View Full Version : Tanking Need help programming a simulator

12-12-2010, 05:19 PM
I won't be able to work on this at all this week because it's finals for me, so I'm going to be turning to my studies (one last hard semester, next semester will be a virtual breeze compared to this one, so I just gotta hunker down this one last week, then I'm pretty much home free), but I have an idea for a new simulator very similar to the one in my 3.3.3 EHP thread, to try and simulate whether avoidance or stamina is better now for tanks.

One thing I've noticed in heroics is that healers have massive mana issues. Some of it may be healers not used to it, some may be players not used to not being able to take subsidiary damage, etc. etc., but whatever the reason, mana is definitely an issue now. With mana an issue, avoidance becomes far more important than in WotLK, imo.

However, just HOW important?

I need help though, I'm not a very good programmer, and I also need a lot of parses from different classes. I would like to do all of this with excel macros, so if you're really good with those I could use your help.

I also need combat parses of HEROIC fights or RAIDS only, with appropriately geared/spec'd healers of every class. Especially fights that are less about healing raid damage than healing tank damage.

I also need information from healers in conjunction with these parses, I need mana efficiency numbers, what spells they use when, etc.

I will fully admit I have NOT researched these aspects, especially b/c I have been playing my warlock. All I have noticed is that there hasn't been a big "okay cata's out avoidance vs. stam again guys!" thread (thank god because I feel like it would just turn into a crap fest since there doesn't seem to be a lot of real solid math done in it yet), and I think this is going to become another question, and I do not have the answer for it.

So, it is my intent to build a simulator that simulates tank damage intake and compares it to healer mana output. Do a hit run-series like the EHP thread did, and then when I tank tanks X amount of hits in a row, probably depending on tank HP and how hard a boss hits, then a healer drops their bigger heal spell that expells more mana, calculate in what regen rates do exist, and then if a healer runs out of mana and a tank takes damage you can either see how long it will take for that to happen by running the simulation until the tank dies, or have it set to a certain length of time to see if a tank will die in a given fight. Run this simulation thousands of times, and you should be able to get some quantifiable numbers.

Granted, this bares several "real world" fallicies, such as healers healing DPS that take damage, boss special abilities, etc. etc. So there's gonna hafta been some extrapolation here, but I hope that if we can get a lot of people working on this, we can get some real data out there that has some solidity behind it before everyone starts going "LOL N00B WHY YOU GEMMING AVOIDANCE?!" or "stam? what do you think this is WotLK!??!"

I'd really like to avoid the inflammatory threads as much as possible and instead turn this into a very scientific and educational discussion.

Any help is much appreciated. People great at programming, please PM me if you think you can help over the next month or so to put this together.

People that want to provide parses: I really prefer WoL reports, and it is free. please PM me the info, and a summary would be very nice (total healer mana, regen rates, etc. etc., as much info as you can give me without me having to dig it all apart by hand will help me a lot).

Anyone else that has input on what this simulator could look like, feel free to PM me or post here, I think this could be a very fun project.

But ya, let the theorycrafting begin! LOL.

12-13-2010, 03:45 AM
It is an interesting idea and while I am a bit too busy to offer any help immediately I am curious about what conclusions can be drawn from this. One thing you can do with enough parses (but how do you get enough) is it check how often tanks die compared to how they are geared. Lets say do a graph of tanks deaths per boss fighton on one axis and the tanks avoidance on the other axis. Will probably take a buss load of logs to give any decent data but should really say something about EH vs avoidance.

12-13-2010, 08:23 AM
Well I think the only problem with that is that sometimes logs or graphs like that don't really show the hole story. Say you have two tanks, one that isn't as good skill wise but stacks avoidance but with a terribad healer, and one that is really good skill wise that stacks EHP and has an amazing healer. Even if the EHP tank is technically "Gearing wrong" (again this is a dataless assumption, I have no idea if going for avoidance or EHP is better at this point, this is just a hypothetical scenario) but uses tons of cooldowns and the healer is really pro and clutch, then they "doing it wrong" tank might life while the less skilled/aware tank might be "doing it right" so to speak. Taking into account the human element is really hard to do, so I'd rather try and take the human element out of the math, and add it back in during the analysis.

12-14-2010, 01:02 AM
With enough datalogs the human part should not be a problem but getting that much data is probably the problem. Does mana regen show in the logs btw or does it have to be calculated?

12-14-2010, 01:14 AM
I'm not sure, my guess is it has to be calculated, but I can neither confirm nor deny this at the moment.