Hunter Spreadsheets -- In development - Page 159

SenorPez · 08/04/08, 10:55 AM

Thank you very much Cheeky, Lactose, and the rest of the Hunter community here. Tools like this spreadsheet helped me really understand the "nuts and bolts" of what separates the average Hunter from those that excel. (No, I didn't intend a pun.) Your hard work on the spreadsheet was certainly a factor in improving my game, and I thank you for it!

Faerdael · 08/04/08, 11:13 AM

Thank you so very much for all your hard work Cheeky. Your contributions to hunter theorycraft have been eclipsed by no other, and we all are in your pixilated debt.

Regarding the spreadsheet, shot rotations and haste are a tad nebulous and people should understand the behavior.

Shot rotations should be taken slightly with a grain of salt; they are "artificially" modled and not 100% representative of in-game play. For example, if you equip a chicken bow and add 556 haste, you will end up with a 1.5second auto shot. You can then select a 3:2 rotation and it will populate auto, steady, steady, auto, steady, even though this would be an impossible result given any accepted hunter macro. Thus it is wise to understand your in-game behavior when working out shot rotations, although it is a good start.

Looking at quickshots tab, you will generally see that by about the sixth shot, your distance between steadies is pretty consistently 1.5 seconds apart, I am not altogether clear why the initial few shots differ in this regard, but as far as overvaluing haste, perhaps slightly, but less than you might suppose. You really start to see diminishing returns based on haste; there really is a "soft" cap to this on paper.

The slight overvaluation of haste regarding GCD is possibly counterweighted by the affect of latency as calculated on the spreadsheet, since using a /cast /cast rotation in a continual spam manner could possibly be incurring the detriments of latency at the beginning of the spam cycle and not in between each shot, as the spreadsheet indicates. (I will have look for the post(s) to link here sorry).

Also, when dealing with mulitple haste buffs, it is a roll of the dice as to how much of these buffs will actually overlap during play, and result in non-maximized use of haste buffs. The spreadsheet plays the odds here and comes up with some averages for chance for duplicity, but again, it can never be a literal translation of the game results. I personally use [The Blade of Harbingers] and a [Dragonspine Trophy] and I have never felt I was misled by the spreadsheet, and I am glad I made these choices, which I otherwise would not have made.

So in short, the spreadsheet has been a brilliant guide, but its an aid to the gameplay and not "how-to" guide. To really understand the working of any armor set, you just need to get the stuff hot.

Sebudai · 08/04/08, 12:00 PM

I appreciate the post, Faerdael, and I'm well aware of what you're saying. However, I think you're looking past what I'm claiming here, probably because you're used to dealing with posters that seem to possess the intelligence of a tree stump. I don't blame you. Anyway, the spreadsheet doesn't "slightly misjudge" the GCD of Steady Shot early in the rotation and rectify the problem later on. It simply does not account for the GCD of Steady Shot at any point in the rotation. I understand that certain game play aspects are impossible to model with perfect accuracy, but this is just an error. An error that is affecting the value of haste in a substantial way. I know, because I used math. The spreadsheet is overvaluing haste rating by about 26% for me.

Anyway, I've done my part in alerting others of this. I wasn't really expecting attention to detail to be an attribute of the hunter community in general.

Dralmoo · 08/04/08, 12:11 PM

edited because i can't count, and I did some more testing.

Actually, no matter how much haste I set it's not like you ever start chaining 0.5 second steadies. Even with 2000 haste, I see the second steady coming 1.2s after the first, which is clearly in error, but then an average of 1.7s between the rest, when using the 1:1 button. The extra 0.2s is from the latency calculation. The actual cast time of each steady is 0.15 seconds at this haste value.

Sebudai, do you see something different?

Action	Start	Time Since Last
		
Steady	0.50	
Steady	1.70	1.20
Steady	3.40	1.70
Steady	5.10	1.70
Steady	6.80	1.70
Steady	8.50	1.70
Steady	10.20	1.70
Steady	11.90	1.70
Steady	13.60	1.70
Steady	15.30	1.70
Steady	17.00	1.70
Steady	18.70	1.70

Cheeky · 08/04/08, 5:15 PM

Originally Posted by Sebudai

I appreciate the post, Faerdael, and I'm well aware of what you're saying. However, I think you're looking past what I'm claiming here, probably because you're used to dealing with posters that seem to possess the intelligence of a tree stump. I don't blame you. Anyway, the spreadsheet doesn't "slightly misjudge" the GCD of Steady Shot early in the rotation and rectify the problem later on. It simply does not account for the GCD of Steady Shot at any point in the rotation. I understand that certain game play aspects are impossible to model with perfect accuracy, but this is just an error. An error that is affecting the value of haste in a substantial way. I know, because I used math. The spreadsheet is overvaluing haste rating by about 26% for me.

Anyway, I've done my part in alerting others of this. I wasn't really expecting attention to detail to be an attribute of the hunter community in general.

There are separate GCD calculations being performed in the Shot Rotations, independent of the Steady Shot cool down (which is allowed to go below 1.5s) As was shown above, you can get the first Steady less than 1.5s after you start (since I assume your GCD is available at the start of the cycle) but you will not be able to increase their frequency beyond 1.5 + latency.

This does provide an effective cap on haste. You'll see some small variance in DPS (less than 0.5 DPS) with massive haste numbers just because of the way I implemented the Shot Rotations to actually start and end at times that may not coincide with the GCD. But that's noise of even less effect than rounding errors would be.

In short, haste is modeled as completely as my limited intellect allowed me to do it. Hopefully someone else can come up with a better scheme, but the problem seems much easier when it's not you having to account for everything.

I'd like to see your math that proves I'm off by 26%, but honestly it sounds like the basis for a new spreadsheet to me. Good luck with that!

I'd also like to let everyone know that I am touched by the comments left. It was a lot of fun (when I wasn't busting my head against my monitor over trying to figure out how to model interfering shots or the fucking [Talon of Al'ar] - worst trinket ever) working on this project. The list of names of everyone who contributed to the final product is extensive, and reads very similar to the list of everyone who's ever posted to the thread or the one at TKA.

Sebudai · 08/05/08, 6:37 AM

Sorry, Cheeky. I was being a little snarky, but it wasn't aimed at you at all. I was really just looking for a quick "you're wrong and this is why..." or a "you're right" and not a "well, spreadsheets and simulators are naturally somewhat inaccurate because it is impossible to perfectly model real game conditions" etc. I'm well aware of the limitations of a spreadsheets. Anyway, this is a copy of my current 1:1.4 rotation:

#	Action	Damage	        Mana    Start	Stop
1	Auto	1,584.71	0	0.00	0.50
2	Steady	1,745.54	102	0.50	1.80
3	Multi	1,859.01	255	1.80	2.30
4	Auto	1,584.71	0	2.43	2.93
5	Steady	1,745.54	102	3.30	4.61
6	Auto	1,584.71	0	4.87	5.37
7	Steady	1,745.54	102	5.37	6.67
8	Auto	1,584.71	0	7.30	7.80
9	Steady	1,745.54	102	7.80	9.11
10	Arcane	1,613.85	213	9.11	9.11
11	Auto	1,584.71	0	9.74	10.24
12	Steady	1,745.54	102	10.61	11.91
13	Auto	1,584.71	0	12.17	12.67
14	Steady	1,745.54	102	12.67	13.98
15	Auto	1,584.71	0	14.61	15.11
16	Steady	1,745.54	102	15.11	16.41
17	Multi	1,859.01	255	16.41	16.91
18	Auto	1,584.71	0	17.04	17.54
19	Steady	1,745.54	102	17.91	19.22
20	Auto	1,584.71	0	19.48	19.98
21	Steady	1,745.54	102	19.98	21.28
22	Auto	1,584.71	0	21.91	22.41
23	Steady	1,745.54	102	22.41	23.72
24	Arcane	1,613.85	213	23.72	23.72
25	Auto	1,584.71	0	24.35	24.85
26	Steady	1,745.54	102	25.22	26.52
27	Auto	1,584.71	0	26.78	27.28
28	Steady	1,745.54	102	27.28	28.59
29	Auto	1,584.71	0	29.22	29.72
30	Steady	1,745.54	102	29.72	31.02
31	Multi	1,859.01	255	31.02	31.52
32	Auto	1,584.71	0	31.65	32.15
33	Steady	1,745.54	102	32.52	33.83
34	Auto	1,584.71	0	34.09	34.59
35	Steady	1,745.54	102	34.59	35.89
36	Auto	1,584.71	0	36.52	37.02
37	Steady	1,745.54	102	37.02	38.33
38	Arcane	1,613.85	213	38.33	38.33
39	Auto	1,584.71	0	38.96	39.46
40	Steady	1,745.54	102	39.83	41.13
41	Auto	1,584.71	0	41.39	41.89
42	Steady	1,745.54	102	41.89	43.20
43	Auto	1,584.71	0	43.83	44.33
44	Steady	1,745.54	102	44.33	45.63
45	Multi	1,859.01	255	45.63	46.13
46	Auto	1,584.71	0	46.26	46.76
47	Steady	1,745.54	102	47.13	48.43
48	Auto	1,584.71	0	48.70	49.20
49	Steady	1,745.54	102	49.20	50.50
50	Auto	1,584.71	0	51.13	51.63

As far as I can tell, the spreadsheet thinks I can start things like Arcane Shot or Multi-shot 1.3 seconds after I cast Steady Shot, and not 1.5. If I had more haste it would be even less accurate. Anyway, am I interpreting all of this incorrectly? I'm not looking for anyone to fix the spreadsheet, I just want confirmation as to whether or not this error exists so I can tell if I need to account for it. I appreciate the assistance so far, guys.

Cheeky · 08/05/08, 10:25 AM

Originally Posted by Sebudai

Sorry, Cheeky. I was being a little snarky, but it wasn't aimed at you at all. I was really just looking for a quick "you're wrong and this is why..." or a "you're right" and not a "well, spreadsheets and simulators are naturally somewhat inaccurate because it is impossible to perfectly model real game conditions" etc. I'm well aware of the limitations of a spreadsheets. Anyway, this is a copy of my current 1:1.4 rotation:

As far as I can tell, the spreadsheet thinks I can start things like Arcane Shot or Multi-shot 1.3 seconds after I cast Steady Shot, and not 1.5. If I had more haste it would be even less accurate. Anyway, am I interpreting all of this incorrectly? I'm not looking for anyone to fix the spreadsheet, I just want confirmation as to whether or not this error exists so I can tell if I need to account for it. I appreciate the assistance so far, guys.

Could it be that you have the non-castsequence option selected? In that case the spreadsheet will recover GCD during the cast of Auto Shot, making it seem like you can chain 2 specials closer than 1.5s. This mimics what you can get in game, but the GCD should eventually catch up to you again and start limiting total special shot throughput to 1/1.5s.

The last row of the matrix is the GCD, take a look at what's going on there. See if what is happening algebraically makes sense to you. This was all part of my effort to model the client-side GCD changes (and the whole reason the /cast /cast macro seems to work.) I could have screwed something up, since I very rarely use other than a 1:1 rotation myself.

You may be wrong, you may be right. I dunno. There's some tricky math in there in regards to the GCD, and I'm perfectly capable of having fucked it all up.

DOOA · 08/05/08, 3:08 PM

Cheeky,
I lament this being the last issue of your spreadsheet.
Any chance you show how to add items so we can have drop downs for new stuff?
Currently I use Custom, but having a drop down would be much faster for trying combinations.

Cheeky · 08/05/08, 3:12 PM

Originally Posted by DOOA

Cheeky,
I lament this being the last issue of your spreadsheet.
Any chance you show how to add items so we can have drop downs for new stuff?
Currently I use Custom, but having a drop down would be much faster for trying combinations.

Unhide the item sheets, insert new rows, and fill in the blanks.

brigwyn · 08/05/08, 3:32 PM

Cheeky and Lactose,
You know how all of us over at The Hunting Lodge much less the entire hunter community appreciate all of the hardwork the both of you have done for us hunters. So a simple thank you is not enough but is really all we can offer.

Since I'm not sure how long Elitist Jerks or either of you will continue hosting the spreadsheet or this forum. I just wanted to let you know that I'm willing to host it, keeping this original version preserved and also am willing to offer anyone that's willing to pick up the torch - a forum and hosting site to keep this project alive. Of course this is only if this is acceptable to the both of you.

- Brig
The Hunting Lodge Forums

beathoven · 08/06/08, 2:56 AM

A big thanks from myself to Lactose and Cheeky as well. Your work really made a huge difference, be it on the spreadsheet or detailed investigation of hunter mechanics. So thank you both for your efforts and have fun in whatever you are going to do in the future.

Sebudai · 08/06/08, 5:11 AM

Originally Posted by Cheeky

Could it be that you have the non-castsequence option selected? In that case the spreadsheet will recover GCD during the cast of Auto Shot, making it seem like you can chain 2 specials closer than 1.5s. This mimics what you can get in game, but the GCD should eventually catch up to you again and start limiting total special shot throughput to 1/1.5s.

The last row of the matrix is the GCD, take a look at what's going on there. See if what is happening algebraically makes sense to you. This was all part of my effort to model the client-side GCD changes (and the whole reason the /cast /cast macro seems to work.) I could have screwed something up, since I very rarely use other than a 1:1 rotation myself.

You may be wrong, you may be right. I dunno. There's some tricky math in there in regards to the GCD, and I'm perfectly capable of having fucked it all up.

Yep, that's exactly what it was. I had Hand-Weaved selected because that's what I have been doing in-game. Switching that to Cast Sequence seems to put everything in order. Thanks for the help, and for the spreadsheet in general. I appreciate it.

-EDIT-
I guess I really have no idea why the Castsequence option accounts for the GCD and the other two options don't. I don't get it. I mean, if I hand-weave my shots in-game I get to ignore the GCD? I don't understand. What behavior is this attempting to mimic?

Bikiniwax · 08/06/08, 10:29 AM

Thanks to Cheeky and Lactose for all your hard work. This spreadsheet has been THE reference tool for informed hunters everywhere.

Nophshlah · 08/06/08, 12:19 PM

Originally Posted by Sebudai

I guess I really have no idea why the Castsequence option accounts for the GCD and the other two options don't. I don't get it. I mean, if I hand-weave my shots in-game I get to ignore the GCD? I don't understand. What behavior is this attempting to mimic?

In one of the patches, 2.2 I believe, they changed how the GCD worked. The client now allows you to start specials early, even if the server side does not think the full 1.5 seconds have elapsed. This ONLY works with hand-weaving. A /castsequence macro will always wait for the full 1.5 seconds before attempting to fire the next shot.

Tsook · 08/06/08, 4:09 PM

Originally Posted by Sebudai

I guess I really have no idea why the Castsequence option accounts for the GCD and the other two options don't. I don't get it. I mean, if I hand-weave my shots in-game I get to ignore the GCD? I don't understand. What behavior is this attempting to mimic?

Since 2.3 special shots started after autoshot's channel time starts but before it finishes are delayed until after the auto - but their GCD is not. This means, assuming the .5 second auto cast time we usually do, in theory you can get up to a .5 second 'head start' on the GCD from the steady that immediately follows an auto, assuming the GCD is up before the auto. It's one of the reasons Cheeky theorized the 3:2 rotation 'works' -- if you look at the space between two autos where you get two steady shots, you're essentially offsetting the GCD for the first one back into time you couldn't have been doing anything anyway (auto cast time), which allows you to start the second steady shot before you should have been able to.

Using .00001 as an arbitrary unit for 'very short':

time action
0.00000 auto cast time starts
0.00001 steady1 GCD starts
0.50000 auto cast time finishes, steady cast time starts
1.50001 steady1 GCD finishes
1.50002 steady2 GCD and cast time starts (because of the 'spell queueing' that was put in for all casters in 2.2 or whatever, but we couldn't really take advantage of until 2.3)
1.60001 steady1 cast time finishes (assuming exactly 1.1 second steady, obviously this part will vary slightly)
2.60002 steady2 cast time finishes
2.60003 auto cast time starts
3.00002 steady2 GCD finishes (if I was continuing this you'd start a new steady cast time/GCD here, gaining .1 on the next GCD)
3.10003 auto cast time finishes

It's.. pretty much bullshit, and why I've avoided using the 3:2 macro. As Cheeky said when it first came out: "this is making me want to quit hunter theorycrafting."