Lv.99 MNK TP Set

Answer:



I had the same question TY Austar: Some times it is hard to find good answer to good questions though the wall of flame that often fills the forum window. I almost missed your short post but glad I scrolled back up and saw it.

Thats funny. Math is hard. Average has no meaning. And MNK should get marches and cureV so we can delete all other jobs.

You have no idea what you're talking about. Pchan is saying that someone using a TP set with less TA/QA but more attack/STR will probably win 9 out of ten parses, because they'll hit for more, even if they're potentially hitting fewer times. One out of ten parses, those TA/QA procs will trigger and you'll have done more damage than normal, hence the "spike in damage." The point he's trying to make is that you shouldn't rely on that 1% QA from Windbufffet when you could benefit constantly from the STR (and PDT, and Subtle Blow) from Black Belt, or other stats on another belt. He's saying the more reliable your damage is (IE, the less variance you have between parses), the better the TP set it.

Whether he's right or not is another question entirely, but his train of thought is not hard to follow.

/facepalm

What if you are fighting a blob, facing south, formless strikes is down, on windsday,waning crescent and QA/TA doesn't proc? Then you fight a skele facing north, iceday, new moon and QA procs 3 times in a row?

Fixed for accuracy.

You're forgetting moon phases taint

I'm not arguing about whether Pchan's logic is good or not, just that it's not hard to follow, and the guy I quoted completely misinterpreted what Pchan said.

I regurgitated what he said. I didn't say he was right.

So should we remove all ta and da from tp sets also so we have even less variance?

18 MNKs wearing nothing but DEX and Haste gear being controlled by 2 people sounds about right

ftfy

May as well go ahead get used to the squawking of the post-update world.

Variance measures the average quadratic variation between the current DPS you are doing and the average DPS you will be getting in the end. Current math is done by assuming mobs have infinite HP and that you fight them an infinite amount of time. In reality mobs have between 5k and 10k HP sometimes even less, so you actually don't care of ***like quadruple attack. When it procs, cool story bro, when it doesn't proc it's inferior to anything. You cannot expect a QA to proc more than once a fight and on average it's a lot less likely than once a fight. Therefore even though the wierdest setups can lead to slightly increased damage (we're talking less than 1% increase on average) if you change the model to a realistic one where mobs have 10k HP (say), the advantage is becoming useless : it's pointless to win a parse 10% of the time by a large margin. It's better to win in 90% of the time by a small margin.

Using the above philosophy, what's your preferred MNK TP set?

It's "Pointless" to win a parse on the other nine too since reducing variance won't change your dot in a way that will cause 9/10 voidwatch mobs to die a grand total of 1 second faster. So people will gear for the reason people always gear for.......maximized damage over time because that leads to higher epeen. It's the same reason people make glowing weapons, so people will look and say 'wow' right?

There is nothing new under the sun. People have always been primarily concerned about two things: 1) Health. 2) Being important. There is no practical difference in killspeed when 1/12th of your DD's has either a .1 advantage in dot with high variance or .1 fewer with less variance.

But in any situation where fighting lasts long enough over one or multiple mobs, the higher dot option with more variance will be better. The variance evens out over time.

Obviously it's spharai setup. The 40 attack bonus and the fact that shijin spiral has less variance (no crit) than victory smite makes it superior generally. The best TP set doesn't exist but I can give you an idea of a best TP set in the following scenario :

Not using berserk (damage taken nerf soon), using impetus, focus, /war, agressor, RCB as food, 2 x marches from relic horn and 2x minuets from relic horn, dia2, boost dex, haste.

Best average damage :

spharai/ potestas
usukane+1 / rancor / moonshade / brutal
tantra +2 body / nomkahpa / rajas / epona /
atheling / twilight / byakko /usukane+1

Least variance amongst best averages damage sets :

byakko -> tantra +2
twilight -> black belt
(probably)

Also I know idiots understood that one should avoid DA/TA/QA gear, but what I said is that when gear sets have comparable damage output you should look at the least variance set.

So what I'll suggest to you noobs is to use motenten spreadsheet and try out 4-5 set that are the best DPS and then take the one that doesn't have thaumas coat, windbuffet belt and has a BB, if it exists.

edit : forgot to say that byakko should be augmented to the best (which makes it dumb to use it in the first place if you look at the augments) and that I'm running my numbers under the new ratio cap assumption.

Except with thaumas coat and WB belt you have over a 10% chance to multihit. What are you fighting that doesn't last 10 attack rounds (where on average you would benefit from multihit). Therefore, variance is low enough to go with what the spreadsheet says does the most damage.....which is NOT your setup.

Don't argue math with me. I'm the best.

No, you are not the best. You get laughed out of every BG discussion due to egregious mistakes that favor your bias. You let your opinion sway your inputs rather than letting math dictate your decisions. You have never once admitted you were wrong even in the face of overwhelming evidence. In short, don't come into the kiddie pool demanding respect when you have already been exposed as someone who just can't grasp simple critiques that show how ridiculous some of your statements are.

At the risk of putting words in his mouth, it would be, but necklaces are weak and Rancor is so far ahead any potentially more stable option that it ends up being a wash.

For this reason, I don't think pchan would object to, say, a SAM/DRG/DRK TPing in mala.


Also, I think pchan's larger point is that math is simply a way of describing and organizing data. It is not a method of evaluating it. You don't get an endrun around critical thinking just because you understand how to read statistics.

Pretty sure Kvazz was using the invisible sarcasm font.

he'd just ask why they aren't all on MNK





Guy math is all about evaluating data, that's its purpose. Critical thinking is what gets you the stats to use in the math.

It's really not, though. The belief that it is is how we come up with silly arguments like obesity is good for you since the obese are less likely to die of frequently-weight-related illness, which is a conclusion based on data that doesn't control for the fact that the obese are more likely to be regularly seeing doctors who can identify and treat them, an important fact the study here doesn't control for precisely because the researchers couldn't quantify it.

Pchan's point is that DPS models are based on mobs with infinite health. The fact that such models are the only way we can calculate DPS doesn't make him wrong.

I fail to see how a monster with infinite health affects the outcome of a calculation. It accurately represents the chance of occurrence and average gain of x stat over y time regardless of the number of monsters being fought. Whether a QA procs more than once per fight or not the average value of the QA procs over all fought monsters is still superior to other options, that's why it's called an average value and that's why it's considered the superior option.

Just because you notice a difference from a different stat more often doesn't mean the sum of its parts is superior to the gains from the infrequent stat

There you go again. Of course it doesn't affect the outcome of a calculation.

The fact that no monster actually has infinite health and most in fact have very little health certainly affects the relevance of that outcome.

Determining how much it does and how much you ought to care is where the critical thinking comes in.

EDIT: for the record, I'm not convinced by pchan here, mostly because I DON'T care about the monsters with 10k health; situations where I'm fighting 10k health monsters and my killspeed matters in my experience value uptime-on-target and efficiency way more than they value raw damage, so why wouldn't I gear for the 100k+ health fights where 1 or 2% DPS actually matters? But that evaluation itself is in no way determined by the math; it is a balancing test where the math is one factor.

no... it doesn't. that's the entire point of my post. unless you're constantly fighting monsters with so little health that a QA proc on the TP phase is not speeding up your kill, then it's still the superior option over any stretch of time regardless of how many fights it's broken up into. Individual monster HP is mostly irrelevant when it's a measure of the effectiveness of a value over time.

a monster with such little health (so little that the QA is not being fully utilized) wouldn't care what you're wearing to begin with, it's going to die quickly regardless.

No, it doesn't matter. At least not to the extent you think it does. Whether one mob has 1,000,000,000 hit points or 100 mobs have 1,000,000 hit points the amount of multiattack will, over time, be the same. The only difference will be if a mob dies on the first hit of what would be a multi attack, thereby nullifying the multiattack.

So yes, for a WB/thaumas set up about 10% of the time you will multiattack. That doesn't mean 10% of the time the multiattack will be superflous considering:

1-12 other DD may get the killshot
Your weaponskill may get the killshot
The mob may die on the last hit of the multiattack instead of the first of what would have been a multiattack.

The % of those things NOT happening (i.e. you do kill the mob with the first round of what would have been a multiattack) is smaller than the overall DOT difference in the spreadsheets. Therefore, the thaum body setup wins. In other words: the best is the best no matter how you try to rationalize it.

Individual monster HP is the only relevant issue inasmuch as the whole point of DPS is to get rid of it.

...which leads us to:


Swell, but why do I care about the effectiveness of that value, then?

What? Circular reasoning isn't going to win you this debate.
Why would I be arguing to begin with if I didn't believe DA/TA/QA resulted in faster kills over time? Attempting to breed a nonexistent issue in the middle of a losing battle is silly, unless you actually believe that DA/TA/QA doesn't increase killspeed because a monster has 5000 HP, in which case I just feel sorry for you I guess? Willing to help you understand though.

Why would I use the inferior, lower attack frequency armor when increasing my attack frequency via DA/TA/QA is superior for killing things over time?

The value of the less frequent event is still superior because it carries more weight than the more frequent event(s). The amount of HP that the monster has in measuring the relevance of stat vs stat in terms of one dropping off is not a factor, that that is why you were calling HP into question to begin with. Look at individual monster's HP pools as part of the communal HP pool of several monsters and suddenly average values and infinite HPs make more sense.

Well, generally you wouldn't, because this isn't WoW; the stuff with DA/TA/QA is often so far ahead of other actual options that it isn't really a balancing test at all. See: mala.

Pchan is referring to one of the precious few situations where it really WOULD seem close, which is BB (and other gear that's individually more damage than Usu+1) vs. windbuffet for TP; is the multiattack on Windbuffet really helping you more than the pretty significant attack and possibly STR you'd be getting from other options?

(the answer is, "sometimes")

Slight devil's advocate, but one thing to consider is how much a particular job benefits from the TP overflow that occurs from the multihit procs. Smite benefits a lot but overflow on Shijin could barely be called a "benefit".

Anything can be rationalized down to "sometimes" and/or "situationally", what's often left out is how often that sometimes actually occurs. It isn't frequent.


Before that's misinterpreted by someone and turned back around onto DA/TA/QA, it's a different matter altogether.