Norwest

No no, please continue. I really appreciate when people take the time to go over my work and challenge my assumptions. I want to get this working right, and if no one nitpicks my work I'll never know if I've got it right.


Yes, that's what I'm assuming. You may be right on this that I'm 'misusing the mean' as you say. You don't have the same chance to refresh Hawk on the 7th hit as on the 2nd hit. I need to mull this over.

 

The Iron Colonel

Strictly speaking, you DO have the same chance to hit on the 7th hit as the 2nd hit. The difference you're referring to is "What is the chance that I will get 7 non-procs consecutively" versus "What is the chance the I will get 2 non-procs consecutively". The probability of each independent event is the same, the probability of consecutive events is different.

I haven't hashed it all out yet, but my feelings are that you'd have to ask "what is the probability that I haven't YET had a proc in a series of shots?" and "what is the probability that this shot will be a proc?". The second question is easy to answer (10% proc chance) whereas the first appears simple, but is actually complex when you consider whether or not you're within the duration of a prior proc. I don't have good answers, but just a little insight.

As for the number of shots in a proc, I think you're on the right track. The first proc will lose time equal to the unmodified attack speed before the first auto shot, which you've accounted for. The subsequent procs, if overlapping such that the effects are contiguous, won't lose the first shot.

 

• Glaurong

It is interesting to see you guys thinking out loud, running through some/most of the stuff I did before I decided to just simulate it.

I'm sure there is a way to accurately describe it but I don't have a degree in statistics so it was well beyond my scope.

 

Norwest

Gaul just a question, are your numbers at the start on a per-second basis or are they per shot?

 

The Iron Colonel

Yeah Glau, as I said before I think simulating it is probably the easiest method to do this (which you've already done). It is an interesting academic question, though, so if someone is able to decipher things and create a nice little equation or set of equations, I'll happily support that effort.

Getting back to the original question of "how do we model iAotH?", I want to reiterate my support for determining average up time and then weighting a QS and non-QS rotation accordingly. It won't be perfectly accurate, but it'll be more representative of the potential alterations required to the shot rotation while QS is up.

 

Norwest

Alright, the average number of shots before a reproc isn't that hard. It's a power series and it's not too hard. I'm going to put the numbers in and see what it gives. For a 2.70 it was formerly 4 bonus shots, now it's 3.58.

Edit: Frack I need to redo it for the first proc too! So much for getting any real work done today, The Iron Colonel this is all your fault you realize that?

 

• Glaurong

My simulation at the start was to find a way to determine uptime, given different auto shot speeds.

The flaw in the simulation is that the proc gives you a whole number of shots gained (non continuous), the simulation is continuous. This non-continuous nature of the proc (in relation to weapon speed) is part of what makes it difficult.

 

Breakerone

Hey Cheeky,

just a small thing, the base Agi for a lvl 70 orc is 148 and not 150, and base int is 74, not 78.
And I think you got the dive training point costs wrong. Should be the same as dash.
Also, would it be possible to make a checkbox for deciding if someone wants the modeled IAotH or not? Not that I dont want it, but if I could turn it off, I could recheck the rest with my own calculations to be sure that we have the same base.

Edit: Oh, and if you want to go for perfection, you might want to add the 5% damage to the pet if the owner is an orc

 

Cheeky

1 - Orc base stats fixed!
2 - Dive cost fixed!
3 - The new way I model IAotH (upcoming version) allows for the Hunter to build 2 shot rotations. One in normal speed, and one with quick shots. It then ratios DPS, MPS, and shots per second between the 2 based on IAotH uptime. This should meet your needs. One way to remove Quick Shots in the current released version is to turn Aspect to Viper and then hand-code in +155 RAP on the gear page.
4 - Orc racial to pet damage added!

Thank you for your comments and suggestions.

Cheeky of Havoc

 

Norwest

Alright cheeky,if you are modelling it as 2 rotations, my calculations still work provided there is no clipping. I added a compact version of my calculations that don't take up half a page. I redid my power series calculations so it doesn't rely on modelling anymore, it just gives average Hawk uptime.

I made the corrections to the average number of shots added by a re-proc. I did that by calculating the relative probabilities each shot will re-proc, then assumed that one of the shots DID proc.

I'm fairly confident of my numbers. Compared to Glauron's numbers they are a bit high though. I blame this on the fact that final shots in the rotation extend the bonus of iAotH past 12sec. Without knowing Glauron's calculations I can't really say.

Here's the updated sheet:
http://www.mediafire.com/?8yennqun0dz
any feedback is greatly appreciated.

 

Cheeky

Thanks Norwest. I'll look over these equations and move them to my spreadsheet. By separating out the 2 rotations like I did I can compute autoshots/second for both rotations. This will include clipping, so we get an accurate pricture of how frequently auto shots will be fired, rather than relying on the base weapon speed and static haste.

Cheeky of Havoc

 

Cheeky

Ok, I've looked this over and I have a few areas I need help understanding. In cell B15, what is this equation trying to do, and where do you derive it from? Also I'm at complete loss to understand what C17 is doing too.

In cell B9 - where you are calculating the number of shots in the re-proc of Quick Shots you subtract off the time of a hasted shot, then divide what is left by the hasted speed and round up. Wouldn't you get the same result by just dividing by the hasted speed and taking the floor?

Ex. 12 seconds and 2.5 hasted speed - Original equation = roundup((12 - 2.5)/2.5) vs. new = floor(12/2.5). Both yield a result of 4. Doing the same thing on a boundary condition, a hasted speed of 2 works the same: Original = roundup((12-2)/2) vs. new = floor(12/2). Both yield 6.

I am not saying any of your numbers are wrong. I just don't understand your math enough to know if they are right.

Cheeky of Havoc

 

Vec

I've tried to compare lactose's and cheeky's spreadsheets because I felt if they came up with close to the same results that would be a good thing . I had to add a few things into lactose's spreadsheet, such as blessing of kings buff and the value of +hit rating compared to other stats.

Basically what I found is everything seems to be about the same, except the value of +rating. With my gear and buffs, lactose's sheet showed 1 +hit to be worth about 2.15 ap, while cheeky's showed 1 +hit to be worth almost 3 ap. This is a pretty big difference, especially considering that everything else,(in terms of comparing stat choices), seemed to be almost identical between the 2 sheets. Now since I had to manual add this into lactose's spreadsheet, I could have made a mistake, but I don't think that's the case. I was thinking it could be how each spreadsheet handles the "roll" system differently for crits/hits/misses, etc. I'm definately not an expert on this, so maybe someone else has some insight on it.

 

◊ Lactose

I'm using a 1-roll system.
25% crit --> 25% of total attacks = crits.

A page or two back Cheeky went through and compared some stuff in our spreadsheets, you might find some reasoning for the observed behaviour there?

Look, Lactose, we'd rather you didn't eradicate the whole human race.
- Sam & Max

 

Vec

Somehow I missed some of those posts. I think it's pretty clear to me now.

 

Norwest

The difficulty of modeling iAotH is that when you chain procs the last one gets overwritten. The question becomes: how many shots did the first proc get off before it was overwritten? That's what B15/B16 is. I previously had that number as half the length of the 1st proc, since I assumed the 2nd would on average proc halfway through the 1st. But in reality, it procs slightly earlier. The calculations are on the 'Average Shots' Sheet.

C17 is the average number of Hastened shots from the first proc, until the last hastened shot. Basically that's how long a chain of procs lasts on average. That number is really the key to the whole calculation.


Yes, you'd get the same result. My formulae are often a bit backwards.

 

Skogix

I just downloaded your spreadsheet (Cheeky) and i find it really interesting, but i get some weird error at some weird places.

Ranged Weapon Steelhawk Crossbow Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 0 0 0

Waist Ebon Netherscale Belt 418 0 0 18 13 0 44 0 20 0 7 0 0
Enchant None Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 Err:504 0 0



Tried all combinations with gear i could think of and all different bow / waists. Downloaded the sheet several times and looked for earlier versions without success. Working as intended or is it just a "extra function"?

 

Cheeky

I develop this tool in Microsoft Excel 2000, and have verified it works in 2003 as well. Anything else is "Use at your own risk."

Cheeky of Havoc

 

Cheeky

Version 14 is released.

Major (and minor) features:

• Finishing shot used to cycle the timing back to the initial shot in all shot rotations.
• Major Improved Aspect of the Hawk changes. I now allow Hunters to specify rotations for both in and out of Quick Shots and ratio the DPS/MPS/shot-per-second between them based on the IAotH uptime.
• Weapon buffs (mana oils)
• Pet Avoidance ability can be trained (not modeled in any way though)
• 2.1 PTR gear stats - Courtesy of Vaxum from www.TKASomething.com
• New automatic shot rotation to generate 2 specials/auto
• Resilience and armor penetration added. (Vaxum did leg work on tracking down stats)
• Moved methodology to 1-roll system, so your crit percentage is an absolute of all fired shots. (No more "crit a miss" effect)

Cheeky of Havoc

 

Groggan

Cheeky, do you have any plans to implement Monster Slaying/Humanoid Slaying? I realize it would be a pain with regard to having to either note the type of mob the dps is for (and likely would be most intuitive if a total was added for targets under the effects of slaying vs. not), but it seems as though the slaying talents should have a not-insignificant effect on DPS (particularly for a full SV hunter as they approach higher and higher crit rates).

 

Cheeky

Yeah, I can do that pretty easily. I'll have it in the next version, which should be early next week. I figure that gives people plenty of time to point out all the mistakes in the latest (15 - fixed a few bugs in 14) version up.

What I'll do is provide a drop down menu for target type. If it matches one of the types you have slaying points in I'll increase damage accordingly.

Cheeky of Havoc

 

Utters

I don't know if this has been posted but on the buffs section turning on TSA doesn't seem to add any AP on Cheeky's spreadsheet.

 

Fearspect

I believe it is assumed to be calculated (TSA) if you already have the talent, the buff section will only add it in to calculate if you are not already talented in that way.

Now, regarding EW, I am having a little trouble with the current model:

1 - (Chance of no proc)^(attempts)

Now, this doesn't really produce a true uptime, at least I don't think so. What it shows is the chance that it will fire off within a period of 7 seconds, but does not really factor in refreshes and such.

I can't figure it out myself I don't think. Is this a combination of Discrete Math with a Rate problem?

The reason is that with points tight in SV, I am wondering if I can get away with 2/3 Expose weakness instead of 3/3 (It seems like greatly diminished returns per point, but I want the raid to benefit from the most possible uptime).