Hunter Spreadsheets -- In development - Page 8

Lactose · 05/02/07, 12:07 PM

I'll edit my initial post to include links to other spreadsheets listed in this thread.
It'll be a bit later today though, lots of stuff on my mind.

matthra · 05/02/07, 12:40 PM

You guys rock, thanks for tkaing the time to put these together.

The Iron Colonel · 05/02/07, 1:20 PM

One feature that might be nice would be one to account for stacking on hunter's mark. As you're already figuring out the timing on the first 50 (if I recall correctly) shots of the rotation, it wouldn't be difficult to account for hunter's mark at each shot interval. I include this only for completeness - I'm not sure it would really make that large a difference.

Edit: sorry, missed that you had released the new version a few posts back. I guess this could go in the next release then.

Ernesti · 05/02/07, 2:09 PM

Great spreadsheet Cheeky.

There is one flaw in the new expose weakness uptime calculation. If I put 0 points in it, it gives 100% of the 25% agi bonus. And as you put more points to it, the proc bonus gets smaller. I.e. 229 AP with 0 points in EW and 200 AP with 3 points.

Cheeky · 05/02/07, 2:29 PM

Originally Posted by Ernesti

Great spreadsheet Cheeky.

There is one flaw in the new expose weakness uptime calculation. If I put 0 points in it, it gives 100% of the 25% agi bonus. And as you put more points to it, the proc bonus gets smaller. I.e. 229 AP with 0 points in EW and 200 AP with 3 points.

I must not have saved the fix for that when I did it....

Re-released with that fixed again. My original equation tried to do too much. I have broken it down now into 1/4 Agility, EW up time, and effective benefit.

Sorry about the confusion, I need better source control on my end!

Keltan · 05/02/07, 3:44 PM

Cheeky, have you worked 2.10 Kill Command into your spreadsheet yet?

I just took a look at your latest spreadsheet (12:21, May 2nd) and crit rating is coming up as a smaller dps increase than it was in the "version 11" of your spreadsheet. I was expecting a solid increase in the dps value of crit rating with Kill Command being removed from the GCD and now actually worth using for max dps. Insted, it's actually showing up as weaker.

Rius · 05/02/07, 4:43 PM

Cheeky:

Looks like you've got Flask of Mighty Restoration giving 25 Mana, instead of 25 MP5. Just a head's up =) Thanks for your work on this so far.

Cheeky · 05/02/07, 6:05 PM

Originally Posted by Keltan

Cheeky, have you worked 2.10 Kill Command into your spreadsheet yet?

I just took a look at your latest spreadsheet (12:21, May 2nd) and crit rating is coming up as a smaller dps increase than it was in the "version 11" of your spreadsheet. I was expecting a solid increase in the dps value of crit rating with Kill Command being removed from the GCD and now actually worth using for max dps. Insted, it's actually showing up as weaker.

If you'd like to send me a pm with specific talents/gear/gems/enchants I can look at it and maybe see why the numbers might be different. I have a copy of version 11 still, so I can compare everything.

Lactose · 05/02/07, 7:39 PM

Check first post =)
Anything else someone would want me to edit in?

Norwest · 05/03/07, 9:48 AM

Very nice spreadsheet.

I'd really like to improve the iAotH calculations, I'm too German to accept the approximations. I'm going to try and math it from an individual shot standpoint. I want to improve the calculations, no offence is intended to those who did the current calculations.

But first I want to clarify the mechanics of iAotH so we are all on the same page.

- iAoth has a 10% chance to proc on auto shot
- It procs the instant the shot leaves the bow, so travel time is not an issue.
- The first shot (the shot it procs off) does NOT receive the haste bonus (I tested this awhile ago, but not 100% sure)
- If the proc expires directly after an auto, the auto will still receive the haste bonus, the next auto will not be hastened.

Lactose · 05/03/07, 9:56 AM

I reccommend you read Glaurong's posts on the first page.

Norwest · 05/03/07, 10:12 AM

I've read it, and I have a problem with it. The effect of iAotH with bow speed is not continuous. For example, with a 2.80 and Serpent's swiftness you get 7 fast shots inside a proc. With a 2.70 you get 8 fast shots. That makes the difference between 2.70 -> 2.80 is much larger than 2.60 -> 2.70 and the current formula doesn't capture that. I think I can improve on Glaurong's calculations.

The only thing I'm unsure on is this:
- The first shot (the shot it procs off) does NOT receive the haste bonus (I tested this awhile ago, but not 100% sure)

But I'll make up a spreadsheet that toggles that.

Lurchington · 05/03/07, 10:25 AM

As of now, I'm leaving my item list spreadsheet, coinciding with letting my account lapse. It's not terribly complicated, so if anyone out there enjoyed it and would be interested in keeping it updated, make sure to post it here

It was a good way to procrastinate.

Cheeky · 05/03/07, 10:26 AM

Originally Posted by Norwest

I've read it, and I have a problem with it. The effect of iAotH with bow speed is not continuous. For example, with a 2.80 and Serpent's swiftness you get 7 fast shots inside a proc. With a 2.70 you get 8 fast shots. That makes the difference between 2.70 -> 2.80 is much larger than 2.60 -> 2.70 and the current formula doesn't capture that. I think I can improve on Glaurong's calculations.

The only thing I'm unsure on is this:
- The first shot (the shot it procs off) does NOT receive the haste bonus (I tested this awhile ago, but not 100% sure)

But I'll make up a spreadsheet that toggles that.

I'm not really happy with what we have for Improved Aspect of the Hawk either, but we need something that can average out the effects over time. Glaurong simulated millions of battles to derive his equations, I trust the results based on his methodology. I do agree that modeling an all-or-nothing effect with a average is not ideal.

I hope you do come up with a better way to model it. If you do I'd love to incorporate it into my spreadsheet. Please post your model and thought process here. There are a lot of smart people (Lactose, Glau, etc.) who might be able to help you out.

The Iron Colonel · 05/03/07, 10:41 AM

Originally Posted by Cheeky

I'm not really happy with what we have for Improved Aspect of the Hawk either, but we need something that can average out the effects over time. Glaurong simulated millions of battles to derive his equations, I trust the results based on his methodology. I do agree that modeling an all-or-nothing effect with a average is not ideal.

I hope you do come up with a better way to model it. If you do I'd love to incorporate it into my spreadsheet. Please post your model and thought process here. There are a lot of smart people (Lactose, Glau, etc.) who might be able to help you out.

Time averaging is a valid method of modeling a probabilistic effect over a very large sample size. Unfortunately, I think the issue here is that given the mechanics that hunters have (i.e. shot rotations that do not proportionately scale down with haste due to GCD/clipping problems) you will see increasing divergence between theory and practice as speed scales down. It seems to me that the best method of modeling would be discrete modeling (which is essentially what Glau did in his simulations). Unfortunately, discrete simulation can't be done easily by a spreadsheet. Unless someone wants to write simulation software (which, to me, would be the most accurate way to do things, but also the least convenient to revise) that takes information from a character sheet and then simulates an arbitrarily large number of combat sequences to determine average dps, we pretty much are going to have to live with approximate models based on continuous functions.

Perhaps this belongs in the hunter mechanics thread, but I'll share it here [admin move it if deemed necessary]; one thought that has been rolling around inside my skull for a while is modeling the shot rotation as overlapping waveforms. I haven't fleshed out the idea completely (or, in my opinion, adequately enough to warrant sharing yet, but nonetheless...), but depending on your first shot you can model autoshot as either a sine or cosine function of attack speed (For example, values of 1 would correspond to autoshot fires whereas values of -1 would be 90 degrees out of phase with auto shot - the ideal time to fire a special shot). Likewise, any shot in your rotation can be modeled using sine or cosine (again, depending on which shot you lead with; ideally values of 1 would correspond to a value of -1 on the autoshot waveform). The real bugaboo seems, to me, to be determining how to best use such a model. Perhaps someone can share thoughts regarding this (or am I way out in left field on this one)?