Kraken Club Offhand Hits Per Round Simulation

Right... As you increase sample size, the error propagation approaches zero. If you understand basic stats, that makes sense. Error propagation is contravariant to sample size.

The longer you hit the wall for, the smaller your variance gets, approaching 0.


Because my error is 0 when you propagate it, so it does in effect take it into account.

Thats the point of the Sanity Check cells. As long as they are equal to 100%, propagated error formula will logically produce 0% as the error, due to contravarience.


I didnt bother to add it, but I could if someone cared.

Follow up attack on Raetic weapons scales with MP so if I really cared, I would need to punch in a Max MP cell for it.

And you think the 7 DPS difference you get with some items on a spreadsheet constitutes as "better?" what if the calculation is skewed because you have an average 12.1 tp return instead of an average 11.8 tp return and it thinks you need one less hit to reach your desired TP value? Not to mention numerous other advantages a simulation provides.

Congrats, you wrote a borderline unintelligible line of code that doesnt even do what my sheet does.

You arent capping your hit rate at 8 there, and you arent apply MH/OH accuracy to the hits, and you have assumed kclub offhand, which my sheet does not assume, and you dont have OAx support for mainhand.

So not even remotely close to what the sheet does. Try again.

Someone doesnt know how the DPS spreadsheets works, lol.

You should maybe go actually look over how the DPS spreadsheet works before saying stuff like this, because you're outting yourself as "I dont know what I am talking about"

First off, I presume you mean 121 and 118 TP return, because it isnt 2015 anymore last I checked.

Second, the DPS spreadsheets account for over TP rounds and hits to achieve TP to WS, and TP return from WSing as well, all factored in with accuracy.

Thats why on the first page of the spreadsheet there's a cell for your TP threshold to WS, since twiddling that you may realize some WSes you actually dont want to WS as soon as you hit 1000, IE Savage Blade, sometimes.

The spreadsheet accounts for that, it is a complete simulation that simulates every possible hit iteration and covers the entire spread of all types of attack rounds, its pretty close to all encompassing (Though I think most DPS spreadsheets still do not have support for Follow up attack round weapons yet, because very very few people care about them)

You'll never approach that value in FFXI. I can run 10K fights at an hour each and still have variance greater than what you'll accept.

You aren't propagating anything, its just an arithmetic mean.

I am capping my 8 hits there. I know it's way over your head since you can't just copy and paste it from someone else's' work.

But, I just wrote it quickly to make a point that I know the formula and your sheet doesn't mean anything.

I'm talking about additional hits. You know, the ones that naturally are 10 but with store tp would be 12.8 or 11.8, normally floored of course, but your spreadsheet doesn't do that.

except the TP return values are naturally wrong, which means your actual WS values are also wrong, and potentially the amount of hits required to get there are wrong.

It isn't a simulation. You very clearly don't know what a simulation is or how it operates.

On average? Yes. For every instance where RNG screwage might make you miss a notable TP milestone or such more often than normal, a different situation or set of RNG rolls will likely cancel it out.

The spreadsheets take into account things like the chance your set has to barely miss your preferred TP threshold, anyway. They aren't so simple as "this is X number of hits and X amount of STP, which will result in X average damage WS"; they actually do a variant of your iteration method to get the chances of ending up with each amount of TP after each round, and calculate your DPS based on a combination of all of these possibilities.

A set with high sensitivity to variance will negatively impact your DPS and can be worked around just fine with a spreadsheet.

There is no variance in the spreadsheets. And how would you know if a set was high variance or not in a spreadsheet built like that? You don't. Why? because it only gives you averages.

I double checked and ran your code and, no, it really does not appear to be capping at 8.


You might want to go look at a list of the maintainers of DPS spreadsheets, before you continue outing yourself as not knowing what you are talking about.

Every time you post, you look more and more foolish.

But by all means, please feel free to show me where your formula accounts for OH/MH acc, and where it rounds down to 8 hits (and also applies OH acc properly to the truncated hits)

The fact your code doesnt have a >, <, Min, Max, or 8 anywhere in it kind of displays the code smell instantly.

You're out of your depth here, instead of acting rude towards people who tried their best to explain the faults in your approach, you should consider listening to people way more seasoned than you on the topic.

I get that its easy to get emotionally attached to something you have created, I've written plenty of blocks of code in my time that I felt proud of, and its a wonderful feeling to create something that feels useful.

But the number one rule of being a good programmer is listening to the feedback, the moment you put emotion before logic and stop listening to your seniors who are taking the time to point out how to improve your code, you have become a bad programmer.

Stop and take a breath. Your program is decent, and it serves the purpose of demonstrating the variance of DPS in small bursts.

But if the goal is to simulate choices of "which is better, A or B" it does not serve that purpose then.

When you have variance that actually hurts your average (say, a set where you want to WS at as close to 1000 TP as possible, but it puts you at 990 TP half the time and 1200 TP the other half of the time) will manifest in the spreadsheet as lower average DPS.

As for variances that don't negatively impact your average DPS, I'll say that for general situations (as there are always exceptions, like if you're trying to do a 6-step SC in Vagary), I personally don't think this type of variance is actually something that is very much worth measuring or basing your sets around to reduce. Over time, every time you have a low roll you're equally likely to have a high roll that cancels it out, and if your groups are killing things so skin of the teeth that an unusually unlucky set of DPS rolls could cause a loss (or other negative result), it's just as likely that you could be on the other side and have an unusually lucky set of DPS rolls cause a win (or other positive result).

3 qa, 12 ta, and 25 da gives me 4.83362221, looks like it's capping at 8 to me.



maintaining someones work doesn't mean you made it.

it doesn't need one, this is an perfect average that you're so obsessed with.


nah, that would take more than a few minutes. i'll let you figure out how i'm capping at 8 hits before i waste my time with that.

as for the rest of what you said, its pretty apparent you're the one incapable of listening. its actually flabbergasting that you are against having a tool that mirrors actual gameplay. like seriously. why wouldn't you want something to help you gear around real situations instead of only fighting a wall for infinity with no regards to situations such as maintaining AM3? i think you're just afraid your "maintenance" won't get used anymore honestly.

that's exactly what it does. you wouldn't know, since you've probably never used my full DPS simulations and if you did, wouldn't know how to interpret the information

not necessarily. the law of averages would say so, but you aren't approaching infinity in this game. most fights are over in several minutes.


if that's the case, I doubt it's one or two pieces of gear. maybe in a one man zerg you can get a streak of bad luck, but a streak of good luck won't cancel it out. look at it this way, if on average you beat it, the bad luck streak is a loss. the average is a win, and anything over average is a win. reducing the variance towards the average is going to increase your win rate.

Once again, outing yourself as not knowing anything about the DPS Spreadsheets at all.

Stop posting, go spend a day or two actually learning how they work from start to finish, step through the cells and observe.

You'll learn a lot.


If you believe the TP returns from WSes on the DPS spreadsheets are incorrect, please make a post detailing where/how this is, preferably by creating an example sheet and uploading it and going "See here in this cell? It says its getting this TP return but it should be this according to my math"

Otherwise, if you cannot do that, then I have no idea what you are talking about. Last I checked the TP return from WSes on the sheet is correct and Ive never heard anyone mention they are wrong.


When you simplify the error propagation formula for a 0% error, you just end up with 0/1, which is 0. Anyone with basic stats knowledge knows this. Thus I am "in effect" doing my error propagation, because it is just zero, and therefor can just be simplified out of everything.


It kind of sounds like you dont know what "approach" means, have you taken any calculus courses...? Im guessing no, based on the way you just tried to use the word "approach" incorrectly.



Correct and thus...


You have successfully answered your own question.



Bless your heart, lol. So when called out on the lack of it, you first went "Yeah, it totally does it" then backpedaled and have now gone "Oh well, actually it doesnt need it" and then backpedaled harder going "Oh and well, if I actually wanted to really do what I said I did it would actually take more work, which inherently means I have now acknowledged replicating what you did in the sheets is not so easy"

So, thanks for acknowledging what I made isn't "nothing" haha.


I already answered this question in detail, twice now. Please go back and read if you want to understand why. Your tool is not useful for making gear choices. Thats all there is to it. Whether that matters to you is up to you.


If you say so, lol.

Then, as I stated, your tool fails at this task, as if you are getting a 25% variance in DPS just by re-running it twice, that is completely and utterly useless for anyone trying to compare gear choices that are in the range of modifying DPS by 1%

Pretty simple. If a user is comparing a piece of gear that will modify their DPS by 1%, and your program will vary its output wildly by 25% or whatever each time they run it, that just does literally nothing for the user.

Not sure what about that is hard to wrap your head around.

You two should just kiss.

It's true that a limited sample size won't ever get you to the actual average, but that doesn't matter, because you're also equally likely to have a streak of better rolls than average as you are to have a worse streak of rolls than average.

In good groups the personal variance of peoples' TP sets aren't going to make the difference between winning or losing; the difference between victory or loss is going to depend on how many significant active errors are made (WHM glancing away for 5 seconds and letting a DD die, the buffer forgetting soul voice, a DD hitting the wrong mob, etc), not anything to do with the variance in the DPSs' sets.

In bad groups you're just as likely to be in one where you barely lose on average as you are to barely win on average, so having a high variance could open up more chances for victory just as much as having a low variance could shut down chances for loss.

If you think having the exact same TP return is correct, I don't know what else to tell you.

Lack of what? i never said that little bit accounted for accuracy. I just said the formula wasn't hard and doesn't take that long. It does implement 8 hit per round cap, you're just too stupid to see it. Did you take any algebra perhaps? in fact, you can even do this in the spread sheet, since it isn't random. well, people that don't just copy someones work can at least.

Are you an idiot? I don't run it twice, it runs it however many times and for however long I want it to. 10,000 fights at 3 minutes each? no problem. 10,000 fights at 5 minutes each? no problem. 10 minutes? Sure. The variance is between each individual run of those 10,000 runs. Again, it's very apparent you don't know ***about anything. Stop trying to act like you know ***about math or code just because you modify a few cells someone else already made.

And low variance with minimal to no DPS change is ideal.

Not always. A DD that does 1000 DPS on average with 20% variance will never pass a challenge that requires 1250 DPS, but a DD that does 1000 DPS and has a variance of 30% will pass on occasion, and a DD that has a variance of 40% will pass more often.

I still stand by my statement that the true E-peen DPS stat above all is Meva, and no DPS simulation or spreadsheet at the moment does, or I expect ever will, simulate it.

The DPS that resists that paralyga/terror will be the one that wins the parse, always. Its why Rune Fencers often have surprisingly high DPS compared to their DDs in the party, despite the fact they dont typically rock nearly as much aggressive stats.

When everyone else is paralyzed or terrored or has like 3 debuffs on them and the Run is like "Haha Resolution goes BRRRRRRRRRR", it will often soar up the parse meter.

This is why in my opinion, the Malignance set will often outperform anything else on jobs that can use it. Even if it has much worse stats, if its any type of NM that pops off a bunch of enfeebling AoE ***, resisting some of that stuff will do so much more for your DPS than any amount of aggressive stats.

You also may be surprised to discover that in some fights, Indi-Vex may actually result in much much better runs compared to something like Fury.

I will often recommend Frailty+Vex for any fight where AoE *** debuffs are common, over Frailty+Fury.

Seriously, give it a try sometime, you might be surprised at how much it makes fights turn into a cakewalk. Lotta HELM NMs suddenly become a lot less of a pain if you do Vex.

And can fail horribly, since you want to look at both sides. If you're relying on nothing but high rolls to beat a fight, you have other issues.

And Austar: You have demonstrated without a doubt you are not only substantially out of your depth, but you are also completely and utterly unwilling to read or listen to feedback. This concludes our discussion.

Enjoy your mediocrity.

This post is very on-topic in this thread, I'm impressed.

And if you are relying on avoiding bad rolls to not lose a fight, you have the same issues. Both situations should be avoided equally, but on average cancel each other out when in the realm when you cannot avoid them.

In the realm where you can avoid them, then the variance has zero effect on whether you win or lose and thus isn't a problem to be avoided.

I agree that if you're relying on average rolls the whole fight, you have a problem. But I'm not saying you should, I'm saying you reduce the likely hood of getting those. For reasons beyond just winning or losing a fight, it can even be helpful for maintaining a SC, not relying on that extra TA proc to get the 5th or 6th ws off for example.

well when you want to act smart again, copy and paste my one line of code and just change the q,t,d to some cells and you'll see how I got 8 hit cap.

Doing 6-step skillchains is one of the few exceptions I mentioned earlier for minimizing variance being something worthwhile to go for, but that's generally not something that is of interest unless putting a set together for Vagary or Omen farming.

In a good group where your DPS is comfortably over the minimum, variance is neutral because it will never mean the difference between victory or loss, as the difference in DPS variance is very small compared to other sources of variance in FFXI (human error, mob's choice of TP moves, spell resists, etc). With high variance, you might clear Kin a minute slower out of 20 if you are unlucky that run, but you might clear him a minute faster if you're lucky that run, and even maximum reasonable unluckiness isn't going to cause a loss.

In a bad group, variance is also neutral because while a large variance might on occasion cause a loss if your average DPS is just a hairsbreadth above the required amount and you get unlucky rolls, a large variance could also cause a win if your average DPS is barely below the required amount and you get lucky. The range of situations cancel each other out.

I feel like this has stopped mattering as well now that summoner can just 6 step with Astral conduit with basically zero variance.

tldr; soralin is an absolute moron and trying to build his own cred/ego by talking ***about things that are above his head

when that failed, he fell back on personal insults, strawman about m.eva, and shitty meme

excellent thread