Magic Accuracy & Day of the Week

[Wizerd]

My question's pretty simple, but I decided to make a new thread here to ask a few other questions regarding magic accuracy and testing methods, since I'm making a new thread anyway.

Primary question: If you're casting a spell of the element that the current day is weak to, do you get an accuracy bonus?

For example, on Iceday, do you get an accuracy bonus for casting Fire spells?

I always assumed not, and that the only elements affected were Ice (random +potency and possibly a random accuracy boost), and Wind (random -potency and possibly a random accuracy sink). However, in my persuit of testing and quantifying Magic Accuracy, I ran into an outrageous result I didn't expect. My sample sizes are still quite low, and I haven't had a chance to test another element to see if I got similar results.

My basis currently stands on a few assumptions: Magic Accuracy being calculated similar to accuracy, and Even Match monsters (specifically Qiqirn RNGs in Aydeewa) having Magic Evasion levels similar to what would make sense for their job, when comparing to sensible Melee Evasion levels. That is to say, skill over 200 would be considered 0.9 Magic Accuracy, with each point before 200 being 1 point. INT would be 0.5 Magic Accuracy, and of course anything that said Magic Accuracy itself would be just that.

My original tests were Water, Fire, and Blizzard on Even Match Qiqirn Archaeologists. I didn't research their INT or MND or anything, but I do know that all lv75 of this same mob will spawn with the same stats, as with other monsters in the game. The character I'm testing on has 4 Ice Accuracy merits, which is why I tested Ice alongside Water and Fire (which Qiqirns apparently have no known weakness to).

I did a series of 200 casts with the first two elements, all on neutral days and in neutral weather. Both sets came out very close, when considering the amount of nukes that landed for full damage vs the amount that didn't (58.5% and 59%, or 117/200 and 118/200). This is pretty much exactly what I expected, and predicted two paragraphs earlier in my current document, a day before doing the tests. I figured I pretty much had it nailed down in preliminary terms: 267 Magic Accuracy naked (238 Elemental Magic Skill and 66 INT) vs roughly 300 Magic Evasion, when worked out Melee Accuracy style comes out to 58.5% (correct me if I'm wrong here).

However, the first snag came when I started casting Blizzard. I only managed to get 133 casts off before having to pause when it switched to Firesday, but it's sitting at 73.68%, which is considerably higher than 12 Magic Accuracy would theoretically give. Instead, it gave around 15% more, which could be simplified to 12% due to low sample sizes. I kind of shrugged this one off due to the low sample size, however, due to the fact that after only 100 casts, the first 2 elements were showing higher results too.

The second snag, and the one this post is mainly about, came when Firesday started. I figured I had at least a very rough baseline for the first two elements, so I wanted to toss on an NQ staff and do a bunch of casts to see if it was closer to Magic Accuracy +10 or +10%. Obviously I couldn't use Fire, so I started with Water (Water beats fire, as we all know, but has no known bonuses on Firesday).

With a base of 267 Magic Accuracy, the theoretical Accuracies should be either 277 or 293, respectively. In order, the full damage rate should either be 63.5% or 71.5%.

Unfortunately, the result I came out with after 100 casts (low sample size, but this was enough of an extreme to make me post this) was 87%. 87/100 of the casts hit for full damage, with 9 half resists and 4 quarter resists.

There are two explanations I could think of: Some of the Water casts were given an accuracy bonus for being cast on Firesday, or Magic Accuracy has more of an impact on a per-point basis than Accuracy.

To extend on the latter, I'll bring Accuracy up first. Adding two points of Accuracy to your setup increases your Hit Rate by ~1%. One point being ~0.5%. The Accuracy vs Evasion match-up on an Even Match monster gives you a Hit Rate of 75%. What I am suggesting is that the match-up is the same for magic, but instead, one point of Magic Accuracy gives you roughly a 1% increase in your ability to land spells full-on.

This would facilitate both the 4 Ice Accuracy merits and the idea that the NQ staff I was using gives +10% Magic Accuracy (rather than +10 Magic Accuracy), given the small sample sizes. It would also explain why spells are so difficult to land on some monsters, and so ridiculously easy on others - there would be a much more narrow range of numbers where your Accuracy would truly be affected.

I realize I kind of went on a tangent here, but the original question still applies, and if any of our more math-inclined members want to give me some insight on my results (to reiterate, I realize the sample sizes are still small; I plan on testing more to increase the baseline sizes to at least 500-1000 just to start), I'd be very grateful. I can post the full results of what I have so far if it helps, too.

[Suiram]

The full results would be nice, since I feel pretty confident that modeling Macc as you would model melee accuracy is a flawed premise. What you saw has been seen in the few other tests I've read regarding magic accuracy, where under certain circumstances you see double the predicted bonus. What would be most beneficial would be to figure out how and why that occurs.

For instance, there was that JP flowchart that suggested that above 50% hit rate the benefits from adding skill are doubled. Now if that is true, and if it's calculated after staff bonus, then an otherwise X increase in macc from staff would result in an actual 2X increase since it's doubling the bonuses from skill above a certain threshold.

In any case, I'm glad that someone is continuing research into Macc, and I'm moreover glad that it's someone like you.

Edit: If you are open to suggestions, what I would personally like to see tested, which is along the lines of what I was saying above and a variation on that flowchart I mentioned before which is the closest thing I can think of that falls in line with the minimal data that has been presented in previous tests is something like this:
Step 1: Base Macc (either a constant, or dependant on the spell)
Step 2: +Macc gear and +Staff (1 macc = 1% hit rate? 10%/15% fixed hit rate increase for staff?)
Step 3: Caster INT vs. target INT (The chart suggested it's 1% hit rate per dINT up to +10 dINT, but could be different from +10, and either way, where does it start?)
Step 4: Caster Skill vs. target elemental resistance up to 50% hit rate (chart suggests 2 skill = 1% hit rate)
Step 5: Caster Skill vs. target elemental resistance above 50% hit rate (chart suggests 1 skill = 1% hit rate)

This would just be a starting point, there's a number of things that could be different like whether staff bonus is fixed % hit rate, whether the ratios are accurate 2:1% and 1:1% for INT and Skill, the order of steps 3, 4 and 5, and the location of the inflection points for INT and skill. Or the whole thing could be bogus, but like I said, it would explain a lot of stuff if it did work that way, such as increases in earlier steps doubling the expected bonuses due to pushing skill calculations from step 4 into step 5.

[Not Kuno]

Pretty sure you do. The best way to test it is with Helixes and Obis. They'll last longer (which depends on m.acc) if you have an obi of the day on.

[Wizerd]

As far as the Magic Accuracy stat itself is concerned, I don't have access to a lot of things with that stat on it. Looking at the list on wiki, there doesn't seem to be a whole lot of easy-to-obtain pieces, either. I do have Balrahn's ring on the character, and Chasuble doesn't seem common on AH (at least, none up right now). I have Water, Wind, Light, and Dark grips currently.

I actually have 2 earth accuracy merits on the character as well (my wife was planning Ice+Earth before she quit, for Paralyze and Slow), but I was planning on cancelling those as I got more merits to replace them.

My friend's character has Morrigan's Cuffs, and since MAB has no effect on accuracy, I could use those as well if I tested with his character (a total of +16 Magic Accuracy if using Water - Qiqirns are weak to Wind so I can't test that yet).

However, part of the problem with your suggestion is I have no notable way to get *under* 50% accuracy on the monster in question. I have 238 Elemental Skill on the RDM character I'm using now, and on the same character I have BLM70, but with 4 merits into Elemental Skill, it has 259 I believe, which is actually higher than the naked RDM. Couple that with the fact that level correction would be taking place as well and there's just a whole lot of problems.

Qiqirn Enterprisers are 69-71, so I can easily figure out which ones are lv70 on a lv70 character, but that doesn't solve the problem of a much higher skill and INT - I would be dealing with a Magic Accuracy rate of well over 75%, even naked.

The last thing I thought of was WHM/SCH, using Dark Arts to get the characters skill up, but not up too much, but I just checked and apparently that will give me 246+8, which is, again, a problem.

Any suggestions on getting less than a 50% Magic Accuracy rating on this EM Qirirn, while still being over 20% and from a lv75 character? I don't think WHM/BLM will work, because 114+8(122) plus the Magic Accuracy from INT won't quite be enough. Using the possibly flawed melee accuracy formula, with apparently 62 INT, this would give me 153 Magic Accuracy naked - more than a hundred lower than the naked RDM.


Anyway, here are my full result tables so far:


Monster: Qiqirn Archaeologist
Level: 75
Elemental Skill: 238
INT: 66
Bonus Magic Accuracy: 0 (+12 for Blizzard)
Theoretical Magic Accuracy: 267 (279 for Blizzard)

Water - 200 casts
Full: 117
Half: 53
Quarter: 17
Eighth: 13

Rate of full nuke damage: 58.50%

Fire - 200 casts
Full: 118
Half: 49
Quarter: 15
Eighth: 18

Rate of full nuke damage: 59.00%

Blizzard - 133 casts
Full: 98
Half: 22
Quarter: 6
Eighth: 7

Rate of full nuke damage: 73.68%

Water with Water Staff(NQ) - 100 casts
Full: 87
Half: 9
Quarter: 4
Eighth: 0

Rate of full nuke damage: 87.00%



I also believe we should only be looking at full damage vs not full damage, when it comes to accuracy. Also, in each result set, half resists appeared close to 60% of the time among the resist results.

[Wizerd]

Alright so back to testing. My plans tonight are to redo the Water Staff with Water tests on non-firesday neutral days, up to 200 casts to compare to the first test, as well as 2 more sets of 200 casts to compare to the first set, those being Magic Accuracy +4 vs Elemental skill +4.

I wanted to use higher values for the latter two but both Wise pieces (head and legs) on sale are overpriced and I don't feel like paying a ton for testing; already wasting Shihei. <_<

Edit: After finishing 200 casts of Water with NQ Staff on, and getting two different results, both on Windsday in Wind Weather, I decided that the first result I got was simply random. The following are the two sets of 100 casts I just did:

Water with Water Staff(NQ) - 100 casts
Full: 76
Half: 18
Quarter: 4
Eighth: 2

Rate of full nuke damage: 76.00%

Water with Water Staff(NQ) - 100 casts
Full: 85
Half: 12
Quarter: 3
Eighth: 0

Rate of full nuke damage: 85.00%

[pchan]

So from what I see you average 82.6% on water with NQ water staff, and 58.5% without staff. Sleeping lv 80 puddings with High base skill with and without HQ staff resulted in 48% and 63%. So here is what we observe :

NQ staff gives + 24% on a "low" level mob with no magic resistance on the element.
HQ staff gives + 15% on a high level mob with abvious resistance to the element.

This proves that staff don't give static accuracy% regardless of the mob, but I think noone really thought that. The effect will be greater if the mob resists less.

[Wizerd]

So from what I see you average 82.6% on water with NQ water staff, and 58.5% without staff. Sleeping lv 80 puddings with High base skill with and without HQ staff resulted in 48% and 63%. So here is what we observe :

NQ staff gives + 24% on a "low" level mob with no magic resistance on the element.
HQ staff gives + 15% on a high level mob with abvious resistance to the element.

This proves that staff don't give static accuracy% regardless of the mob, but I think noone really thought that. The effect will be greater if the mob resists less.

The first problem with this post is your results: I assume you were tracking successful sleep casts vs unsuccessful sleep casts, regardless of the duration of sleep. Sleep can last 60 seconds, 30 seconds, 15 seconds, or full out resist. 90, 45, and 22 seconds if we're talking about Sleep II. Unless you waited at least half of the duration each sleep before waking it up and resleeping it, your results are counting all Full hits, Half hits, and Quarter hits as successes.

If we take that and apply it to my results, we get 91.33% without the staff, and 99.33% with it. Just from these results, we can see how flawed that logic is.

The second problem is that you've made a very large generalization about high level and low level monsters. You tested against lv80 monsters, without knowing what kind of level correction Magic Accuracy suffers from casting on higher level monsters. Furthermore, we don't quite know how Magic Accuracy works yet.

Complete tests (I'm aiming for a sample size of 1000 for every test set I do), using different elements and different values to map out what affects what would be required before even thinking about increasing or decreasing the level of monster we test on. And when we do increase or decrease the level of the monster, we have to take a specific controlled result set and replicate it on the new monster level. After figuring out the math on those sets, we have to increase or decrease it by 1 more, to see if the pattern sticks. Then and only then is it possible to accurately gauge mobs of higher level.

Of course, this is all my opinion. Now, if you did record sleep durations and your successes include only full sleeps, then I stand corrected and applaud you for thinking ahead. I believe that Magic Accuracy has a 95% cap on full hits, and about a 99.5%-99.8% cap on hits that aren't fully resisted - very different numbers, and so both are very important to record.

[pchan]

Yes I only accounted for landing or not landing sleep in my tests. This just means the actual landing rate is lower for full duration sleeps.

Also for accounting partial resists of nukes, an easy way I found when I nuked VT wamouracampa is counting the total damage you are dealing and comparing it to the maximum that would occur with no resists. Count 1 for no resist, 0.5 for 1/2 resist etc. From your tests :

(248+ 39*0.5 + 11*0.25 + 2*0.125)/300=90.1% with NQ staff + water.
(117+53*0.5+17*0.25+13*0.125)/200=74.7% without staff.