Theck's MATLAB TPS analysis - WotLK/3.x

[theckhd]

Enchants/Foods:

Setup:
Rotation: 969
Seals: SoV
Glyphs: V
Talents: 3V+3C
Gear: Majiben's progression gear set
File: calc_enchants.m

This is similar to the stat analysis, in that I calculate how much TPS each point of a given stat gives you, and then just do some simple multiplication. I use Majiben's gear set since it's below both hit cap and expertise soft cap.

Code: Select all

Enchant              DPS    TPS
Berserking          51.5  126.3
Accuracy            35.6   89.0
Ti Weapon Chain     28.0   71.1
Superior Potency    23.9   58.6
Potency             22.9   57.4
Greater Potency     18.4   45.1
Mongoose            18.3   40.6
26 Agility          10.8   25.7

Not a lot has changed here. Interestingly, the SoV 5-stack proc seems to have pushed Superior Potency ahead of the old BC Potency enchant in effectiveness. Accuracy and the Titanium Weapon Chain are still the best threat options.

Code: Select all

Food                DPS    TPS
Dragonfin Fillet   48.0  121.0
Snapper Extreme    42.2  107.8
Rhino. Wyrmsteak   40.1   97.2
Fish Feast         38.5   93.0
Bl. Drag. Fillet   18.8   45.8

Nothing surprising here either, Dragonfin Fillet is still the best food, with Snapper Extreme solidly in 2nd place. Rhinolicious Wyrmsteak will drop to half of the shown value once you pass the dodge soft cap. Fish Feast is still a pretty solid threat food for players whose guilds provide it.

TLDR Summary:

• Accuracy and Titanium Weapon Chain are still the best threat enchants
• Dragonfin Fillet is still the best threat food
• The sky is still blue

[Conaan!]

alright, thanks for the explanation theck, makes much more sense

[raenac]

Tons of great information Theck. Is there a graph somewhere for best mitigation and/or avoidance food/weapon enchant?

[culhag]

I'm thinking about modding RatingBuster to display a "Threat" summary in the tooltips, using all this theorycrafting, but I'll have to learn a bit before I can do something really accurate.

EDIT: after staring at calc_stat_tps.m for half an hour, I'm afraid I'll never get past using the static values of TPS per stat point that Theck posted which could be a good enough approximation in 3.1, but won't cut it in 3.2 with the diminishing return on BV.

[knaughty]

Theck,

You've posed TPS numbers for Reckoning 5/5 which show it as being "still worst but actually decent now". It does have an efficiency advantage, of course - zero wasted points to pick it up.

I had a question, however: I thought that Reckoning had variable TPS/talent point - how much TPS per point is 3/5 reckoning? I believe that 3/5 was found to be the "best" level to take Reck to, if you were going to take it.

*edit*

And SotP + Reck would seem to synergise fairly well - what's the TPS of 5/5 SotP + 3/5 Reck ?

*edit2*

/fail - SotP doesn't apply to the swing proc from a 5-stack. Synergy with Reck removed. Annoying.

[theckhd]

The latest talent analysis post covers a few of the comparisons you want. It shows 5S and 5S+5R; 5S+3R should be somewhere in the middle of those, probably just a bit ahead of 1V+3C.

However, to calculate the value per point for the first plot in that post, I just compared 0/5 Reckoning with 5/5 and divided by 5. As you point out, Reckoning's value per point isn't strictly linear. Just to check, I ran some code to see how much DPS & TPS each point of reckoning added individually:

Code: Select all

point #    DPS       TPS
1        17.6890   33.1455
2        17.4973   32.7864
3        17.3060   32.4279
4        17.1150   32.0701
5        16.9244   31.7129

As you can see, it's not linear but it's awfully close. The average TPS is 32.4, and it varies by less than +/- 1 TPS over the full 5-point span. So it should be reasonable to treat it as a fixed value per point. Using those numbers, 5S+3R is going to be about 30-40 TPS ahead of 1V+3C on the talent spec plot (1V+3C is a little over 8k, 5S is a little below 7950, 3R is worth around 96 TPS).

[Dread]

Theck,

You've posed TPS numbers for Reckoning 5/5 which show it as being "still worst but actually decent now". It does have an efficiency advantage, of course - zero wasted points to pick it up.

I had a question, however: I thought that Reckoning had variable TPS/talent point - how much TPS per point is 3/5 reckoning? I believe that 3/5 was found to be the "best" level to take Reck to, if you were going to take it.

*edit*

And SotP + Reck would seem to synergise fairly well - what's the TPS of 5/5 SotP + 3/5 Reck ?

*edit2*

/fail - SotP doesn't apply to the swing proc from a 5-stack. Synergy with Reck removed. Annoying.

Yeah, I originally thought this might be the case as well, but SotP not affecting the 5 stack proc kinda killed it.

[thorn]

theck, do you have any information on (or are you able to plug in numbers for) the difference between using a potion of speed or a potion of wild magic during wings? (or even a powerful rejuvination potion at any point in the fight.)

i seem to have heard tanks talking about potion of speed being preferable, but that sounds pretty inaccurate...

(the same principal applies to people who save wings for bloodlust; i don't imagine there is much of a benefit to that, as opposed to using wings off the pull so it's available for use again before the fight is over.)

[majiben]

theck, do you have any information on (or are you able to plug in numbers for) the difference between using a potion of speed or a potion of wild magic during wings? (or even a powerful rejuvination potion at any point in the fight.)

i seem to have heard tanks talking about potion of speed being preferable, but that sounds pretty inaccurate...

(the same principal applies to people who save wings for bloodlust; i don't imagine there is much of a benefit to that, as opposed to using wings off the pull so it's available for use again before the fight is over.)

Well based upon this graph theck posted a few posts back we can roughly approximate the value of those potions.

Low balling the 10 crit and 11.6 SP values of 10 and 5 Tps respectively, and a value of 7 dps for 10 haste we get the following results:

286.2 Tps for the wild magic potion
350.0 Tps for the haste potion

Even using a high ball value of 11 Tps for crit we only get
306.2 TPS for the wild magic potion

Now my numbers are only approximations from the graph but they put the haste potion solidly in the lead.


Now on to the AW and heroism question: With dps this question is easier, they don't have to worry about DP lockouts. With them it's simply a function of the fight length and how soon heroism is used. If you could fit in one more AW by using it out of sync with heroism then it is worth it to them. But given tank's need to keep CDs available it throws a wrench into the works.

[theckhd]

theck, do you have any information on (or are you able to plug in numbers for) the difference between using a potion of speed or a potion of wild magic during wings? (or even a powerful rejuvination potion at any point in the fight.)

i seem to have heard tanks talking about potion of speed being preferable, but that sounds pretty inaccurate...

(the same principal applies to people who save wings for bloodlust; i don't imagine there is much of a benefit to that, as opposed to using wings off the pull so it's available for use again before the fight is over.)

Well, from this analysis (this plot in particular) we can estimate the TPS increase of each point of crit, haste, and spellpower at around 1, 0.75, and 0.5 respectively. So the potion of speed would give you 500*0.75=375 TPS for 15 seconds, while the potion wild magic would give 200*(1+0.5)=300 TPS for 15 seconds. Wings will increase the benefit of each, but I doubt it will change the order at all.

There was a long thread discussing Heroism timing a while ago. I don't remember if any conclusions were arrived at with respect to Wings timing, but it might be a good read.

<edit> Drat, Maji beat me to it

[majiben]

Too slow theck

But you forgot to factor in that your SP value is for 11.6 SP not 10.

[theckhd]

Too slow theck

But you forgot to factor in that your SP value is for 11.6 SP not 10.

I pulled the numbers off of the "equal value" plot, which compares 1 crit to 1 Haste to 1 SP.

[majiben]

Ah, well that plot shows 1 SP significantly below .5 TPS but 11.6 SP is practically flush with 5 TPS on the line graph but the end result is the same. The speed potion leads.

[thorn]

hmm. alright. originally my intuition told me that 200SP+200crit > 500 haste, but when people less lazy than me ( ) actually do the math, you're right, haste pot is what i will stock up on.

[knaughty]

Theck, I imagine your reck numbers assume a fast (suboptimal-for-reck) weapon?

Surprised by how linear the plot is. I remember it being a lot less linear back when I was working out my Brutalus spec. Then again, maybe our maths was just wrong back in the Olden Days before we had a MATLAB expert trolling the boards.