about Damage per second

_Skai_ - Raging Tide wrote: »

Because as the crit rate increases, the crit numbers in a large sample increase, thus making the overall DPS higher.

For example, let's use small static numbers and solid crit numbers, with a solid APS of 5 attacks per second.

For this example, (remember I'm leaving a lot of variables out, just to make my point here clear), the min and max damage will be 200.

The crit rate in the first sample will be 20%, so 1 in 5 hits (again, statistically speaking) will be a crit. This sample will have 20 hits.

bennymars wrote: »

Is chance works that way?. with 20% chance of doing critical damage in 5 hits. in every 5 hits you deal 1 critical damage?. is that how chance works?. works in a fixed manner?

in 5 hits with 20% chance of doing critical damage, its even possible you don't land even 1 crit in 5 hits. or you may not even land 1 crit in 10 hits...

this law of large numbers that i previously read, it applies only in a larger scale. but reality speaking, do you do hits in a larger scale here?.

Do you not know what the bolded text means?

Edit: And in argument, in 5 hits with 20% chance of doing critical damage, it's possible that ALL 5 hits are crits. It's possible that all of the next 10 hits are crits. That's why I said it's better calculated on a large scale to find the average DPS, which is what we were aiming from the start.

This is the same with things that aren't statistically labeled, such as the proc rate (% chance) of things such as Sacrificial Strike and God of Frenzy. If you do a sample of 10 hits and GoF procs 0 out of 10 hits, then would 0% be a good number to put the proc rate for GoF at? No.

I was just making it statistically accurate for him to see. I know my example was flat out crappy, but he wanted to know why increasing a crit rate would increase average DPS, and that's the first thing that came to my mind to try to explain it.

And it makes me wonder... why do you even care about "100% accurate" information in regard to DPS? Even if we're not talking about the % crit rate, look at the min and max damages.

You're not going to ALWAYS hit the same number when the min and max damages are not the same. That is also "a form of chance".

DPS = ((MinAtk + MaxAtk) / 2) * (CritFactor * RageIncrease) * APS * (1 + ATKLV/100))

This completely ignores any possibility of the enemy being able to evade attacks.

bennymars wrote: »

If we follow that formula. the crit% in that formula is no longer a "chance". it works 100% at the given percentage.... and it took the whole nature of crit% which is "chance".....

i hope maybe in some ways. you consider chance in that formula. that way.. it will have a good accurate results...

Yes then you can have non-average results b:chuckle

But to calculate average DPS you have to assume that 20% crit is a 1 critical strike out of 5 hits or youll be doing simple theorycrafting instead of calculating average dps.

_Skai_ - Raging Tide wrote: »

@Benny Through research through the years, that formula has been accepted as the way to calculate average DPS. Sure crit hits are always a chance, but over any sample of hits, with any % crit rate, you can either: crit on all of the hits, not crit on any of the hits, crit higher than the statistical average, crit lower than the statistical average, or crit on the statistical average.

It's random yes, but that's the only way to calculate average DPS. If you take that out of the formula, you'll be getting a false average DPS that is in fact your DPS with 0% crit. If you take out your maximum damage as well as the crit, then the resulting calculation would be your minimum DPS. But it won't be the same the other way around if you take out your minimum DPS and crit from the formula.

Well, the average only really serves as a tool for comparison. The reason for this is that there are weapons which have ridiculously high maximum damage, e.g. R8, and weapons which have a very small damage range, e.g. Barrier Thorn: Nirvana, so you really cannot compare them fairly with either minimum or maximum.

For practical purposes, I would look at both the minimum and the average. The maximum amount you can deal is pretty much irrelevant outside of PvP and considering that most assassins are gung-ho for attack speed, the average represents better.

The thing is, no one is claiming that criticals aren't a random thing. The formula just gives you an expected value, which at best serves as a means of comparing different weaponry and making observations of damages, e.g. whether they were substantially higher than the average.

Also a point worth noting is Power Dash. Power Dash gives you 40% Critical, which will always put you over 50%, meaning that you are more likely to crit a hit than not.

Starlicious - Lost City wrote: »

@Benny...Let me word this in a way you can understand.

Yes you are correct that crit% as well as other factors are uncertain and chance values. But the issue is that you need to account for these uncertain values to have any accuracy at all.

On a small sample size assuming 20 total attacks it could be very inaccurate. But as your sample size increases to 10000 100000 or 1000000 the law of averages makes those uncertain variables work like they were a fixed variable.

Ouch ouch ouch! Terminology! Law of Averages is bunch of bullcrap, what you're looking for is law of large numbers.

_Skai_ - Raging Tide wrote: »

Exactly. Average DPS calculations are there mainly for comparing different weapons' average DPS while accounting for every attribute an Assassin has. It's a very useful tool for rising Assassins or math lovers to compare theoretical average DPS's of high-costing weapons so that they can determine what goal to set. That's why I'm wondering why benny is so nit-picky about how average DPS is calculated.

If anything instead of arguing about how "wrong" the average DPS formula is, benny should go in game and do a DPS calculation in real time by recording all of his damages with every variable recorded, and calculate the average DPS over the course of an inhuman sample.

From there derive his own DPS formula with said numbers.

Sorry for the small rant. b:surrender

The problem is simply that DPS is an average. Averages are a terrible tool to compare just about anything because they flatten all results and erase the differences. An average is only the final tool you use too put a hierarchie between the last remaining (and usually tight result) solutions you have.

However, the pwi community tend to use this average as 1st and main criterium. I'm glad finally some others came to notice how increadibly dumb this is.

It's simple to adjust the "formula" really. Just have to change the approach, rearrange a bit and, for ***** sake, keep intermediate results.

For instance, personally, I would opt for a probabilistic approach since I like spike damage :
- calculate your minimum and maximum base hits (min and max damage from calc then factor in att lvls and slaying lvls).
- create a few significant segments, like 4 as starters (you would have a "low end damage", 2 "average hits" and 1 "high end damage" segments), and take an average of them. The more segments you make, the less you flatten your results.
- take your probability for normal hit, to crit, to zerk and to zerk crit. Apply them on each segment average.
- apply your aps to each probabilty corrected segment average and addition the collumn. But also apply your aps on each average before probability correction.
- now you have a simple crossheet that allows you to see your dps, but also it's dispersion, your minimum, your potential and even factored in your dph in every way.

Idk, for me it looks so simple, but it seems many don't understand how toxic averages are for ones analysis.

Empu - Sanctuary wrote: »

The problem is simply that DPS is an average. Averages are a terrible tool to compare just about anything because they flatten all results and erase the differences. An average is only the final tool you use too put a hierarchie between the last remaining (and usually tight result) solutions you have.

However, the pwi community tend to use this average as 1st and main criterium. I'm glad finally some others came to notice how increadibly dumb this is.

It's simple to adjust the "formula" really. Just have to change the approach, rearrange a bit and, for ***** sake, keep intermediate results.

For instance, personally, I would opt for a probabilistic approach since I like spike damage :
- calculate your minimum and maximum base hits (min and max damage from calc then factor in att lvls and slaying lvls).
- create a few significant segments, like 4 as starters (you would have a "low end damage", 2 "average hits" and 1 "high end damage" segments), and take an average of them. The more segments you make, the less you flatten your results.
- take your probability for normal hit, to crit, to zerk and to zerk crit. Apply them on each segment average.
- apply your aps to each probabilty corrected segment average and addition the collumn. But also apply your aps on each average before probability correction.
- now you have a simple crossheet that allows you to see your dps, but also it's dispersion, your minimum, your potential and even factored in your dph in every way.

Idk, for me it looks so simple, but it seems many don't understand how toxic averages are for ones analysis.

Considering that about 95% of the population of PWI fails at middle school level math (since that's what counting average DPS really amounts to), I don't think that most of them would have the skills to do that either, which isn't essentially any different.

And besides, even for people who could do that, such as myself, it's just too much effort to bother with.

Why in the world are there 6 pages on the topic of the definition of DPS?

Elenacostel - Heavens Tear wrote: »

Why in the world are there 6 pages on the topic of the definition of DPS?

Because some newbie is irritated that the average DPS formula isn't 100% accurate for all given situations and is essentially the average. Yeah, it's a tautology.

Quilue - Sanctuary wrote: »

When you hit at 5 times per second, you pretty much just take average DPS and run with it.

Well, that was kinda my point: high APS results in massive samples, thus the average is a pretty good representation.

Empu - Sanctuary wrote: »

But the number of hits needed to kill a specific boss is still the same...

The question would be : why even calculate dps, as it stands for nothing...

Not necessarily true.

For example, I got roughly identical DPS with a +6 Ocean Supreme Dagger and a +6 Barrier Thorn: Nirvana. The former has 2.5 APS, whereas the latter has 3.33. Obviously, they'll kill the boss in an identical time, but the amount of hits will be greatly different.

And it's perfectly possible to increase your APS while decreasing your DPS, such as using Deicide on an Assassin. Or going from a highly refined Rank 8 to a unrefined Barrier Thorn: Nirvana.

Olbaze - Sanctuary wrote: »

Ouch ouch ouch! Terminology! Law of Averages is bunch of bullcrap, what you're looking for is law of large numbers.

Sorry same basic idea just picked the one with the flawed basic principle...b:surrender

from wat i've been reading to what i've been doing in PWI, Talk all u want but take it from a sage sin's point of view, 3 sparks, sage subsea+ sage powerdash + TM= 9 seconds 96% crit FTW, 5aps Yan the traitor gone in 5 seconds. :O, so wat now ;P