(Wasn't sure whether to put this thread here of Sorc forum ... since I guess many people will be interested in this I put it here, so more people can see it)
Okay so yesterday I finished my first 3 piece epic Abishai set on my Air Savant and I was thinking "OMG wow finally ... now my chain lightning will rock even more ... AWESOME!!!!!" ... So went to DD (hard) to do some Eardweller farming ... after a few casts I noticed the numbers seemed a bit off ... switched the set on and off several times and there seemed to be no difference, but well we all know that "just a feeling that something is way off" is not enough. Thus I did some serious testing:
Test conditions:
.) Each specific combination was tested exactly 100 times (when I got more than 100 I ignored the numbers past 100)
.) I did the testing in normal Irestone Inlet, simply because there are plenty of mobs that are not resistant to elec or cold and the mobs are so easy (and there are enough "safe spots"), so it was no problem to stop after each ~20 kills to write down all numbers from the combat log (the combat log has a very small size ... if you kill too many mobs at once numbers will get lost). Additionally since every mob in there dies from one cast, this gave me the nice feature that I could control if I oversaw any numbers by comparing the amount of damage numbers I copied out and the number of kills.
.) The test was done on a level 20 first life sorc (air savant)
.) Since Maximize and Empower blow the numbers up even more - creating more variance - I turned both off for all the tests => only active metamagic was Heighten
.) My sorc has fully maxed out elec and cold lines
.) I always used superior VIII clickies (ToD belts)
Tested spells:
1.) Chainlightning: No ToD set, no Abishai set, Major Lore, Superior Spark VIII
2.) Chainlightning: Air ToD set, no Abishai set, Major Lore, Superior Spark VIII
3.) Chainlightning: Air ToD set, 3 piece Abishai set, Major Lore, Superior Spark VIII
4.) Polar Ray: Air ToD set, 3 piece Abishai set, Major Lore, Superior Freeze VIII
5.) Eladars: Air Tod set, 3 piece Abishai set, Major Lore, Superior Spark VIII
For the calculation of the expected damage values I used the following (widely as correct accepted) formular:
Min damage = (MIN_OF(CL, Max_CL) * MIN_DAMAGE_PER_CL + LOWEST_POSSIBLE_FIXED_DAMAGE) * 2,25 * 1,2
Max damage = (MIN_OF(CL, Max_CL) * MAX_DAMAGE_PER_CL + HIGHEST_POSSIBLE_FIXED_DAMAGE) * 2,25 * 1,2
Average damage = (Min_damage + Max_damage) / 2
In the following calculation you will notice that the actual min/max values will be far higher/lower than the calculated ones. This is very simply due to the reason that the more dice are rolled, the more average the result will be. The exception to this is Eladars, since it only uses one dice.
Also notice I rounded all calculated results to one decimal. DPS in the following calc means average damage per hit target including target mobs saving 5% of the time for halve damage and including crits.
1.) Chainlightning: No ToD set, no Abishai set, Major Lore, Superior Spark VIII
Expected stats: CL 26, max CL: infinity
Expected damage: 280,8 to 421,2, average: 351
Expected crit damage: 772,2 to 1158,3, average: 965,3
Expected dps: 452,8
Actual damage: 283 to 329, average: 310,7 => ~ -11,5% off
Actual crit damage: 801 to 905, average: 854,2 => ~ -11,5% off
Actual dps: 373,5 => ~ -17,5% off
While I know a sample size of 100 is far from perfect, it still holds a lot of value and errors ranging from 11,5% to 17,5% simply sow that there is something wrong. After a lot of trying around and diverse theories I don't want to delve into here, I came to the conclusion that most probably chainlightning is capped at CL 20. As an Air Savant I have a default CL 26 and +3 max CL on all elec spells. Thus I redid the expected calcs for CL 23:
Most probable stats: CL 26, max CL: 23 (20 + 3 Savant)
Most probable damage: 248,4 to 372,6, average: 310,5 => ~0,06% off
Most probable crit damage: 683,1 to 1024,7, average: 853,9 => ~0,03% off
Most probable dps: 440,5 => ~ -6,75% off
As we can see now: The assumption of chainlightning having CL 20 and Savant Pre/Sorc capstone/... working correctly seems to be right (0,03% to 0,06% error is actually incredibly good for a 100 sample size). Some might now wonder why the dps still shows an 6,75% error ... this is simply due to having few crits during those 100 casts ... 100 hit opponents with max elec lines plus major lore should produce ~18 crits ... in this run I only critted 13 times => thus the dps is off ... I don't want to delve too deeply into statistics here ... just let me assure you that a sample size of 100 is okay to check damage and crit values, but its simply not enough to test crit-proc rates. Also during the following tests I then got slightly more than those 18 crits ... it simply evens out over a larger sample size.
Since we can now assume chainlightning having a max CL of 20, I used this for the new expected damage/crits calculations.
2.) Chainlightning: Air ToD set, no Abishai set, Major Lore, Superior Spark VIII
"Newly" expected stats: CL 26, max CL 25
"Newly" expected damage: 270 to 405, average: 337,5
"Newly" expected crit damage: 742,5 to 1113,8, average: 928,1
"Newly" expected dps: 435,4
Actual damage: 302 to 359, average: 335,7 => ~ -0,54% off
Actual crit damage: 898 to 1002, average: 943,8 => ~1,69% off
Actual dps: 440,2 => ~1,1% off
As we can see, the errors aren't as small as in 1.) but still low enough to suggest our assumption was correct.
3.) Chainlightning: Air ToD set, 3 piece Abishai set, Major Lore, Superior Spark VIII
"Newly" expected stats: CL29, max CL 25 => same expected damage/crit damage as in 2.)
Actual damage: 313 to 359, average: 338,1 => 0,19% off
Actual crit damage: 883 to 1002, average: 937,5 => 1,01% off
Actual dps: 432,3 => ~ -0,7% off
Seems our assumption still stands => we can combine test results of 2.) and 3.) to form a larger sample size of 200 => errors decrease to -0,35%/-0,68%/0,4% which is pretty good
4.) Polar Ray: Air ToD set, 3 piece Abishai set, Major Lore, Superior Freeze VIII
Expected stats: CL 23, max CL 25
Expected damage: 248,4 to 372,6, average: 310,5
Expected crit damage: 683,1 to 1024,7, average: 853,9
Expected dps: 408,3
Actual damage: 288 to 326, average: 311,1 => ~0,2% off
Actual crit damage: 816 to 876, average: 845,1 => ~ -1,02% off
Actual dps: 424,4 => ~3,94% off
While a non-perfect match, those numbers still suggest that everything is working as expected for polar ray .. dps a bit off since i critted more than averagely
5.) Eladars: Air Tod set, 3 piece Abishai set, Major Lore, Superior Spark VIII
This one works a bit differently. Since there is only one d6 dice rolled, there should also be only 6 possible different damage numbers popping up. I must admit here that I used a smaller sample size for this spell ... but i got all 6 different values: 56, 59, 62, 64, 67, 70 ... just as expected ... though those 6 numbers were smaller than expected:
Expected stats: CL 29, max CL: infinity
Expected damage: 81 to 94,5, average: 87,8
Expected crit damage: 222,8 to 259,9, average: 241,3
Expected dps: 120
Actual damage: 56 to 70, average: 62,9 => -28,33% off
Actual crit damage: 155 to 185, average: 172,7 => -28,45% off
Actual dps: 78,6 => ~ -34,52% off
These numbers (even for a smaller sample size) strongly show that there is something waayyyyyy off. After some trial and error, I assumed max CL 15 for Eladars:
"Newly" expected damage: 56,7 to 70,2, average: 63,5 => ~ -0,88% off
"Newly" expected crit damage: 155,9 to 193, average: 174,5 => ~ -1,04% off
"Newly" expected dps: 83,4 => ~ -5,83% off
This very closely matches the numbers from the test. Only dps seems a bit off, but once again i critted less than average, and due to smaller sample size this part of the test is especially prone to inaccuratenesses.
----------------------------------------------------------------------------------------------------------
Summary:
Some new CL caps have been tested and confirmed:
Chainlightning: Max CL 20
Eladars Electric Surge: Max CL 15
Since Niacs Biting Cold is basically the very same as Eladars I guess it has the same level cap of 15, although i havent tested it yet.
Also Polar rays max CL being at least 23 has been confirmed (Description says 25)
---------------------------------------------------------------------------------------------------------
Conclusion:
With these new test results (assuming that other spell descriptions are correct) there are only two damage dealing arcane spells that have a maximum CL higher than 20:
Black Dragon Bolt [Conjuration]: 25
Polar Ray [Evocation]: 26
Only one of them is an evocation and thus affected by the 3 piece Abishai set (and from what I read on the forums the 5 piece set doesn't grant a +3 CL bonus to conjuration and instead gives a +3 bonus to DC, can't test that myself since I don't have the 5 piece set): Polar Ray ... an ice spell.
So lets look at the usefulness of 3 piece epic Abishai set for savants:
Air Savant: Raises Polar Ray dps (max+emp+capstone+superior lore+superior clickie) from ~831 to ~955
Earth Savant: same
Fire Savant: Raises Polar Ray dps from ~582 to ~706 ... although how many fire savants pick up polar ray?^^
Water Savant: Raises Polar Ray dps (same as the others + ToD set) from ~1080 to ~1205
From what I read so far the 3 piece epic set should stack with arcane augmentation IX (epic Staff of Arcane Power/T3 Alchemical) ... in that case it would have the following effects:
Air/Earth Savant: Raises Polar Ray dps from ~914 to ~1038
Fire Savant: Raises Polar Ray dps from ~665 to ~789
Water Savant: Raises Polar Ray dps from ~1163 to ~1246
I guess in the end it all comes down to how much you use Polar Ray. I for one use it quite rarely, since it is single target spell and for boss fights DOTs + SLAs are more sp efficient.
Well thanks for the read and I hope I could help some other people with their gear layout planning with this... all in all the set still got its uses, especially since part of it are the helmet and cloak (+7 Cha, +7 Con, +5 Resists) and all of them have sweet slots ... but I m not sure anymore if I would have put that much effort in it, if I had known that the DOTs and chainlightning have low max CLs when their description would not indicate anything like that =/