If anyone wants a framework to build a simulator on top of, I built most of one in Python about a year ago. It's intended for a pre-BC dual wielding Warrior, but it wouldn't be too hard to adapt. Things like Flurry and Windfury are already built in, though not the WF cooldown - that would be very simple to add however.

linky

It's not well documented and for that I apologize, I got distracted by the last fall semester and never finished/polished it, but I'm fairly sure the main combat engine works. If it would be helpful for anyone else's sim projects, feel free to PM asking me how to tinker with it. The executable is fight.py, there aren't (I don't think) any command line params; most of the specifiable inputs are either in a text file or constants in fight.py.

I'll take a look at that framework tomorrow, basicaly what I did was write the framework in C++ ( event dispatcher pattern) and then wrote a shaman on it, with his own list of talents (implemented only the flurry atm); may be I will get some ideas from your, I know python but I like more c++ if I have to do something for my amusement.

So...basically...a setup of crit and weapon that no one, EVER, are going to use to dps?

I'm not searching for the best combo of weapon speed, or %crit, I just doing some setups and see if the simulation confirms the reality (call WOW reality is an oxymoron )

First off, apologies for going over the top. There *was* some elementary misunderstanding of probability going on here, but not by you.

Now to turn to this result - well you don't need to go to that lengths to prove it. Part of the problem was I didn't realise how many words were being used on something (a) so simple and (b) already known.

Considering a case where you're autoattacking, no flurry, no specials, no parries.

WF is linear with weapon speeds above 3.0 , with a p[roc chance per hit of 1/5

WF is linear with weapon speeds between 1.5 and 3.0, with a proc chance per hit of 1/6. This is because there is exactly one ineligible swing after each WF proc

WF is linear with weapon speed between 1.0 and 1.5, with a proc chance per hit of 1/7. This is because there are exactly two ineligible swings after each WF proc.


So much is trivial - but it doesn't get us very far except as part of a more advanced simulation, and is already built into the existing sims. So I'm not quite sure what the purpose of this thread is.


I do however have a question about WF and cooldowns. I know that the cooldown only applies to the self-buff version of WF, and not to the totem. However, does the self-buff cooldown affect the totem proc chance?

That is, if a shaman casts the self-buff version on their offhand while using WF totem on the main hand - will WF procs from the (self-buffed) offhand trigger a cooldown and prevent the (totem-buffed) main hand from proccing?

If not, it seems to me that in raids, where the shaman will be dropping WF totem, a faster offhand might be workable, since you'd only have to worry about off-hand procs eating other off-hand procs. With DW miss rates, this wouldn't be a huge issue.

Incorrect, for big numbers you will get a WF every five swing, but not like this:
PROC
-
-
-
-
PROC
etc

Out of 1000 swings you'll get 200 procs (if you ignore the 3s CD) but it will sometimes look like this
PROC
-
PROC
PROC
-
-
etc

You're twisting statistics, you're saying that you'll get a 6 every 6. throw of the dice, but you could stil get 5 6's in a row. It's not that big of a difference, but it matters. Especially when you have a "made up" situation.

Someone has probably stated this before, but I'll do it again:
3s swing -> Fever and bigger procs
1s swing -> More small procs

When you introduce the 3s CD, more then 3 times as many small procs will be nullified by the CD. And even if there was no CD, specials would tip the scale in favor of slow weapons.

WF Totem doesn't have a cooldown at all, so it does not interact with WF Weapon in any way. However what you are proposing would result in a net loss of DPS for the shaman. Winfury Totem is strictly inferior to Windfury Weapon, even with the cooldown on the weapon enchant. We get 2 attacks instead of 1 with a much larger attack power bonus.


@dragkhar: Look, here's the issue. Regardless of whether your logic is correct, the argument you are making at this point in time is completely moot. To say that you are not considering crit (all your examples were for cases with 0% crit, and therefore no flurry either), and only considering a single-wield scenario (at weapon speeds that itemization does not exist for) makes this argument worthless to the public.

Worse though is that your posts now serve to confuse the windfury mechanic and present misinformation to shaman who come here seeking info on how to itemize and play their character At first glance through this thread a new shaman (and we're getting plenty of them visiting this particular forum lately) will come to the conclusion that he should just Dual Wield two daggers with windfury - which is a bad decision.

I spoke with Yes last night and he said that your source code is written extremely well. You might have a great model and sim there, but not right now. Right now it seems that you are struggling with some underlying assumptions of how to model things. You apparently came to some conclusions based on these flawed assumptions and you spent your time in this thread defending your conclusions soley based on your simulation. I would strongly encourage you to go pick apart Disquette, Pater or Tornhoof's simulators and examine their approach.

To be fair, this joke argument isn't math. It's argument by violation of lexical scope. It's the same kind of argument as:

God is love
Love is blind
Ray Charles is blind
Therefore Ray Charles is God

And the source of confusion comes from "is" meaning "equals" in some statements and "has the property of" in others. (In your example, the overloaded word is "longer".)

I'm not saying my argument has any merit, the point was that it was farcical.

Correct and accurate information is important to testing of any scientific model. Something that is not being done here in this thread.

I was talking about average, but never mind.

Ah right, so when they're dropping WF totem for the melee group, they're still putting the self-buff WF on their main hand? Good to know.

I posted my result on my first page to have a feedback in order to validate the simulator, but it seems that
most of readers didn't catch my objective (my fault for sure). I'm not trying to demonstrate anything to anyone, I'm not suggesting that you have to use a slow weapon, or you have to use a 2H weapons instead of DWing.

So please do not give me feedbacks of a teenager like I already received, once for all:
"this is a stupid post and you are stupid for posting it."

So thanks to Lactose, Hell and even Yes, I refined the combat simulator framework fixing the glitch about WF not accounting the extra damage due the weapon speed ( 34 * speed), I'm also using the time resolution that blizzard seems to use, 15 ms. On my first post also the maths I shown was incorrect due to a misunderstanding, to have the right formulas refer the one Hell written after that post, btw my formula was
valid only for speed > 3.0.

I implemented on the framework a Shaman using a single weapon, WF and Flurry are considered.

The graph reports on X-axis the speed from 0.1 to 8.0, on Y-axis is reported the DPS obtained with the weapon. The weapon is a 100 DPS one. Bottom line to top: 0%, 10%, 15%, 20%, 25%, 30%, 33% crit.



I believe now the graph reports correctly the espected damage; next step to validate the simulator si to introduce another weapon as offhand in order to see the interaction the two swings have each other (I'll try to have a 3D graph).

If your end-state goal is the modeling of DW weapons, you should probably consider making your output graph more useful by limiting the x-axis to relevant weapon speeds. Showing what the damage is at speed 8.0 isn't useful to anyone.

I'll do it, do you believe a [0.1 ; 4.0] is a reasonable range ?

1.3 to 3.0 is a reasonable range. There are no relevant 1h weapons outside that range.

Sure but the simulator doesn't have a weapon wow database ( may be in the future ) so I will do the simulations with a larger range than actually available weapons and then I will cut the graph in a range that reflect the existing weapons (just for the one interested in use it as a guide to choose in game weapons combo).

But if you want to use the simulator to estimate damage when you have certain haste effects on, you'll need speeds faster than 1.3. I think a lower bound of 0.5 would be sufficient for that.

To aid debugging your code, here's where you expect to see the discontinuities. I've listed all the "jumps" that fall within the range [0.5, 4]. Obviously you'd have to run the simulator for many hours to see them all clearly. As it stands, everything below about 1.5s is getting blurred out due to statistical fluctuations and/or the resolution of your ticker.


WITH NO CRIT/FLURRY
3s
1.5s
1s
0.75s
0.6s
0.5s


UNDER THE INFLUENCE OF FLURRY
3.9s
1.95s
1.3s
0.975s
0.78s
0.65s
0.557s


WHEN 0 < FLURRY UPTIME < 100%
3.9s
3s
1.95s
1.5s
1.3s
1s
0.975s
0.78s
0.75s
0.65s
0.6s
0.557s
0.5s

Edit: Reading back, I see your simulator already does throw the dice for each swing

Taking this further, assuming you are using a 30% Haste, are you applying Haste correctly?

This is common error to do when calculating hastes, example with 3.9 speed weapon, 30% Haste:
3.9*(1-Haste%)
3.9*0.7 = 2.73
This is incorrect.


This is how to correctly calculate hastes, same setup as above:
3.9/(1+Haste%)
3.9/1.30 = 3
Again, this is the correct way.

Lactose, thank you for this, I'm using the incorrect way indeed, reading the Flurry tooltip it say: "Increases your attack speed by 30%", so I just reduced by 30% the swing period, doing theWeaponSpeed * 0.7, I will replace it with theWeaponSpeed / 1.30.

As soon the framework is stable enough, I will put it on google code so errors like this will be spotted soon. I'm focusing at the moment on the framework and using a Shaman to see if the framework itself have all the features needed to simulate a character (I had to start from somewhere).

Questions:

I introduced a second weapon (offhand) and now I have a problem I need to solve.
I roll the dice at each swing, what happens if the two weapons shall do a swing at same time,
and both have a WF proc? Wich one I have to consider, MH or OH?

As anyone studied already the probability density function of blizzard random numbers
generator ?