at this point would just want to take the blue pill and pretend they haven't totally wasted their time buying these AA...

Palarran

05-17-2003, 07:02 AM

Nah, my experiences with Phinigel have convinced me there's a very real effect with LR5.

Now, this data is not very scientific--there are a lot of variables unfortunately, especially related to timing--but I'm convinced there is some significance to this. My unbuffed AC was 878 on Feb 22, 897 on May 9; it was significantly higher for the first few seconds of each fight. I used hand of ro in every fight, but only used ro's illumination in some of the fights. The 2/22 fights were all against the same Phinigel, I didn't succeed until the final fight for that day. I did not gain any defensive AA's except in the combat agility/lightning reflexes line.

Feb 22: (no CA or LR, AC 878 unbuffed)
Hand of Ro+Ro's Illumination
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=20 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Phinigel+autropos'>Phinigel autropos</a> - Damage taken : 7,649</td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>crush</td><td>10:06:30</td><td>10:10:24</td><td>00:03:55</td><td>10,411</td><td>100.0%</td><td>44.30</td><td>114</td><td>50</td><td>69.5%</td><td>0.70</td><td>161 (5)</td><td>60 (36)</td><td>91</td><td>38%</td><td>62%</td><td> </td><td> </td><td> </td><td>24</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>[b]Total:</td><td>10:06:30</td><td>10:10:24</td><td>00:03:55</td><td>10,411</td><td>100.0%</td><td>44.30</td><td>114</td><td>50</td><td>69.5%</td><td>0.70</td><td>161</td><td>60</td><td>91</td><td> </td><td> </td><td> </td><td> </td><td> </td><td> </td></tr></table>

Hand of Ro ONLY
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=20 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Phinigel+autropos'>Phinigel autropos</a> - Damage taken : 9,038</td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>crush</td><td>10:59:10</td><td>11:01:18</td><td>00:02:09</td><td>7,769</td><td>100.0%</td><td>60.22</td><td>73</td><td>36</td><td>67.%</td><td>0.84</td><td>161 (10)</td><td>60 (10)</td><td>106</td><td>32%</td><td>68%</td><td> </td><td> </td><td> </td><td>20</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>[b]Total:</td><td>10:59:10</td><td>11:01:18</td><td>00:02:09</td><td>7,769</td><td>100.0%</td><td>60.22</td><td>73</td><td>36</td><td>67.%</td><td>0.84</td><td>161</td><td>60</td><td>106</td><td> </td><td> </td><td> </td><td> </td><td> </td><td> </td></tr></table>

Hand of Ro + Ro's Illumination
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=20 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Phinigel+autropos'>Phinigel autropos</a> - Damage taken : 9,482</td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>crush</td><td>11:25:30</td><td>11:28:03</td><td>00:02:34</td><td>6,537</td><td>100.0%</td><td>42.45</td><td>74</td><td>44</td><td>62.7%</td><td>0.77</td><td>161 (3)</td><td>60 (30)</td><td>88</td><td>43%</td><td>57%</td><td> </td><td> </td><td> </td><td>21</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>[b]Total:</td><td>11:25:30</td><td>11:28:03</td><td>00:02:34</td><td>6,537</td><td>100.0%</td><td>42.45</td><td>74</td><td>44</td><td>62.7%</td><td>0.77</td><td>161</td><td>60</td><td>88</td><td> </td><td> </td><td> </td><td> </td><td> </td><td> </td></tr></table>

Hand of Ro + Ro's Illumination
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=20 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Phinigel+autropos'>Phinigel autropos</a> - Damage taken : 14,678</td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>crush</td><td>11:49:13</td><td>11:52:08</td><td>00:02:56</td><td>6,266</td><td>100.0%</td><td>35.60</td><td>75</td><td>37</td><td>67.%</td><td>0.64</td><td>156 (1)</td><td>60 (38)</td><td>84</td><td>49%</td><td>51%</td><td> </td><td> </td><td> </td><td>24</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>[b]Total:</td><td>11:49:13</td><td>11:52:08</td><td>00:02:56</td><td>6,266</td><td>100.0%</td><td>35.60</td><td>75</td><td>37</td><td>67.%</td><td>0.64</td><td>156</td><td>60</td><td>84</td><td> </td><td> </td><td> </td><td> </td><td> </td><td> </td></tr></table>

And from May 9 (CA3/LR5, AC 897 unbuffed)
Hand of Ro ONLY
<table width = '100%' border='1' cellspacing='0' cellpadding='3' bordercolor='#0000ff' bordercolorlight='#000000' bordercolordark='#ffffff' frame='border' rules='all'> <tr> <td colspan=20 bgcolor=#800000><a href='http://eqbeastiary.allakhazam.com/search.shtml?name=Phinigel+autropos'>Phinigel autropos</a> - Damage taken : 12,648</td></tr> <tr bgcolor=#cbbfa5 style='color: #000000;'> <td>Attack</td><td>Start</td><td>End</td><td>Duration</td><td>Dmg</td><td>Dmg%</td><td>DPS</td><td>Hit</td><td>Miss</td><td>Hit%</td><td>HPS</td><td>MaxH</td><td>MinH</td><td>AvgH</td><td>% 1</td><td>% 2</td><td>% 3</td><td>% 4</td><td>% 4+</td><td>DLY</td></tr> <tr bgcolor=#f9fac0 style='color: #000000;'> <td>crush</td><td>07:57:03</td><td>07:59:30</td><td>00:02:28</td><td>5,371</td><td>100.0%</td><td>36.29</td><td>57</td><td>59</td><td>49.1%</td><td>0.78</td><td>161 (1)</td><td>60 (14)</td><td>94</td><td>36%</td><td>63%</td><td>1%</td><td> </td><td> </td><td>21</td></tr> <tr bgcolor=#dddddd style='color: #000000;'> <td>[b]Total:</td><td>07:57:03</td><td>07:59:30</td><td>00:02:28</td><td>5,371</td><td>100.0%</td><td>36.29</td><td>57</td><td>59</td><td>49.1%</td><td>0.78</td><td>161</td><td>60</td><td>94</td><td> </td><td> </td><td> </td><td> </td><td> </td><td> </td></tr></table>

I'll add more LR5 results if people want (I'm missing a couple recent logs for some reason, which were similar to the fight I posted--the ~50% miss rate did not appear to be a fluke--but I'll go fight Phinigel again). The point of this data was not so much to give some hard numbers on LR5, but to show why some people like me believe LR5 has value. An increase in miss percentage from ~33% to ~50% is very significant and noticeable.

Weoden

05-17-2003, 07:28 AM

I am 99% sure that % hit/miss is not determined by ac or attack so debuffing either won't change % hit/miss but will lower/raise average damage taken.

Palarran

05-17-2003, 08:30 AM

Me too, but I included the information on that 1% chance that it would make a difference.

Actually, there's a reasonable chance that the AC bonus from AGI goes into avoidance AC rather than mitigation AC...unfortunately I haven't kept track of my AGI.

Quelm

05-17-2003, 09:11 AM

Weoden said:<blockquote>Quote:<hr>
k, damage taken/done is broken down into 20 distinct numbers where there is a minimum possible number and maxium possible number and there is an increment between each number.

example: min damage is 10 and max is 100 but the increment is 1. you might see 50, 51... 69.

If your ac went up you might see 40, 41... 59.

If your ac went up a lot then you might see 11, 11, 11 11, 12,13 ... 36.

or if your were nakid then 80, 81, 82... 100

the point is that ac will shift where the distribution of these 20 possibly numbers are. Each number occurs randomly but there are only 20 possible numbers.
[/quote]

As I understand it, the only things that can change the actual damage values for a given mob are runes and Defensive Discipline. Otherwise, mob damage (for a given attack type) comes in the form (Damage Bonus + random(1, 20)*Damage Interval). This gives 20 discrete values, with the minimum hit being DB+DI and the maximum hit being DB+20*DI.

Characters with low AC relative to some function of mob ATK, level and the price of gas in San Diego will take a lot of hits for maximum or near maximum damage. The mode is one type of average, the most frequently observed value in a distribution. The modal hit value for an ill-equipped character up against a god might be DB+20*DI. Taking "mode 20" hits hurts.

A higher AC character, especially one of the tank classes, will see more hits for minimum damage. A great tank in PoV may see a distribution with a mode of 3, and rarely ever get hit for maximum damage.

ATK debuffs, AC, and Combat Stability and Innate Defense all factor in to the distribution of hits. In my third trial, the mist panther did less damage when debuffed, and didn't even hit for maximum damage once in 13+ minutes unslowed. Also, I wasn't implying that LR has an effect on max hits, that's the realm of AC and points in CS & ID.

Tudamorf said:<blockquote>Quote:<hr>Tanking is primarily a function of your level versus the target level.[/quote]

I agree that level plays a part, but I think it is fairly clear that AC is important as well. Weoden's experiences in xp groups indicate that points in CA, LR, CS and ID make a difference as well. My experience, many hours spent in Tactics with pets and unslowed mobs, tells me that LR/ID help. These are random samples, at all hours of the day, same character, same gear, and I don't take as much damage. The individual encounters are too brief to merit close analysis, but the end results have been far fewer deaths and a less stressful gaming experience.

Tudamorf

05-17-2003, 11:48 AM

Quelm says: Weoden's experiences in xp groups indicate that points in CA, LR, CS and ID make a difference as well. My experience, many hours spent in Tactics with pets and unslowed mobs, tells me that LR/ID help.

There could be many reasons why you survive longer. The most important is your experience with the zone -- the better you get, the more you know how to avoid dying and the easier the zone seems, even though your equipment is identical. This doesn't necessarily have anything to do with defensive skills.

I just don't put any stock in subjective "feelings" about small variances in large data sets (including my own), because it's impossible for us to accurately perceive it. I lost count of how many tanks during the Kunark era insisted that higher level haste spells gave them more haste (which I proved to be false through testing), or of how many people early in final insisted that stats are the most important aspect of any character (which I also proved to be false through testing). People will believe whatever they want to believe. We need objective testing to really get at the truth.

Maody

05-19-2003, 12:50 AM

<blockquote>Quote:<hr>Two words: Placebo Effect. You believe that the skill is doing something, and therefore you psychologically place more emphasis on the random streaks of misses.[/quote]

Correct.

Maybe you missed my point. I wrote that i have had enough of number crunching in my life. I did it all: quantitative and qualitative research, sample design, scales design, correlation-, regression- and path analysis, propabilty math,parameter estimation and so on.

Thats why i regret to do so in EQ. Its a fantasy game played in my spare time. I dont care if a mob hit me average 240.7 or 241.3. There is no need to turn the game into a science with statistically valid measuring of game rules. Actually i think all those efficency calculations are destroying alot of the game flavor and it is kind of useless to spent time on measuring things who are only one programmers key stroke (or next patch, next expansion) away from becoming obsolete.

But if you like to do so, its ok for me.

I am just adding to this thread that with >300 days played, i know the *feeling* to play my char and i am *feeling* when something has changed. I would still advise that LR5 is a good investment for Druids after you got all the "basic" PoP AA Skills and for my gameplay it values higher than enhanced root or wrath of wild. But again, thats only my personal opinion and not a topic to write a phd thesis on. ;)

Gohbhined

05-20-2003, 08:18 AM

"Two words: Placebo Effect. "

No such effect. Read Scientific American...last year reported on a large study that there was no such thing as a "placebo effect", and any studies that use a placebo group in their studies are wasting their time.

TeriMoon

05-20-2003, 11:21 AM

This has been a very interesting discussion.

Tudamorf

05-20-2003, 02:24 PM

Gohbhined says: No such effect.

I'm not familiar with the article to which you are referring, but telling a researcher that the placebo effect doesn't exist is like telling a physicist that the atom doesn't exist. The psychological tendency to believe what one wants to believe (when faced with ambiguous or random data) is not only an axiom of science, but is also present in all of recorded human history.

Weoden

05-20-2003, 07:26 PM

There is a current thread on steel warriors about CA. Apparently you can avoid the "natural" proc of the mob with CA. In other words if a mob is using a ssoy and it procs you cant avoid that damage but an innnate proc you apparently can.

Maody

05-21-2003, 12:16 AM

Are you able to post the mentioned articles headline, author or the issue of the journal?

Scientific American isnt the common breakfast paper overseas ;)

Romidar

05-21-2003, 06:46 AM

Perhaps you misunderstood what you read. The scientific american article that you're referring to (I think) is about the fact that the placebo effect is "all in the mind." That is, there are neurological correlates to the placebo effect.

The reason for including a control group isn't to observe a placebo effect - it's so that you have a comparison group to determine statistical significance of the hypothesized treatment effect. That is, if you want to determine whether a drug has an effect, you need to give one group the drug and the other group an injection of saline. If you want to see if there's a placebo effect you can either:

a. measure your dependent variable before doing the test and see if you get improvement in both groups (but hopefully more in the treatement group).
b. have 3 groups - one that gets no treatment at all, one that gets the injection of saline and one that gets the real drug.

Note that the first design is flawed (other things could have happened just in the passage of time that influences your DV - e.g., the people could just have "healed" naturally so you get some improvement in the "placebo" group). The third group (no treatment) in design b allows you to determine if there's a statistically significant impact of just getting the shot as opposed to the passage of time. Little or no research includes such a group because the hypothesis being tested involves the drug, not the placebo effect.

It would be ludicrious for anyone to conclude that there is no such thing as the placebo effect - it is found in multiple fields of research involving people (medicine, psychology, etc.). Many early studies with human beings were discredited because they didn't include such a control group. The existence of placebo effect has usually been argued to show that either:

1. the human mind (belief, faith, whatever) impacts physiology or
2. any sort of self-report of symptoms, feelings, perceptions etc. is suspect. For example, an easy way to show a placebo effect is to examine an analgesic. Just giving someone a pill can be enough to make them REPORT feeling less pain, even if it's a sugar pill. Did they're pain "really" lessen? That's an ontological question - pain is "in the mind" so if the perception of pain decreased, then the pain decreased.

I was also one of the people who long ago showed that stats had little impact on performance (my focus was on strength and dexterity, something paladins believed had an impact on accuracy and damage). We STILL get people who show up on the boards and will swear up and down that they are more accurate with higher dexterity.

Of course you don't have to crunch numbers to play the game, but please keep in mind that "feelings" or "personal experience" are correct when it comes to any sort of semi-random event as often as a broken clock tells the correct time. If it weren't for people who were willing to test and crunch the numbers, people would still be wearing gold black pearl chokers and serpentine bracers as good equipment. ;)

iegil

05-22-2003, 02:52 AM

Phinigel autropos - these parse's are more consistent with what I've seen since buying LR5.

A lot more misses. I truly don't care what a "bugged" panther in wakening lands parses. Go parse a real mob. Get a cleric and tank a mob in Halls of Honor with/without it.

As long as your parsing "bugged" Velious era content, I doubt you'll see the "affect" your expecting from this AA.

Tudamorf

05-22-2003, 07:24 PM

iegil says: Phinigel autropos - these parse's are more consistent with what I've seen since buying LR5. A lot more misses. I truly don't care what a "bugged" panther in wakening lands parses. Go parse a real mob.

So a level 52 old world NPC is "real" but a level 60 Velious era NPC is "bugged"? No level 65 buying this skill could care less about how much they get hit by some light blue old world NPC. They want to know whether it makes a difference against the level of NPC they typically face.

Palarran

05-23-2003, 10:54 PM

Some level 65 druids do care about that sort of thing...me, for instance. :P

In any case I think his point was that "real" mobs seemed to behave more like the parses from Phinigel suggested, not that Phinigel himself was a "real" mob.

I'm curious, how often does the mist panther hit for max?

Tudamorf

05-24-2003, 12:19 AM

Palarran says: I'm curious, how often does the mist panther hit for max?

For me, 11.2% of the hits were maximum (115).

In any case I think his point was that "real" mobs seemed to behave more like the parses from Phinigel suggested, not that Phinigel himself was a "real" mob.

And my point is that he was begging the question. If we knew how effective the skill was against "real" NPCs, we wouldn't be having this conversation. The fact is, we don't really know (except for the touchy-feely "hey I spent a lot of points so it must r0x0r and hey I got missed 4 times in a row yesterday" types of comments) and we're trying to find out.

Callahad

05-26-2003, 07:03 AM

*emerges from the shadows*

Interesting thread =)

I would like to say that every little thing that can make a difference is good. In EQ I see how amazing even small differences become significant. I heard talk of the RZ encounter... Well, if you survived aggro once in 5 encounters (instead of none), and every caster had LR5 on the same raid and as such did the same, don't you think that would be significant? I agree you are not supposed to get aggro, but as you know, stuff happens. And having some leeway to recover from stuff is nice, don't you agree?

How many times have I heard hp > all?? Well, if that's the case, even a total 1% decrease in hits taken is significant.

I have ID5, LR5. I got no parses, but I gotta say, if there is a placebo effect, it's rather big. I used to die if a frog in PoS broke charm and hit on me. Not anymore. Now, I usually take the risk if I know I am going to be hit a couple times. This is not once or twice. This is overall, over I don't know how many hits from them (before and after)...

Callahad

iegil

05-26-2003, 02:37 PM

my point was....

I can go tank an orc pawn and show that LR5 has no difference on me due to the orc pawns low attack value.

You chose a mob with a low attack value. (Doesn't do an appropriate amount of damage for his con)

You get skewed results.

You chose a mob with a real attack value. (Does an appropritiate amount of damage for his con)

You get real results.

So tudamorf, please pick a mob that hits you for 300-500 damage range (IE Level 63 ish) and run the test again. Until then your basically dribbling yourself and ruining peoples opinions with misinformation.

BTW, Whats the max hit on phinergal? 161? And you can already see LR5 starting to make a difference?

As far as me, any way you slice it, I'm a better druid with LR5.

I can now tank in PoD/PoN/PoV somewhat in HoH/PoS/BoT.

I've tanked Ragefire/ToV/Sontalek/Yelinak (Tried Dain but died)

AS A DRUID

So Tudamorf you wanna keep parsing mist panthers and providing misinformation, feel free but if you look at the list of stuff I've tanked, I'm sure you can find something to parse that has a real attack number. BTW, most of the mobs there you won't last long on without LR5.

Tudamorf

05-26-2003, 07:35 PM

iegil says: You chose a mob with a low attack value.

Maximum hit value != low ability to hit. If you look at percentage of hits on the mist panther (50-70%), it is pretty close to the PoP stuff. Also, the distribution of hits (i.e., a histogram of hit values) has the typical U-shape you'd see while fighting a level 60. The only real difference I can see is that the maximum hit.

Phinigel by comparison is much lower level, and can barely hit the target. If you look at the last two Phinigel runs, he only hit for maximum ONCE out of over 100 attacks (< 1% of the time). And he hit for minimum a large percentage of the time -- up to 34% on the penultimate test. There's no histogram displayed, but I'd be willing to bet it's skewed towards the minimum hit values. This is all due to Phinigel's low level.

Compare that to the mist panther, who hit for maximum about 11% of the time and minimum about 20% of the time, again due to its much higher level. The mist panther's ability to attack is much greater than Phinigel's, even though his maximum attack amount is lower.

So Tudamorf you wanna keep parsing mist panthers and providing misinformation, feel free but if you look at the list of stuff I've tanked, I'm sure you can find something to parse that has a real attack number.

Given that maximum attack amount is irrelevant to the test (since we're measuring attack distribution), your flame of my methods is totally unwarranted. Other than flaming and pulling unsubstantiated statements out of you-know-where, I don't see your post adding anything useful to this discussion.

Weoden

05-26-2003, 09:28 PM

There are some things I agree with here. First higher level mobs may have a different damage table. Also, pop mobs may have a table and % chance to hit that is higher than previous expansion mobs. So getting a pop mob that hits reasonably might be a bit of a chore.

Next, CA is supposed to allow you to avoid innate procs. So phingel might be good indicator for this aspect of CA but being a lower/old world mob... CA might not have a proper test asside from saying that CA works.

Tiane

05-26-2003, 10:55 PM

That's silly. If the effect is not demonstrable via logged data sets to a statistical relevance (i.e. greater than what would normally be seen by any random set) then for all intents and purposes, the ability has no effect.

As it sits, all the evidence we have shows that it has little to no effect on a druid. The only people denying this offer only anecdotal subjective "evidence." Burden of proof is on the accuser; if you want to disregard Tuda's data, then provide your own data sets rather than "yeah well it felt like I dodged more" opinions.

Tuda's data is in line with everything that has been derived about the combat system in EQ to date. You can believe what you like, but over the long term CA/LR on a druid has only the smallest of effects, the rest is attributable to the streakiness of the EQ random number generator.

Oh and btw, saying you tanked "so-and-so" means nothing, there's an entire raid support mechanism involved. Pre-pop, for fun we had a Magician tank TVK and some other things in VT... he was level 60 and no CA/LR available at the time.

Tia

iegil

05-27-2003, 05:28 AM

Tudamorf, your absolutely right. I brought you no statistical analysis. I don't have a cleric of appropriate level flagged to park behind me in any zone that is worthy of parsing nor do I feel the overwelming desire to collect the data. We raid in excess of 35 hours a week in my guild and thats plenty of time in game for me without getting a cleric to post 60th level.

You call my flame unwarranted. My point is as any entry level statistics student is taught.

"You can manipulate the numbers to prove any point you want by selecting your datasets to support it."

Congratulations, you've succesfully done this.

You've dismissed the Phinergal tests as they are 4 expansions old. I dismiss yours as 3 expansions old (and a bugged mob, you've all admitted it doesn't hit for as much damage as it should for its level) The pack of lemmings doing these parses says thats a respected mob for testing. Your still lemmings.

I challenge you to parse a POP mob that has a real attack value, and whatever values are appropriate for its level. You call it a flame. You stand by your original numbers. Go parse something in Halls of Honor, Tactics or the Plane of Valor with LR5 and I'll agree with your numbers for this skills usefullness.

Until then, I believe your spreading bad information from bugged mobs.

Respectfully,
Iegil Eyriewind
Fury's Edge

AmonraSet

05-27-2003, 06:03 AM

The real problem with the numbers is that the tests themselves dont really provide dont provide much information on the real question of how much affect CA/LR have on PoP mobs.

The tests show that CA/LR have a minimal effect on haze panthers, and a more significant effect on Phinigels, but I doubt anyone really cares how much effect they have when tanking those mobs.

What most people are interested in are what affect CA/LR have on PoP level mobs. Unfortunately this can only be tested on PoP mobs, which hit rather hard to easily get much data from.

Its rather like wanting to know whether Europeans prefer Pepsi or Coca Cola. Its easy to go out onto the street, find some Americans and ask them, then try to apply those results to Europeans. Unfortunately the test may not be much use. The only reliable way to do it would be to fly to Europe and ask people there but thats much more difficult to do.

Quelm

05-27-2003, 09:04 AM

<blockquote>Quote:<hr>As it sits, all the evidence we have shows that it has little to no effect on a druid. [/quote]

No, as it sits, there's about a 10% decrease in mist panther hits on a druid with LR5, as compared to one with 32 less AC and CA3. My test ran for 30 minutes of unslowed hits, or about 1k swings. Since these tests, there hasn't been any more data, and the conclusion was,"Well, the EQ random number generator is streaky, I'm not sure I trust those numbers." This is the same technique used in the very first post to discard data that didn't fit well with the hypothesis of the tester.

Yes, a larger sample size would be better. Yes, it would be interesting to see the test run on a PoP mob. All we have so far is some data showing fewer hits with LR5 than without, and a LOT of speculation.

Tudamorf

05-27-2003, 04:07 PM

iegil says: You've dismissed the Phinergal tests as they are 4 expansions old.

I didn't dismiss them. I explained that they are less relevant to the ultimate question because Phinigel is about level 51-52 whereas the mist panther is level 60. To a level 65, level 51-52 is nearly green, whereas level 60 is mid-blue and a good target level for XP gain.

"You can manipulate the numbers to prove any point you want by selecting your datasets to support it."

I did not cherry pick datasets or manipulate data here. Everything I collected I presented in my chart in full. I set out a methodology and followed it exactly for each test, keeping as many variables as I could constant (target, HP, AC, AGI, buffs). Although I went in to the tests guessing there would be a more significant outcome, there is no way I could have subjectively skewed it while collecting the results. In fact, the results were contrary to my expectations.

I challenge you to parse a POP mob that has a real attack value, and whatever values are appropriate for its level.

I'm hoping to do that with my druid, who is now sitting on CA3 only, when I have some spare time to play and collect a few more AA. In the last week I haven't had time to play at all, and I don't anticipate getting the necesary AA for at least a week or two.

I doubt I'll do it in Tier 3, but in Tier 2, probably PoV against a crawler or golem. I'm hoping to collect all the AA first, then do the test against the same crawler or golem to eliminate any variation in NPC level or stats.

As usual, I will welcome any scientific debate as to the meaning of the results or my methodology. However, comments like "you're wrong but I have no evidence to prove it" won't advance the discussion.

Pariah777

05-27-2003, 05:07 PM

I've been following this discussion with interest, as I was looking at what to buy to increase my survivablity against Tier 2 and 3 mobs when (not if) I get hit.

I know, I know, manage agro, I know how to manage agro, but it's not as fun as playing right along the edge of nuking/healing/dotting!

So I have CA3 atm, and working on getting LR 1-5. Seeing how I've already spent 12 AA on this, I'm going to go through with it all, and hope that people who say it doesn't avoid enough hits to be statistically significant have left something out of their calculations, especially seeing that there's no way to get AA back :)

I see both points in the discussion, but unless you have a guild full of number crunchers that giggle like schoolgirls about damage parsing, atk values and agi affecting AC avoidance vs. item AC affecting mitigation, you probably won't convince enough support clerics and healers to spend hours letting Guardians of Marr and Crystalline Golems beat on the poor druid 10,000 times. ( Statistically, what's the number of samples that is necessary to remove most random variables from the considerations? From what I remember of my lab statistics procedures, it's quite high). Also you would have to randomly sample a variety of mobs of the same name, because they can spawn at varying levels, thus also with the assumption the varying levels carry with them a variable atk value and stats.

I wish you all luck in your scientific investigations, and will be more than happy to read all the results that you dutifully post, as this directly concerns my AA decisions. I would humbly suggest that indeed, not many of my encounters are with Phingel or the mist panther these days, and might hope that whoever feels the strongest need to statistically analyze the CA/LR AA ability would direct the majority of their efforts to mobs that are more commonly encountered by your average 65th level PoP druid, such as the aforementioned ones, forest/desert/sea giants, loroks, Diaku warriors/death knights, etc, etc.

Good luck in your work, and I look forward to more test results !

Tuliani

05-28-2003, 04:51 AM

I post here because a good friend asked me whether the damage mitigation or the damage avoidance aa's are worth it:
my answer is a definite YES!!! Mine are all maxed, but the progression has been evident. As I recall, I worked the damage avoidance skills up first, then mitigation.

From my simple observations (no scientific data gathering) the % mitigation and % avoidance on each hit is random. Sometimes a plane of fire mob will summon me and will miss me 5 out of 6 hits (to me that is survivability), other times it seems same mob might hit me for full damage of nearly 600 only 1 out of 6 hits, other hits land damage being as low as 120's (that is survivability also). Hey, sometimes they never miss! But if a hit lands for 123 and not 600, mitigation has to be a good aa! (and sometimes none are mitigated..oh well)

Seems I get summoned so many times, nukes or heals -- we need an agro-reduction aa!!! Some of the other classes get one of these. :) (I'd buy it in a flash)

Oldoaktree

05-28-2003, 08:30 AM

...with no levels of CA or LR.

That is the problem.

Perception of change is not change.

At some point I will still do my parse, but atm I am just going to wait for my AC to go up and work on the AA that will have a more directly measureable benefit - better healing, more nuke crits, faster nuking, etc, etc.

iegil

06-04-2003, 05:19 AM

Tudamorf, I am glad to see your going to parse a Valor mob. Those I have noticed a signifigant difference in since I bought LR5.

Could be the strength of my toon, or just random luck... but I'm not afraid to tank those for 20 seconds or more while the chanter mezz's and recharms. (cept if they are hasted, VQ is a brutal haste)

Iegil

LonelronVZ

06-18-2003, 12:55 PM

I have LR5 and have had it for quite awhile . I noticed a big difference at CR3/LR1 . It has made a difference of me surving summons and "just getting to close" to a mob . Like Tuliani , the aviodence is random . Or works like crits .. you see a bunch in a row , followed by a dry period . I notice this especially fighting the same mobs.

LR5 hasn't been a waste of AA , just an awful lot of AA to spend . If i were to do agian , I'd work up to LR1 then go to CS3 . Then back to LR .

Gwynet Woodsister

06-28-2003, 06:10 AM

Anyone else bothered doing parses?

I got CA1/LR5 in 4 days and I am still not convinced I see a difference. Of course, sometimes I get missed a lot, but it happened before too and I still get hit 5 or 7 times in a row pretty frequently.

The impression I got is that low level mobs miss me a lot, but regular exp mobs still hit me as much.

*edit - sig nerfed. Please see board rules about sig sizes.

Padain Fain

07-03-2003, 04:19 AM

I'm posting on this thread as a 65 Warrior and a 65 Druid.

The evidence collected by the contributors on the Steelwarrior forums indicate the following:

CA3 increases the total number of attacks that miss you by a factor of 10% (I believe).
LR increases the number of attacks that MISS you by aproximately 0.5% per rank.

On PoP exp type mobs a warrior gets hit aproximately 50% of the time. With CA3 this increases to ~55% and with LR5 to ~57.5%. These numbers aren't hard and fast for every fight but over a very large data set they are the averages.

When you get hit by a mob your mitigation AC (i.e. the sum of the AC of your armor) determines the distribution of hits from minimum to maximum damage. The higher your AC the less you get for on average - more lower hits. No matter how high your AC you can still get hit for max however. CS and ID further increase this mitigation effect towards lower hits. I have no solid numbers but the total damage taken is reduced in a similar way to CA/LR.

People who claim they can see the difference between having and not having these AA's are merely seeing a Placebo affect. The only way to see their affect is to parse extensively. The same goes for most AA's that change something by a couple of percent, you cannot see the changes as you fight.

Are they worth having for a Druid? Sure, if you have already completed the more worthy AA's, your PoP Crits, ND3, SoTW, Radiant Cure etc etc, then yes taking less damage is always a good thing. I'll probably get them one day, but I won't even start working on them til I have 180+ AA's for my Druid, as opposed to my Warrior who got them all complete at 100 or so.

<a href="http://www.magelo.com/eq_view_profile.html?num=265387" target="top">Marshall Sarelin Seis`Bua</a>
<a href="http://www.magelo.com/eq_view_profile.html?num=226868" target="top">Venerable Padain Fain</a>
<a href="http://www.legendsoftherealm.com/" target="top">Legends of the Realm</a>
<a href="http://www.innoruuk.com/" target="top">Innoruuk</a>

Gwynet Woodsister

07-03-2003, 10:40 AM

/nod

I believe that it prolly helps a bit from time to time but that it really doesn't make such a difference at all. Maybe I get one more miss from time to time and maybe sometimes it saves my life but really I haven't seen much of a difference.

I don't consider it a waste of AAs as there wasn't so many AAs I still really wanted anyway, but there are for sure some more useful AAs out there.

Weoden

07-18-2003, 05:16 PM

A former guild mate of mine did some testing and he posted CA vs ID on the old warriors board. here is a link:
pub14.ezboard.com/ftheste...=1&stop=20 (http://pub14.ezboard.com/fthesteelwarriorthearena.showMessageRange?topicID= 2742.topic&start=1&stop=20)

Anyway, a summary of what he said was consisted of saying that ID and CA mitigated about the same damage but issues like innate procs and and ae would make CA possibly better than ID.

Sscaley was an mt for much of Ntov and SSra and emporer... up to RZ when the SoB jumped servers.

Rhyae I

07-21-2003, 10:30 AM

I don't spend time parsing, I just see what I've experienced so far.
LR5 makes a difference. And on high end encounters you had better have it.
I live much longer with it than before I had it, including tanking nameds on occasion after being summoned. Or when its your job to keep the chanters standing and the mobs really resent you and decide you need a pounding.
It helps you live long enough for someone to get a heal on you, when before you would drop like a rock.
IMHO it's not a waste at all, it's a must.

Palarran

07-21-2003, 10:58 AM

The question of course is what priority should be given to CA/LR. I don't think anyone would argue that the skill does absolutely nothing. Assuming that you're going to be getting AA's anyway, and not exp grinding only to get specific abilities...I'd say it's somewhere in the 100-200 area, probably closer to 200 for most people. Most of the archetype skills, a few of the class abilities, and some of the PoP abilities are more important. After those are out of the way, CA/LR are well worth the AA's.

Casai

07-21-2003, 01:04 PM

I play warrior who has those and in PoP its makes a HUGE difference. For druid I'd think it would make even bigger difference since you definatley have to somehow survive a few hits everyone now and then before mez hits or tank lost agro or kite went bad you name it and the class definatley does not have the hp to survive more then a few hits so you better bet your money on the miss.

As for parse that started this thread anyone who does mele parsing can say outright 500 hit sample proves nothing in EQ.

Tudamorf

07-25-2003, 01:17 AM

Casai says: As for parse that started this thread anyone who does mele parsing can say outright 500 hit sample proves nothing in EQ.

Well clearly, your vague speculation and use of ALL CAPS on the adjectives is far more scientifically enlightening. <img src=http://lag9.com/rolleyes.gif>

Casai

07-25-2003, 01:36 PM

thanks for sharing Tudamorf? your last post was indeed much more helpfull and on the issue then mine.

As been noted by others your disparity of "% Hits" just in the parses of LR5 should tell you that the sample needs to be much larger.

Tudamorf

07-26-2003, 12:22 AM

Let me put it a different way: I and the other empiricists on this thread have collected thousands of data points. One hundred years of statistics theory says that these numbers do have significance, even if there is a margin of error and some ultimate uncertainty.

You presented zero data points. Instead, you posted on a thread that was focused on the collection and analysis of data by spouting a conclusion that was (a) directly contrary to at least some of the data, with no explanation as to why and (b) unsupported by data of your own.

Then, you mindlessly attacked another set of data because you thought it had too few data points, even though you had no explanation for why that number was too low. No, it really doesn't say "500 hit sample proves nothing in EQ" on page one of that statistics textbook, and 500 samples can often tell you a lot about your subject. (Hint: many studies in medicine and sociology draw conclusions from even smaller samples.)

What you did was preaching, not discussing. If you disagree with a conclusion here, I'm sure every reader will welcome logical argument or added empirical data. The more we get of both, the closer we get to the right answer. But by spouting conclusions without backing them up, you move this thread a step backwards when we should be moving forwards.

As been noted by others your disparity of "% Hits" just in the parses of LR5 should tell you that the sample needs to be much larger.

Thank you, some logical argument. Yes, more data points would increase the certainty of the null hypothesis (the hypothesis that LR has no effect on hit percentage). But it will never be absolutely certain, even if you have a million data points because you can't prove the negative. How much certainty does it take to satisfy you? Or the typical gamer?

Also, the fact that the hypothesized effect is either zero or buried in the noise of the data speaks volumes about its desirability to the average druid. If you can't really notice the effect, because it's non-existent or too small to notice, should you care about it? Should you really spend a month's worth of AAs on that drop in the bucket?

AmonraSet

07-28-2003, 05:59 AM

Okay, so you want some facts to support Casai’s argument that 500 is too few data points (warning, it gets quite complex ahead).

Firstly we need to consider the nature of the testing. Each result can be either a hit or a miss, and there must be a certain probability of each attack being one or the other. I will assume that this probability remains constant throughout the duration of each test which seems to be a reasonable hypothesis (and also otherwise it becomes far beyond my skill to analyse the situation). As such the hit/miss pattern should follow a binomial distribution. A standard scientific measurement of error margins (for example the +/- percentages you will see on most polls) is that the error margin covers a range for which, if you repeated the experiment many times, you would expect 90% of the results to fall within this range.

If we assume that the hit percentage is 72%, then over 500 tests the error margin will be approximately +/- 3.3% (i.e. up to 75.3% and down to 68.7%). Given this, we can see that all of the results posted, with the exception of the first LR5 test would fall within the error margin. So it would be possible to draw the conclusion that CA/LR have no impact whatsoever.

As a rough rule of thumb, in order to be able to detect a trend in data movements above the margin for error, the expected movement in the results should really be at least double the margin for error. Therefore 500 samples would be reasonable if we were looking for a 6.6% or greater movement in the hitrate from each point of CA or LR, but I think everyone can agree that the actual change is going to be far less than this.

If 1,000 samples were taken rather than 500, the error margin drops to approximately +/- 2.6% which is rather better, but still seems insufficient to measure the expected movements.

However, given the above, the first LR5 test showing a 62.7% hit rate is quite telling. If the hit rate had remained unchanged at 72% for all of the tests, then the odds of randomly getting a result as low as 62.7% are extremely small. About 0.0005% in fact.

Therefore we can conclude that the hitrate does appear to have gone down, but cannot really draw any accurate conclusions as to how much. If we were to assume that the first result of 74.8 was at the very top of the error margin, and that the first test for LR5 was at the very bottom of the error margin then the underlying hitrate would have changed by at least 71.6% to 66.4%. However this is very much falling into the trap of trying to draw too many conclusions from insufficient data.

Finally, it may be possible to draw slightly better conclusions by taking all of the results posted and accounting for the actual hit probability varying slowly by an unknown amount as CA/LR changes. However it is far beyond my skill to do so.

Tudamorf

07-28-2003, 10:01 PM

AmonraSet says: Okay, so you want some facts to support Casai’s argument that 500 is too few data points (warning, it gets quite complex ahead)...

Well, you're assuming independent trials (i.e., that any variation outside of the control variable is purely random). This is not true with EQ, since the "random" numbers are very streaky and time-sensitive. Therefore the formulae you are using do not apply and give you bizarre results (like an occurrence being extremely unlikely outside of a change in the control variable, when in fact it is very likely). Dependent trials make the situation much stickier and more complicated to quanity, but in general, it means you need even more data to get meaningful results.

It also means that it is a lot more difficult to reject the null hypothesis, i.e., the conclusion that CA/LR has no effect whatsoever on hit percentages, because you can't tell whether the variation in the data was due to the variable you are studying, or something else that's linking the dependent trials. In other words, was that just a streak caused by a buggy RNG, or a genuine effect?

So yes, we need more samples to be more certain. And more types of data from different sources to help dilute the effect of extraneous variables. But when you propose a hypothesis (e.g., CA/LR affects hit percentages), it is your burden to prove the hypothesis, not everyone else's burden to prove the negative. So far, the evidence in the support looks pretty dubious.

Eoghan007

07-29-2003, 12:55 AM

All I can say is that with LR5 and CS3 now I can tank mobs in places like HOH, BOT and POV etc. Without them, I am at real risk of becoming a green splat on the ground.

I think the whole problem is the Random Number Generator in EQ and the streakiness of it. We all know what it can do to trade skills. Think about what its doing next time you get hit 38 out of 40 times.

Prior to LR5/CS3 the problem with my HP pool was that it would be very streaky, I could tank for a few seconds (say to draw off a mob to root it) but if I was MT it would look fine, healer would get comfortable and all of a sudden Woosh Ranger down.

With the AA skills my HP is still subject to sudden drops but overall its alot more stable making it easier to see when the heal needs to be cast.

Both skills (agility and stability) are well worth the AA spent for those who are prone to sudden agro, they may not work EVERY time but when they do they can definitly prevent you from becoming a green smear on the floor.

Tiane

07-29-2003, 03:32 AM

Yes well, a ranger has a completely different set of defensive skills and tables, so it's really not relevant for a class which operates on the priest defense tables.

Fact is anyone with 6k of buffed hp can tank a slowed mob in any tier 1-3 zone. I do it all the time and I have none of any LR or CS or any of those skills at all.

The Drunken Mage

07-29-2003, 08:50 PM

<blockquote>Quote:<hr>
I doubt I'll do it in Tier 3, but in Tier 2, probably PoV against a crawler or golem. I'm hoping to collect all the AA first, then do the test against the same crawler or golem to eliminate any variation in NPC level or stats.[/quote]

That would be a good test. I would also increase the sample size at each point (minimum of 1000).

500 is simply too small. You mention the "streaky nature of the RNG" yourself. <blockquote>Quote:<hr>The low hit percentage on my initial test was likely the result of a streaky RNG during that particular time period.[/quote]Because of that, you redid the final data set because it didn't match your expectations.

You kind of gloss over that saying that it is "within the margin of error."

The difference between your data sets (7.1 delta in miss percentage) is actually about a 23.5% difference in the results. (7.1/30.2 = .235, or if you prefer 37.3/30.2 = 1.235). If your margin of error is >20% then your test isn't proving anything.

If the sample size permits a fluctuation of 20% of the final value, then it is too small. You acknowledged that by redoing that sample. How can you throw out one sample as invalid due to sample size and the streaky nature of the RNG and yet insist that the other samples (of the exact same size) are valid?

I think the biggest problem is you got too attached to your results while defending them. In your initial post you say, "Overall, if the mist panther data is representative of a typical fight against a PoP-level NPC, then LR5 is a waste of 27 AAXP. That was great. But the moment somone started challenging you on that, and saying that your results were possibly not valid because of the very concern you listed in your initial post, you got very defensive and dismissed their concerns.

You had a decent initial test. You recognized that there were two problems. One was the assumption that the mob you were testing on was representative of PoP mobs. Someone else's testing showed that the results possibly vary by mob. Therefore, for the tests to be useful, they need to be done on a representative mob. Your next problem was the small sample size. You realized it was a problem (or you wouldn't have redone a data point) but you didn't take into account that the sample size problem casts a pretty large shadow over the entire test.

Melee folks parse DPS, damage taken, and other related items (crit rates, proc rates, etc) extensively. One thing we've learned is that when dealing with EQ's RNG is that sample sizes need to be very large to be valid.

I commend you on doing the test in the first place. And I hope that you do your second test. But I also hope that you can step back from it for a minute and take the criticisms of your methodology to heart and design your next test to be more valid.

iegil

08-02-2003, 05:05 AM

We had an interesting discussion in guild the other night about lightning reflexes and why some people get more benefit from it than others.

The cleric in my group was like I have it and I die all the time anyways. I don't understand why the other cleric lives so much.

So we started looking for the reason.

One of our warriors made the comment that LR5 is based against your "dodge" skill. Most wis casters have it capped fairly low. A druids is capped at 85 I believe.

everquest.allakhazam.com/...item=18099 (http://everquest.allakhazam.com/db/item.html?item=18099)

Is the item I have that changed my dodge above the cap. I just finished Combat Stability 3 and am really happy with the results.

Iegil

L1ndara

08-02-2003, 10:15 AM

One of our warriors made the comment that LR5 is based against your "dodge" skill. Most wis casters have it capped fairly low. A druids is capped at 85 I believe.

Is the item I have that changed my dodge above the cap. I just finished Combat Stability 3 and am really happy with the results.

Well, it's pretty clear when your dodge skill succeeds since you get the message, the bonus you get from that item will be pretty much unnoticeable. Also I don't think your dodge skill matters at all for LR.

Palarran

08-03-2003, 01:25 AM

However, having high agility _may_ help. If agility is the only non-AA way to improve your avoidance AC, and LR adds a percentage to that avoidance AC, it would be more noticeable the higher your agi is.

Gwynet Woodsister

08-03-2003, 07:37 AM

I have yet to see any testing or figure proving that agility has anything to do with avoidance.

iegil

08-04-2003, 02:20 AM

Dodge is the only difference when the guild buffs out at 305 stats...

Sylphan2

08-06-2003, 04:58 AM

"I have yet to see any testing or figure proving that agility has anything to do with avoidance."

Here's the full formula for AC (not counting AA adjustments):

Avoidance = (T from table) + trunc(Defense * 16/9), but never less than 0
Caster Mitigation = trunc(Buffs/3) + trunc(Defense/2) + (Equipment + 1)
Anyone else's Mitigation = trunc(Buffs/4) + trunc(Defense/3) + trunc(Equipment * 4/3)
Displayed AC = trunc( (Avoidance + Mitigation) * 1000 / 847 )

The table being referred to cross-references your level and agility. The table values range from -24 for 1 agi to 65 for a high level character with 240-255 agi. I never tested with over 255 agi after PoP came out. Other threads and other boards are more appropriate for going into great detail... but I bring this up to point at the two portions. The two factors listed under avoidance are grouped together because of the 'never less than 0' portion. By extensive testing with very low agility numbers (for instance by carrying large quanities of copper pieces), you can demonstrate that the number is rounded up to zero after adding those two factors together, and before adding in any other factors. Also notice that those two factors are the same for all classes, while the other three factors are different between casters and warriors.

It's still possible that they round those two factors up to zero, and THEN add in the amount from buffs and call that avoidance. There's no proof that it doesn't happen. Then mitigation would not be affected by buffs at all, only avoidance. But it doesn't seem plausible that they would round up to zero before adding in all factors.

It's also possible that some portion of the equipment modifier goes to avoidance. Perhaps they round up to zero and THEN add (equipment/3) to avoidance. And add only (equipment) to mitigation. That's possible, but again we don't think it's likely that they would round up to zero before adding in all factors.

Next, it's possible that we just have mitigation and avoidance backwards. Maybe mitigation is affected by agility, and avoidance is affected by gear and buffs. But that just seems silly from a common-sense perspective.

Finally, it's possible that this is all a huge ruse. This entire extensive calculation could be used just to create the AC number that's displayed to us on our screen, and the actual internal avoidance and mitigation calculations are totally different. But that's just paranoid.

It's not actual proof, but that's why we think agility affects avoidance.