Proc rate on Purify Spell

Adroit - Lost City

So I've been hearing a lot of conflicting estimates for purify spell, and figured I'd do a little of my own testing. I haven't finished yet (relatively small sample size) but I figured I'd post what I have so far in case anybody else feels like helping me with testing. The way I tested was I cleared my chat window, added damage to trade chat (and switched to that) and began punching my purify-wielding character listening for the proc. Every time I got a proc I recorded the number of hits that it took to get the proc, and re-cleared my chat window. Any time I lost track of the number of hits I was at, I just counted the number of lines in the chat box. Pretty basic.

So the data I have so far:

Against my wizard (calc in signature) I had an alt 93 bm with almost no gear (no weapon) hitting me-
Number of hits: 519
Number of procs: 61

Estimate for proc rate: ~11.75%
95% confidence interval: 8.98% - 14.52%

For fun I did this same test with my veno that has r8r wep with purify because many people have claimed that the proc rate is different for r8r and r9rr.
Number of hits: 563
Number of procs: 24

Estimate for proc rate: ~4.26%
95% confidence interval: 2.59% - 5.93%

I was really surprised by this, so went back on my wizard and took off all my gear (other than wep/tome) wondering if maybe soulforce or something in my gear had some effect on proc rate.
Number of hits: 500
Number of procs: 69

Estimate for proc rate: ~13.8%
95% confidence interval: 10.77% - 16.82%


I kinda hate to draw any conclusions from such limited data, but it appears as though the proc rate on purify is different on r8r and r9rr weapons.. if anybody cares to help me out with some more data I can narrow this down a bit more and if you do please include any relevant information (maybe calcs + a quick descrip how you did it) in case the proc rate is not static. And for anyone that is curious, the confidence interval was calculated by taking the probability estimate and adding/subtracting the error.. which is Z value (standard score) for 95% (1.96) and multiplying it by the square root of ((probability estimate * (1 - probability estimate)) / number of hits).

Fissile - Archosaur

f:ponder How on earth are your results so much different from mine? I performed a similar test some time ago (right after I got the weapon, so like a week after the expansion) hitting myself with my 101 seeker and it worked out to between 5 and 6%.

Were you in a duel, or PK? f:confused

Adroit - Lost City

Fissile - Archosaur wrote: »

f:ponder How on earth are your results so much different from mine? I performed a similar test some time ago (right after I got the weapon, so like a week after the expansion) hitting myself with my 101 seeker and it worked out to between 5 and 6%.

Were you in a duel, or PK? f:confused

It wasn't a duel.. there are a few things that don't work the same in duels as everywhere else (immune being what I'm thinking of). I just went out of sz in the middle of nowhere and started testing.. I wonder if the rate isn't static and something other than gear/soulforce affects it? This is kinda why I wanted help getting data, I'll probably do some more of my own testing tomorrow.. but seems odd that none of the estimates I've heard seem to match up.

Slivaf - Dreamweaver

Fissile - Archosaur wrote: »

f:ponder How on earth are your results so much different from mine? I performed a similar test some time ago (right after I got the weapon, so like a week after the expansion) hitting myself with my 101 seeker and it worked out to between 5 and 6%.

Were you in a duel, or PK? f:confused

This is just a theory, but the base 'aps' of a bm even without a weapon is 1.25, and seekers have 1.11 at best, albeit it isn't a huge difference, perhaps it is enough that the numbers differ as much as yours do compared to adroit's.

EDIT: My guess as to why the numbers differ... is due to what I just said up there in reply to fissile, there are definitely a lot of 'variables' that need to be 'plugged-in/stated' for data to coincide with what others have, which is why I believe adroit asked for more 'results.' (Attack rate is definitely something that I feel that should be added into each equation when trying to figure out how much it procs.)

Edit 2: I stand corrected after equipping my seekers sword it too has a base attack rate of 1.25 (with SINGLE sword/blade on, 1.00 with daul sword/blade) still, just make sure the attack rates match up with each others data, to get better results/ones that are much more likely to be correct in the end. After thinking about this... I dont stand corrected... to an extent... the g16 wrists is what allows my seekers attack rate to match the blademasters. The - int gears and the attack rate of the attacker i imagine play a huge role in how often this proc happens.

Desdi - Sanctuary

Interesting findings.
If you do more tests in the future I hope you can do more R8r and R9rr comparisons. I did hear a lot of rumors about R8r Purify proc activating fewer times and it seems like it is indeed true, at least from the few things you found.

Joshcja - Sanctuary

Adroit - Lost City wrote: »

So I've been hearing a lot of conflicting estimates for purify spell, and figured I'd do a little of my own testing. I haven't finished yet (relatively small sample size) but I figured I'd post what I have so far in case anybody else feels like helping me with testing. The way I tested was I cleared my chat window, added damage to trade chat (and switched to that) and began punching my purify-wielding character listening for the proc. Every time I got a proc I recorded the number of hits that it took to get the proc, and re-cleared my chat window. Any time I lost track of the number of hits I was at, I just counted the number of lines in the chat box. Pretty basic.

So the data I have so far:

Against my wizard (calc in signature) I had an alt 93 bm with almost no gear (no weapon) hitting me-
Number of hits: 519
Number of procs: 61

Estimate for proc rate: ~11.75%
95% confidence interval: 8.98% - 14.52%

For fun I did this same test with my veno that has r8r wep with purify because many people have claimed that the proc rate is different for r8r and r9rr.
Number of hits: 563
Number of procs: 24

Estimate for proc rate: ~4.26%
95% confidence interval: 2.59% - 5.93%

I was really surprised by this, so went back on my wizard and took off all my gear (other than wep/tome) wondering if maybe soulforce or something in my gear had some effect on proc rate.
Number of hits: 500
Number of procs: 69

Estimate for proc rate: ~13.8%
95% confidence interval: 10.77% - 16.82%


I kinda hate to draw any conclusions from such limited data, but it appears as though the proc rate on purify is different on r8r and r9rr weapons.. if anybody cares to help me out with some more data I can narrow this down a bit more and if you do please include any relevant information (maybe calcs + a quick descrip how you did it) in case the proc rate is not static. And for anyone that is curious, the confidence interval was calculated by taking the probability estimate and adding/subtracting the error.. which is Z value (standard score) for 95% (1.96) and multiplying it by the square root of ((probability estimate * (1 - probability estimate)) / number of hits).

results on mage seem consistant using both of your test sessions we get 12.7%

I doubt purify differs by class however r8r vs r999 procs or a soulforce based comparison may exist

Way to test if its souforce based: take off all your mages gear and shoot it with a level 1 x times. put on gear shoot it x times

skaitavia

Joshcja - Sanctuary wrote: »

results on mage seem consistant using both of your test sessions we get 12.7%

I doubt purify differs by class however r8r vs r999 procs or a soulforce based comparison may exist

Way to test if its souforce based: take off all your mages gear and shoot it with a level 1 x times. put on gear shoot it x times

That would require several test runs to provide a sure average proc rate.

I think once 5k+ hits have been recorded per use case can we decide on a sure proc rate of the spell (on both R8r and R9r2). Other than that this is useful information!

PotatoHeadQR - Dreamweaver

SweetieBot please award 20 points to Adroit - Lost City for calculating confidence intervals. Can we get a calculation for 99% too? (that's the next step right? or is it 98%? or 99.8%? i have a faint memory of an eight xD)

Joshcja - Sanctuary wrote: »

results on mage seem consistant using both of your test sessions we get 12.7%

I doubt purify differs by class however r8r vs r999 procs or a soulforce based comparison may exist

Way to test if its souforce based: take off all your mages gear and shoot it with a level 1 x times. put on gear shoot it x times

on the other hand, it might be based on soulforce difference so it will be nice to do tests with a 100+ char too. but I doubt that, sounds more probable that it's class based

SweetieBot - Lothranis

PotatoHeadQR - Dreamweaver wrote: »

SweetieBot please award 20 points to Adroit - Lost City for calculating confidence intervals. Can we get a calculation for 99% too? (that's the next step right? or is it 98%? or 99.8%? i have a faint memory of an eight xD)

You are limited to awarding only 10 points per request.
PotatoHeadQR - Dreamweaver awards 10 points to Adroit - Lost City!
Adroit - Lost City is now in 11th place for May 2013 with 10 points (18 points overall).
PotatoHeadQR - Dreamweaver can still award another 20 points today.

Check this thread for the current high scores and to learn how to award points to others.

HrunsPanda - Archosaur

SweetieBot please award 10 points to Adroit - Lost City for trying to find facts instead of discussing BS in that other horrible monstrocity of a purify thread

(has nothing to do with your 20 points not working. Im just happy to read posts like this)

SweetieBot - Lothranis

HrunsPanda - Archosaur wrote: »

SweetieBot please award 10 points to Adroit - Lost City for trying to find facts instead of discussing BS in that other horrible monstrocity of a purify thread

HrunsPanda - Archosaur awards 10 points to Adroit - Lost City!
Adroit - Lost City is now in 3rd place for May 2013 with 20 points (28 points overall).
HrunsPanda - Archosaur can still award another 20 points today.

Check this thread for the current high scores and to learn how to award points to others.

Asterelle - Sanctuary_1381265973

A year or two ago I did a test with barehand punching an R8r cleric in a duel and recorded 8 procs in 500 hits. Perhaps low damage attacks have a lower chance though.

HrunsPanda - Archosaur

Asterelle - Sanctuary wrote: »

A year or two ago I did a test with barehand punching an R8r cleric in a duel and recorded 8 procs in 500 hits. Perhaps low damage attacks have a lower chance though.

If theyd make it only 20k hits (before pvp reduction and armor) or better cause the proc, the problem that would solve all the NW problems and not affect r9.3 vs r9.3 balance

Of course, noone will think that is a good idea, because the whole idea of pay to win is that you can trash defenseless noobs 20v1

Flamespirit - Dreamweaver

Assuming that a new player is going to just cash shop their toon it would make sense that r93 would proc higher than r8r considering the vast difference in amount of US dollars.

The idea that weak attacks wouldn't proc it is bad because it helps against aps toons, which are in large part a bunch of really fast weak attacks.

Zanryu - Dreamweaver

Flamespirit - Dreamweaver wrote: »

Assuming that a new player is going to just cash shop their toon it would make sense that r93 would proc higher than r8r considering the vast difference in amount of US dollars.

It's the same proc. Perhaps varying proc rates on different classes would make sense, but not varying proc rates between weapons on the same class.

Azura - Lost City

Your results for your wizard are close to what it says on the database.

On the database it says the chance for prok in r9-3 is 14.29%

HrunsPanda - Archosaur

Azura - Lost City wrote: »

Your results for your wizard are close to what it says on the database.

On the database it says the chance for prok in r9-3 is 14.29%

b:laugh

Zanryu - Dreamweaver

Azura - Lost City wrote: »

Your results for your wizard are close to what it says on the database

Yeah, the weapon also only has a 14.29% chance to decrease channeling by a certain %, oh, and each set of attack levels only have a 14.29% chance of occurring. Yes indeed, that's right folks. When you get blasted by a Wizard's R9r3 weapon there's only a small chance of 40+ attack levels backing it up.


I think the insinuation that you lack reading comprehension was right on the money.

Zsw - Dreamweaver

There are some evidence that purify spell for r8r has a lesser proc rate than r9rr. The game has two "Purify Spell" and the r8r weapon seem to use a different purify spell than r9rr. Although it is uncertain if this is the correct reason, or the only reason.

Now i don't know if this is meant to be public information. Mods please remove it if not o.o.

Adroit - Lost City

I don't have any more data yet, but just did a quick calculation to see how much data I'd need to narrow the width of the confidence interval to .01 (e.g. 95% confidence to be 11.5%-12.5% for example).. and assuming a proc rate of 12%, it'd take over 16k hits to do this. I'm not intending to spend that much time testing this on my own, but I might try a few more smaller tests to see if anything I can think of affects the proc rate (or if it is just static). If I can't find anything that affects the proc rate, it may be possible for a few people to do like 1k trials on their own and combine the data into one big lump that would give a fairly accurate estimation of proc rate. Anyway, back to studying for finals :P

Mystic-Night - Heavens Tear

I already told you my raw data with r9s3. I don't have any data with an r8r weap but I can tell you that on my veno (r8r weap) I notice that purify procs significantly less than I notice it proc with r9s3. It's just a visual thing for me with the r8r, but yea, I notice a difference and it's pretty big. I wouldn't be surprised if the proc rate was 5% in comparison to my data that r9s3 is 10%.