Barspell Test |
||
|
Barspell Test
Fenrir.Motenten said: » Assuming 6% proc rate with 4/5 and 10% proc rate with 5/5. I understand your break downs of the: #1: 1%/2%/3%/5% vs #2: 2%/4%/6%/10% but do we (anyone) have proof that (#2) is the case? or were you saying the data from these two tests alone confirmed it? The data did not confirm it, no, which is why I noted those proc rates as assumptions. That's just what the results would be if the assumed proc rates were valid (and a very far cry from 0.8% chance in 6 spell casts, even with half the proc rate, that Vaness mentioned).
I explained in my earlier post that they'd need about 3000 test samples to really narrow down the proc rate value. As it is, it's somewhere between 8% and 15%, and 10% seems the most likely. Fenrir.Motenten said: » Vaness said: Maybe my comon sense is ***. 5/5 +2 10% procc rate on a mob that may cast 5 to 6 max heavy spell, that aprox give you a 0.8% chance that set procc in those 6 spell. Incorrect. Chance of at least one set proc with 4/5 AF3 vs 6 spellcasts: 31% (maybe one proc every 3 fights) Chance of at least one set proc with 5/5 AF3 vs 6 spellcasts: 47% (maybe one proc every other fight) Assuming 6% proc rate with 4/5 and 10% proc rate with 5/5. You are saying that there is 30% chance of a procc in 6 spell casted when on 100 spell it's 10%. We do not have the same maths I suppose? For me 10% means that aproximatively every 10 spells it will procc once. If there is 6 spell casted, that pretty much makes it a 0.8% chance. Unless you plan holding a mob forever, you will always get more out using clerics then af3 +2.Unless you know mobs that cast 100 spells in 1 fight regulary? you got the 10% part right, but how does the chance of a proc go DOWN when you add multiple chances?
[+]
Ramuh.Lorzy said: » you got the 10% part right, but how does the chance of a proc go DOWN when you add multiple chances? 10/100 aprox = 0.6/6 That means you pretty much have 0 chances of procc if you are fighting a monster that does not spam magic. EDIT: Like if you kill dragua in 3 min. Maybe he will get time to cast 2 maybe 3 stonaja. In that situation you would get more out with clerics then doing the 5/5 +2 Hope I explained this in a understandable way Cerberus.Vaness said: » Ramuh.Lorzy said: » you got the 10% part right, but how does the chance of a proc go DOWN when you add multiple chances? 10/100 aprox = 0.6/6 That means you pretty much have 0 chances of procc if you are fighting a monster that does not spam magic. your probability of having at least one proc is 1 - the chances of having absolutely no procs. so 1 - (0.9)^6 = ~47% as motenten said.
[+]
also, he advocated af3+2 legs only when cleric's do absolutely nothing for you, i.e. floored or capped resist rates both with and without cleric's.
Ramuh.Lorzy said: » Cerberus.Vaness said: » Ramuh.Lorzy said: » you got the 10% part right, but how does the chance of a proc go DOWN when you add multiple chances? 10/100 aprox = 0.6/6 That means you pretty much have 0 chances of procc if you are fighting a monster that does not spam magic. your probability of having at least one proc is 1 - the chances of having absolutely no procs. so 1 - (0.9)^6 = ~47% as motenten said. It's just a speculation of how many times it would procc in 6 spells. Also I don't see any situation where cleric's pants does nothing for you.Unless that perticular person have 100+ resist gears with 2x carol and barspell.But that is like, never D: you go back and forth between what it is and what it isn't.
yes, your expected value for 6 spells is 0.6. but that does not mean you have a near 0 chance to proc on a mob. basically if you're looking at a mob that casts 6 spells, you expect 0.6 procs. but if that's confusing for you, let's double it. on two mobs, 6 spells each, you expect 1.2 procs. so you expect a proc approximately every other mob.
[+]
Since it seems you're having trouble understanding how statistics works, I'll try to step you through the math. But first, some basic corrections:
Quote: You are saying that there is 30% chance of a procc in 6 spell casted when on 100 spell it's 10%. No. 30% (31% actually) applies for the 4/5 build, which I assume has a 6% proc rate, not 10% (that's the 5/5 build). Quote: For me 10% means that aproximatively every 10 spells it will procc once. As a rough understanding, that's fine. However you need to know how the actual math works before you start trying to draw conclusions from that. Quote: If there is 6 spell casted, that pretty much makes it a 0.8% chance. No. I'm not sure how you ended up with 0.8%, but it's nowhere near the correct methodology. Quote: Unless you plan holding a mob forever, you will always get more out using clerics then af3 +2.Unless you know mobs that cast 100 spells in 1 fight regulary? No. Statistics are about long term averages. It's possible a single mob can cast a single spell and you'll get a proc activation. It's also possible that a single mob can cast a hundred spells and you won't get -any- proc activations. Neither of those specific cases matter. The only thing you need to care about is what the average reduction is over long periods of time. Quote: I technically don't change it, 10/100 aprox = 0.6/6 That means you pretty much have 0 chances of procc if you are fighting a monster that does not spam magic. That's very wrong. It sounds like the "haste takes time to proc" misapprehension that was so popular (and ridiculed) several years back. It demonstrates that you're fundamentally mistaking how things work. 10% proc rate means that, yes, on average 10 out of 100 spells will get the proc effect. And yes, out of every 6 spells, 0.6 of them will proc. 0.6 is NOT equal to 0, nor does it mean that you must have 10 spells cast in order for one to proc. 0.6 is not a 'real' number; it's a representation of the fraction of events that could have occurred when your sample size is infinite (as it would need to be if you wanted a 'true' answer). Suppose you fight a mob that casts one spell during the fight. The odds of a proc is 10%, and the odds of no proc is 90%. Taken in simple terms (simple to make it easier to understand; this is not mathematically correct), if you fight 10 of those mobs and each of them casts one spell, that means that one of them will get a proc and the other 9 won't. Now, suppose that proc happened in the third fight. If you look only at those first three fights you would expect that, given 3 spells cast and a 10% proc rate, you'd have an average of 0.3 procs. Yet you got one proc. That's because the data is not the same thing as the expectation. The expectation is an abstract value; the data is an actual count. Probability means that *you don't know* when the proc is going to happen. Therefore you have to work out the probabilities for a given set of conditions. Now, to the math: In order to figure out the odds of at least one occurance of an event happening, it's easiest to work from the probability of it *not* happening. So, with a 10% proc rate, that means that 90% of the time it won't proc. The probability of two events occurring is the product of the probability of each event. The probability of getting a proc twice in two spells is 10% * 10% = 1%. The probability of *not* getting a proc in two spells is 90% * 90% = 81%. You also have the probability of getting a proc on either the first spell but not the second (9%) or the second spell but not the first (9%). Total probability: 81% + 9% + 9% + 1% = 100%. The value I gave was the probability of getting *at least* one proc out of a set number of spells. Since I don't need to know the probability of all possible combinations of procs (ie: proc on first and third spell, but not the others; proc on 5th spell only; etc), I can look at the probability of *just* the chance that no procs occurred whatsoever. That probability, for six spells, is 90% * 90% * 90% * 90% * 90% * 90% (ie: spell did not proc on any of the six spells) = 53.1%. Given that I know that 53% of the time there were no procs at all, I can also conclude that the other 47% of the time I got at least one proc. This does not say that there will or must be one proc made in each fight, but rather that, if you were to fight, say, 1000 mobs, you'd get at least one set proc in 469 of the fights, and you wouldn't get any procs in the other 531 fights. Of course probability also says that you most likely won't get *exactly* that number of fights with and without procs, but that's a separate (more advanced) matter. Hopefully this will help you understand how the math works.
[+]
Quote: haste takes time to proc
[+]
Ramuh.Austar said: » Quote: haste takes time to proc iQQ also
guys why does it matter even if our set bonus procs the DDs will just blame it on shadow ring proc anyways 
[+]
I was wondering in your initial posting with the enhancing neck (+5 earth) were the numbers you have down while you were wearing the neck or after you took off your enhancing gear.
This is what I bar spell in and I can't get higher than 158 (Af3+2) or 159 (Augurs). Fenrir.Motenten said: » If we assume that the +resist rate (m.eva applied to a single element) is used directly counter to the mobs +m.acc, and make use of the m.acc tests that used enspells to determine resist rates (not sure these tests were ever completed, so going off of vague semi-conclusions), we can evaluate the value of +20 resist. I recall some testing a while back, roughly ToAU era, to determine the effect of elemental resistance and it was discovered that there are 'tiers' which doesn't fit the m.eva model. For example, I think at Lv75 if you go from 200 fire resist up to about 220-240, Tiamat's airborne attacks will go from doing ~40-50 damage per hit down to about 0-10 damage per hit. I don't have any links to any of that old testing, but I'm pretty sure they discovered at least two noticeable tiers. One was around 120, and the other around 220-240. However, the testing was all on Tiamat alone, so those tiers could shift depending on the level of your target or who knows what else, as well as almost certainly shifting based on your own level (different tiers at 90 than 75). I'm pretty sure those were just descriptions of "I get mostly 1/2 resists" vs. "I get mostly 1/4 resists" vs. "I get mostly 1/8 resists"
Correct me if any of this is wrong, but I think it summarizes the MEva/MAcc/MHit Rate findings. Recently someone posted test data on mnk's set proc rate, with extremely high sample sets (10k+ samples per test), resulting in proc rates we can be quite confident in. It was implied (though I haven't seen the testing for) that nin and dnc have similar proc rates, and thus it's highly likely that whm's set will follow the same tiers.
The tier estimates for 2/3/4/5 pieces equipped is 2%/4%/7%/10% rather than the previously assumed 2%/4%/6%/10%. This gives the 4th piece a greater boost, but also reduces the benefit of 5/5 sets compared to how they were viewed before. Revisiting my original estimates, rather than average spell damage dropping from 94% base to 90% base, it now drops from 93% base to 90% base, giving a 3.2% reduction in damage rather than 4.2%. Given that a 4.2% reduction in damage appeared to be weaker than what was likely provided by the relic pants, 3.2% is almost certainly weaker than the overall benefit that the relic pants provide. At this point, even given the possible outlier conditions where 5/5 would probably be better, I'd still recommend 4/5+relic pants on pretty much anything you'd use barspells for.
[+]
Fenrir.Motenten said: » Since it seems you're having trouble understanding how statistics works, I'll try to step you through the math. But first, some basic corrections: Quote: You are saying that there is 30% chance of a procc in 6 spell casted when on 100 spell it's 10%. No. 30% (31% actually) applies for the 4/5 build, which I assume has a 6% proc rate, not 10% (that's the 5/5 build). Quote: For me 10% means that aproximatively every 10 spells it will procc once. As a rough understanding, that's fine. However you need to know how the actual math works before you start trying to draw conclusions from that. Quote: If there is 6 spell casted, that pretty much makes it a 0.8% chance. No. I'm not sure how you ended up with 0.8%, but it's nowhere near the correct methodology. Quote: Unless you plan holding a mob forever, you will always get more out using clerics then af3 +2.Unless you know mobs that cast 100 spells in 1 fight regulary? No. Statistics are about long term averages. It's possible a single mob can cast a single spell and you'll get a proc activation. It's also possible that a single mob can cast a hundred spells and you won't get -any- proc activations. Neither of those specific cases matter. The only thing you need to care about is what the average reduction is over long periods of time. Quote: I technically don't change it, 10/100 aprox = 0.6/6 That means you pretty much have 0 chances of procc if you are fighting a monster that does not spam magic. That's very wrong. It sounds like the "haste takes time to proc" misapprehension that was so popular (and ridiculed) several years back. It demonstrates that you're fundamentally mistaking how things work. 10% proc rate means that, yes, on average 10 out of 100 spells will get the proc effect. And yes, out of every 6 spells, 0.6 of them will proc. 0.6 is NOT equal to 0, nor does it mean that you must have 10 spells cast in order for one to proc. 0.6 is not a 'real' number; it's a representation of the fraction of events that could have occurred when your sample size is infinite (as it would need to be if you wanted a 'true' answer). Suppose you fight a mob that casts one spell during the fight. The odds of a proc is 10%, and the odds of no proc is 90%. Taken in simple terms (simple to make it easier to understand; this is not mathematically correct), if you fight 10 of those mobs and each of them casts one spell, that means that one of them will get a proc and the other 9 won't. Now, suppose that proc happened in the third fight. If you look only at those first three fights you would expect that, given 3 spells cast and a 10% proc rate, you'd have an average of 0.3 procs. Yet you got one proc. That's because the data is not the same thing as the expectation. The expectation is an abstract value; the data is an actual count. Probability means that *you don't know* when the proc is going to happen. Therefore you have to work out the probabilities for a given set of conditions. Now, to the math: In order to figure out the odds of at least one occurance of an event happening, it's easiest to work from the probability of it *not* happening. So, with a 10% proc rate, that means that 90% of the time it won't proc. The probability of two events occurring is the product of the probability of each event. The probability of getting a proc twice in two spells is 10% * 10% = 1%. The probability of *not* getting a proc in two spells is 90% * 90% = 81%. You also have the probability of getting a proc on either the first spell but not the second (9%) or the second spell but not the first (9%). Total probability: 81% + 9% + 9% + 1% = 100%. The value I gave was the probability of getting *at least* one proc out of a set number of spells. Since I don't need to know the probability of all possible combinations of procs (ie: proc on first and third spell, but not the others; proc on 5th spell only; etc), I can look at the probability of *just* the chance that no procs occurred whatsoever. That probability, for six spells, is 90% * 90% * 90% * 90% * 90% * 90% (ie: spell did not proc on any of the six spells) = 53.1%. Given that I know that 53% of the time there were no procs at all, I can also conclude that the other 47% of the time I got at least one proc. This does not say that there will or must be one proc made in each fight, but rather that, if you were to fight, say, 1000 mobs, you'd get at least one set proc in 469 of the fights, and you wouldn't get any procs in the other 531 fights. Of course probability also says that you most likely won't get *exactly* that number of fights with and without procs, but that's a separate (more advanced) matter. Hopefully this will help you understand how the math works. As much as I used to be good in maths this was just horrible to read to me. What is the result? I don't make maths with the game only test and use comon sense. My comon sense tells me to use augurs mitts + clerics until I can get the cape (*** HELP I AM TRAPPED IN 2006 PLEASE SEND A TIME MACHINE drop). So plz if I'm wrong speak english or french @_@ Vaness said: My comon sense tells me to use augurs mitts + clerics until I can get the cape (*** HELP I AM TRAPPED IN 2006 PLEASE SEND A TIME MACHINE drop). So plz if I'm wrong speak english or french @_@ Augur's Gloves are +5 Enhancing Magic skill, which *might* give you +2 barspell resistance, which might reduce damage in the long term by maybe 1%. Given that you are using relic pants (per my post above), you'd have the option for 4/5 AF3 set otherwise. If you use Augur's instead, that means 3/5 AF3 set. 3/5 should give a 4% proc rate, while 4/5 should give a 7% proc rate. That reduces average damage from 96% baseline to 93% baseline, or an average of a 3.1% damage reduction. There's basically no way possible that Augur's Gloves would beat the set bonus. The only reason to use Augur's Gloves for barspells is because you only have the AF3+1 hands. Fenrir.Motenten said: » Vaness said: My comon sense tells me to use augurs mitts + clerics until I can get the cape (*** HELP I AM TRAPPED IN 2006 PLEASE SEND A TIME MACHINE drop). So plz if I'm wrong speak english or french @_@ Augur's Gloves are +5 Enhancing Magic skill, which *might* give you +2 barspell resistance, which might reduce damage in the long term by maybe 1%. Given that you are using relic pants (per my post above), you'd have the option for 4/5 AF3 set otherwise. If you use Augur's instead, that means 3/5 AF3 set. 3/5 should give a 4% proc rate, while 4/5 should give a 7% proc rate. That reduces average damage from 96% baseline to 93% baseline, or an average of a 3.1% damage reduction. There's basically no way possible that Augur's Gloves would beat the set bonus. The only reason to use Augur's Gloves for barspells is because you only have the AF3+1 hands. /hey kalilla :3
Cute! Quote: I believe the 4 more skills take the resistance to a new tier (seemed to do half and 3/4 resist more often then others) Assuming you're basing this statement on your tests, I'd want to review the data you saw. The only tests that can directly compare the results of Augur's vs AF3 are tests 2 and 3 of the second set of tests. The other tests changed gear unrelated to the hands slot between them. Looking simply at the Quake data: Test 2 (AF3), Quake: 142 dmg: 1 time 288 dmg: 1 time 570 dmg: 2 times 572 dmg: 3 times 578 dmg: 9 times Total Quake casted: 17 Test 3 (Augur's), Quake: 70 dmg: 4 times 141 dmg: 3 times 142 dmg: 2 times 283 dmg: 3 times 284 dmg: 1 time 298 dmg: 1 time 568 dmg: 7 times 570 dmg: 1 time Total Quake casted: 31 So, full damage Quake casts with AF3 hit 82% of the time, while with Augur's they hit for full damage 26% of the time. Augur's seems better, yes? Except not quite. Note the damage done on each Quake: 568, 570, 572 and 578. There's also half-resist casts that did 283 and 288 damage, which implies 566 and 576 damage on the unresisted version. The different values indicate several different levels of mob, most likely increasing by 2 points per mob level, but maybe a +4 in there depending on stat rounding. For illustration, let's say: lvl 87 - 566 dmg lvl 88 - 568 dmg lvl 89 - 570 dmg lvl 90 - 572 dmg lvl 91 - 576 dmg lvl 92 - 578 dmg That means that the majority of the full damage spells that hit you while using AF3 hands were cast by mobs that were higher level than were casting on you with Augur's Gloves. Let's split out the full damage list and look at what's confirmable: AF3: lvl 92: 9 lvl 91: 1 (half resist) lvl 90: 3 lvl 89: 2 lvl 88 or 89: 1 Avg lvl per cast: 91.13 (ignoring the uncertain one) Augur's: lvl 89: 1 lvl 88: 8 (1 half resist) lvl 87: 10 (3 quarter resists, 4 1/8 resist) lvl 88 or 89: 2 Avg lvl per cast: 87.53 (ignoring the uncertain ones) So on average, mobs casting on you using Augur's were 3.6 levels below the ones casting you on AF3. The difference is even larger if there should also be a mob of a level to cast 574 dmg nukes in there; it goes up to an average 4.3 level gap. Naturally the Augur's test seems to be for lower damage. Therefore your reasoning for selecting Augur's Gloves over Orison Mitts is flawed. There is no evidence to indicate that the +5 Enhancing skill made any notable difference in the final results because the results are so heavily biased by mob level. Note that this should have no relevance to your testing of proc rates since proc rate shouldn't depend on mob level; that half of your testing is fine.
[+]
Fenrir.Motenten said: » Given that I know that 53% of the time there were no procs at all, I can also conclude that the other 47% of the time I got at least one proc. This is the main part Vaness needs to understand for it to make sense. She is still thinking it's 0.8% I believe. The most important thing we've learned is that it seemed promising before that in most situations 4/5 was better than 5/5 because of assumed 2/4/6/10 now that we have pretty much confirmed it's 2/4/7/10 like Morten said 4/5 is even more valuable and 5/5 is less valuable it makes it an over whelming easy choice. So I haven't really been following the topic...
The optimal set would be +2 for everything but pants? I have some questions to this if I may. I read the whole thread so far and even if my native language was english I know I wouldn't understand all this math since I suck at it so bad lol.
So what is found out with this is Cleric's Pantaloons > AF3+2 and logically therefore Cleric's +1 too. You also figured out that the AF3+2 Body wins over the Blessed Briault NQ, due to 10MDB > 5MDB. On the first testing with just bodies it says "only gives MDB with Solace on" but if so, then it's the Solace only that gives you MDB (Which goes along with the description on wiki) and the bodies do nothing for you at all but the Solace does the job, and AF3+2 is only better because it enhances the Solace which pushes your barspells then. What does the "enhances elemental resistance spells" on the Blessed Briault then do for us? Again according to wiki the description says it does give 5MDB and on the discussion page it says the same on posts of 2007 where there was no Solace yet. So I believe your testings but also makes me wonder since it's known for so long (longer than Solace been invented) that the body is supposed to give 5MDB. I'm confused on this. Also I wonder what the Blessed Briault +1 does for us. I tried to figure out the whole "What's so awesome about the HQ?" question before but never really understood. Maybe if NQ = 5MDB then HQ = 6MDB? or =10MDB? Anyone knows? And even if it had 10MDB, would AF3+2 still outperform it then allthough both then reach the 10MDB? Maybe because of the Set effect? Then about the hands discussion. If I understood it right, if I can break a tier to the next barspell number shown on my screen then go for Augur's Gloves but if not rather throw in the AF3+2 since I get higher chances for a set procc then. And as a last question that I was wondering allthough it's a bit offtopic: How can I push up my enfeebling resistance? Is there a way at all? Because all the gear that is known for barspells always is about the elemental barspells but never about the enfeebling part. I thought maybe via the whole Solace thing since it says "grants an additional Magic Defense Bonus to the associated element" and since enfeeblings have their own element too maybe Solace works for this as well? But gearwise I can't really think of something besides throwing Fast Cast and Haste in. Odin.Gollum said: » And even if it had 10MDB, would AF3+2 still outperform it then allthough both then reach the 10MDB? Maybe because of the Set effect? I don't see how the HQ Clerics could beat the AF3+2 even if it did give 10MDB like you said you're getting the set effect. Odin.Gollum said: » Then about the hands discussion. If I understood it right, if I can break a tier to the next barspell number shown on my screen then go for Augur's Gloves but if not rather throw in the AF3+2 since I get higher chances for a set procc then. The 7% set bonus from 4/5 AF3+2 is favorable, it was very close when it was thought the set bonus was 6% I would personally recommend just going with the Orison AF3+2 if you got Cleric's Odin.Gollum said: » On the first testing with just bodies it says "only gives MDB with Solace on" but if so, then it's the Solace only that gives you MDB (Which goes along with the description on wiki) and the bodies do nothing for you at all but the Solace does the job, and AF3+2 is only better because it enhances the Solace which pushes your barspells then. What does the "enhances elemental resistance spells" on the Blessed Briault then do for us? Again according to wiki the description says it does give 5MDB and on the discussion page it says the same on posts of 2007 where there was no Solace yet. So I believe your testings but also makes me wonder since it's known for so long (longer than Solace been invented) that the body is supposed to give 5MDB. I'm confused on this. Odin.Gollum said: » Then about the hands discussion. If I understood it right, if I can break a tier to the next barspell number shown on my screen then go for Augur's Gloves but if not rather throw in the AF3+2 since I get higher chances for a set procc then. Odin.Gollum said: » And as a last question that I was wondering allthough it's a bit offtopic: How can I push up my enfeebling resistance? Is there a way at all? Because all the gear that is known for barspells always is about the elemental barspells but never about the enfeebling part. I thought maybe via the whole Solace thing since it says "grants an additional Magic Defense Bonus to the associated element" and since enfeeblings have their own element too maybe Solace works for this as well? But gearwise I can't really think of something besides throwing Fast Cast and Haste in. Ice Resistance Enhances Resist Paralyze Magic Evasion MND |
||
|
All FFXI content and images © 2002-2024 SQUARE ENIX CO., LTD. FINAL
FANTASY is a registered trademark of Square Enix Co., Ltd.
|
||
