I'm unsure what you mean by "switching out" mastery for haste via custom values. I'm also not sure why that's useful. it's simpler to use the reforging tool to figure out what happens if you switch out stats, and that consistently shows dps gains on various combinations of actual gear pieces (which is how people are actually going to use reforging anyway). Custom values seem to be more helpful for answering questions like, "what happens if I got another 500 mastery?"

It's also worth noting that FD does account for plateaus, or else the values wouldn't vary significantly as you approach areas that shift shots and the like. It also seems that you're overestimating how much of FD's projected gains come from predicted rotation changes. Most of the projected gains come from straightforward sources like autoattack/shot damage, increased wild hunt uptime (additional GftT procs + base regen), and additional arcane shots over long fights.

The uptime point you raised is worth fleshing out -- Mastery puts all of its eggs in the pet basket, which is a benefit whenever hunter uptime drops but is a detriment when the pet loses uptime. Haste, for better or worse, splits the two, so it only gets completely devalued when both hunter and pet have no uptime. This isn't an argument as to which is better, but some people may want to hedge their bets and others might not.

On the first part, we must be looking at different gear sets, because I have the opposite experience in regards to switching out. Whether we do it by custom stats or actual reforging hopefully shouldn't matter. I have chosen to use custom stats to be able to accurately compare all 3 secondary stats.

The quoted part I will concede a very important point which I have not addressed in the OP, and I will do so putting haste and mastery on par with eachother and noting the differences and that each can be a dps gain over the other at different times - all in all we should strive to keep them somewhat balanced balanced. I will also make a note of the strength of each. Before I do this, I just want to make sure whether anyone is in disagreement that crit is the strongest of our secondary stats (if so, lets get that worked out as well).

The question that I keep struggling with is "Why stay with two Cobra Shots?" The more I analyse my own numbers, the more it shows me that the rotation should be, non-hasted, KC.AS.CSx3

The the following spreadsheet work was using the averaged numbers from my 10M BoT run (I do not have 4-pc Tier 11 bonus):

Average Damage / Shot (Crit and Norm) in readable whole-numbers
KC 25000
AS 12500
CS 11000

Now ... my latency is low at <40ms in raid so I have not accounted for that, but here's what I get for the two rotations:

Two Cobra
Three Cobra
Two CS+Three CS+Extra Arcane
From the calculations, anytime your CS is below 1.7 seconds, you are gaining DPS buying adding in a thrid between Kill Commands.

Thoughts?

I think its an interesting mathing out of what is indeed the intent of the OP. Since haste varies so much from reforging options and what exact gear you have, the only real way to see if you should add in another CS is to do the calculations on your own gear.
To contrast what you have done I tried ddoing it with the stat derivation model I use on FeD where the static casttime for CoS is 1.7 seconds and got the following results;

CoS:	13,000
AS:	16,000
KC:	33,000
Casttime:	1.7

The Stat derivation profile is full ilvl 379 gear without 4t11 bonus (I have another with), reforged for crit and hit. The damage numbers are rounded down to whole readable numbers

Doing the "normal" rotation with wait time results in

Ability	Secs	Avg Dmg
KC	1	33,000
AS	1	16,000
CS	1.7	13,000
CS	1.7	13,000
Total	6	75,000

DPS: 12,500

This is just the baseline of course. Then I added in a third CoS, delaying KC by 1.1 seconds:

Ability	Secs	Avg Dmg
KC	1	33,000
AS	1	16,000
CS	1.7	13,000
CS	1.7	13,000
CS	1.7	13,000
R-Secs	7.1	88,000

DPS: ~12,400

The delay is clearly over the threshold with a dps loss of just over 100 (rounded up to readable numbers), just to illustrate, the 0.3 second difference, I did it again with a 1.6 second CoS, which should be achievable without changing the numbers on the abilities, as they were all rounded down:

Ability	Secs	Avg Dmg
KC	1	33,000
AS	1	16,000
CS	1.6	13,000
CS	1.6	13,000
CS	1.6	13,000
R-Secs	6,8	88,000

DPS: ~12,950

Clearly the threshold of delaying KC for an additional CoS lies somewhere between 1.7 second casttime and 1.6 second casttime, for 2 different dmage (and gear) levels, so far the conclusion that delaying KC when you have a CoS casttime of 1.6 or under is supported. Doing a little simplified math, the delay allowed for KC is 1.04 seconds before it is a dps loss for the single rotation. Now adding 1 more CoS causes a lot of increased focus per rotation, and you aren't just gonna sit and let that go to waste, so you will be losing 1 CoS (and the casttime for that) to gain an AS every so often. Taking 2 rotations into account where you add in an extra AS looks like this (with 1.7 second CoS casttime):

Ability	Secs	Avg Dmg
KC	1	33,000
AS	1	16,000
CS	1.7	13,000
CS	1.7	13,000
CS	1.7	13,000
KC	1	33,000
AS	1	16,000
AS	1	16,000
CS	1.7	13,000
CS	1.7	13,000
R-Secs	13.5	179,000

DPS: 13,200

The interesting thing here is that even with a loss in the single rotation, you have a gain by delaying that KC if it lets you fire an additional AS off in the next rotation, and notice that in this rotation the third KC (the one that comes right after this rotation) is delayed by 1.5 seconds compared to the original rotation of KC, AS, 2CoS. The 1.5 second delay is of course over 2 rotations for a 0.75 second delay for each, well below that 1 second we established as a gain previously.
Whether you, personally, should be delaying KC for another CoS, becomes a little more complex if we bring in other focus regenerators, like Invigoration, Fervor, Roar of Recovery and Arcane Torrent if you are a blood elf, because at 1.7 seconds the modelling here shows that you only gain dps by delaying KC if, and only if, that additional CoS is what lets you fire off one more AS in the next rotation. If you would have been able to fire it off anyway (due to focus levels or focus gains) it is a loss as we can see in the first 2 rotational examples here.
Without a bit more work on the numbers at different gear levels, it is too early to say whether you should push KC at 1.6 second casttime for a third CoS, but tehre are strong indications that this is the case. If anyone else could be so kind as to add their numbers so we get real gear for the first 2 rotations {KC, AS, 2CoS} and {KC, AS, 3CoS} with your static casttime (not average, static) I can update the OP if this is show in more than the current 2 cases as a general recommendation.

Hi there can someone please explain why you use 1.7 sec as a cast time for cobra shot? Shouldnt it be more like 1.2 sec? Maybe i got something wrong but if we start with 2 sec than add 10% haste from gear, 15% from focus fire, 10% from haste buff, 3% from talents and at the end we take 0.2 sec from 4 set bonus shouldnt we get 1.2sec?

I am using 1.7 second casttime because that is the static casttime shown by the ilvl 372 model without 4t11 bonus. Also, for 4t11 the 0.2 seconds is taken from the base casttime, so it becomes a 1.8 second cast rather than a 2.0 second cast, meaning that a 4t11 CoS with 3/3 pathing, full FF and WF/HP/IT will have a casttime of 1.38 seconds before haste on gear. Regarding haste on gear, the amount you have depends a lot on your exact gear pickups, but I can tell you that the the baked in haste on the 372 gear (only 2t11) before reforging is only 822 rating or 6.5% haste, not 10%.

We get cobra shot at 1.3 sec cast x 3 = 3.9 + 1 sec arcane = 4.9 sec and there is no delaying on kill comand so shouldnt this be standard rotation + change cobra with arcane if enough focus? Dont know how latency comes in all this but if needed you can always ditch mastery for more haste...

This is currently in the OP, but auto shot fires while moving now (in any aspect) so there is no need to stop. I confirmed this unspeced with no aspect up to be sure.

You are correct, I did not remember this mention of stutter stepping when 4.0.6 rolled around with autoshot while moving. Thank you for making me aware of this, I will remove it immediately.

I´m sorry if this has been brought up earlier but couldn´t find it anywhere. I´ve started playing BM due to the lack of ret pallys in our roster, and I´m struggling at the start of the fight with the order I should be using CDs. AS soon as we pull, I cast SrS, then pop BW +CotW and KCx2 + 7-8 ASs until BW is gone, then Rapid fire + Berserking and starting the common "rotation". I´ve read that we shouldn´t be overlapping BW with dynamic haste effects, as during BW you shouldn´t be firing nothing but KC and AS. But if I check the option in FD "don´t overlap BW and Rapid Fire" it shows a dps decrease.

By the way, I find BM very useful and competitive in fights such valiona HC and omnotron HC, at least for the first trys, while learning the mechanics. Heavy and unpredictable movement fights fits BM playstyle like a glove and pretty much outperforms MM/SV until you master the encounter

This is similar to what I do however I go about the start a bit different. I use Serpent CotW BW KC Arcane Spam, when BW is up still have time left on CotW so I use 1 Cobra(for Serpent) Fervor to get in more AP boosted Arcane Shots. Then proceed with Rapid Fire Cobras to 110. To me it seems a bit of a waste to Cobra during CotW when you can get those Arcane shots out with Fervor on CD asap.