Serancha
04-14-2014, 11:35 AM
In light of the new damage infomation brilliantly brought out by Kalizza, it is likely that a lot of people are going to be wanting to test things. There are so many fascinating features in Arcane Legends, but many of them can change your damage without you realizing, making test results inaccurate. I have been testing with my game-partner since season 2, and we have a pretty solid method for achieving good and accurate data.
So, here's the basics for anyone who wants to join in on testing these game-changing revelations.
Preparing to test
• The key is to use pets /weapons / skills without armor reduction, dot or bleed.
Pets
• If you do not need a pet for your test, do not use one at all.
• If you need to use a pet, use the lowest level one you have that will do the job, and make sure to check that their passive attack will not cause any of the status effects above.
• Try and find pets to use that give ONLY the thing you are trying to test, and not additional stat changes as well, since these would alter your damage output. (The beastmaster pets are usually best for testing, as mythic / arcane have too much other junk on them for accuracy)
• Stat changing pets (str, dex, int) can have an effect on your damage, so are unadvisable in this situation, unless you have a second pet with the exact same stat changes, along with the different variable.
Weapons
• Check the proc of the weapon before you test with it. Many will proc additional damage or some other thing that can skew results without you noticing. Either use a weapon with a very obvious proc, or one that doesn't change your damage / reduce enemy armor when it goes off.
• As a rogue I prefer daggers for doing this kind of thing
Skills
• If you are testing an upgrade on a skill, make sure to take off ALL other upgrades. Perform your control with only the skill activated, and no upgrades at all, then test with only the upgrade you are looking for data on. Other upgrades do different things, and change the way the skill works. We also don't know if combinations of upgrades affect things, so best to keep it simple.
• Unless you are testing a passive, it is best to take off your % based passives like damage and armor - leaving off crit will help your test go faster, since crits are not accurate data. (way too wide a range)
Following the steps above, you will be testing a single variable and not having other factors influencing your dataset.
Counting Damage:
• Unless testing a specific skill, use only uncharged normal attack.
• Do not count crits, only normal attacks,
• Get as large a sample as possible, recording which specific mob takes how much damage each hit. (ie, the first mob in elite Brackenridge Forest / Rooks Nest / Skull Cove / Oltgar etc)
• You will likely have to remap (use partyinv 0 to enter, leave and /partyleave, then /partyinv 0 again) in order to get more samples on the appropriate mob.
• Once you have a decent control sample size, change your variable, and record the same number of hits on the identical mobs.
• For sample size, it is really the more the better. I go for a minimum 25 (usually 50) recordable hits per mob for the control and then again for the variable. Minimum 20 mobs for pve to remove specific oddities. Spreading out over several maps is a good idea too.
• When done, count up all your control hits for one mob, and divide by the total number of hits. This gives your average. Do the same for the rest of your controls
• Count up the second tests (the ones with your variable changed) and find the averages for those.
• Compare the two data sets, and there is your result.
I know others will have different methods, which are welcome to be added (please keep to the accurate ones, since this will be tagged for the guide section.)
I may have missed a couple of points, as I am writing this on the fly, so if anyone sees a gap in the process, please chime in.
Thanks!
So, here's the basics for anyone who wants to join in on testing these game-changing revelations.
Preparing to test
• The key is to use pets /weapons / skills without armor reduction, dot or bleed.
Pets
• If you do not need a pet for your test, do not use one at all.
• If you need to use a pet, use the lowest level one you have that will do the job, and make sure to check that their passive attack will not cause any of the status effects above.
• Try and find pets to use that give ONLY the thing you are trying to test, and not additional stat changes as well, since these would alter your damage output. (The beastmaster pets are usually best for testing, as mythic / arcane have too much other junk on them for accuracy)
• Stat changing pets (str, dex, int) can have an effect on your damage, so are unadvisable in this situation, unless you have a second pet with the exact same stat changes, along with the different variable.
Weapons
• Check the proc of the weapon before you test with it. Many will proc additional damage or some other thing that can skew results without you noticing. Either use a weapon with a very obvious proc, or one that doesn't change your damage / reduce enemy armor when it goes off.
• As a rogue I prefer daggers for doing this kind of thing
Skills
• If you are testing an upgrade on a skill, make sure to take off ALL other upgrades. Perform your control with only the skill activated, and no upgrades at all, then test with only the upgrade you are looking for data on. Other upgrades do different things, and change the way the skill works. We also don't know if combinations of upgrades affect things, so best to keep it simple.
• Unless you are testing a passive, it is best to take off your % based passives like damage and armor - leaving off crit will help your test go faster, since crits are not accurate data. (way too wide a range)
Following the steps above, you will be testing a single variable and not having other factors influencing your dataset.
Counting Damage:
• Unless testing a specific skill, use only uncharged normal attack.
• Do not count crits, only normal attacks,
• Get as large a sample as possible, recording which specific mob takes how much damage each hit. (ie, the first mob in elite Brackenridge Forest / Rooks Nest / Skull Cove / Oltgar etc)
• You will likely have to remap (use partyinv 0 to enter, leave and /partyleave, then /partyinv 0 again) in order to get more samples on the appropriate mob.
• Once you have a decent control sample size, change your variable, and record the same number of hits on the identical mobs.
• For sample size, it is really the more the better. I go for a minimum 25 (usually 50) recordable hits per mob for the control and then again for the variable. Minimum 20 mobs for pve to remove specific oddities. Spreading out over several maps is a good idea too.
• When done, count up all your control hits for one mob, and divide by the total number of hits. This gives your average. Do the same for the rest of your controls
• Count up the second tests (the ones with your variable changed) and find the averages for those.
• Compare the two data sets, and there is your result.
I know others will have different methods, which are welcome to be added (please keep to the accurate ones, since this will be tagged for the guide section.)
I may have missed a couple of points, as I am writing this on the fly, so if anyone sees a gap in the process, please chime in.
Thanks!