Hi! Are there any plans to implement in the wonderfull Shadowcraft-tool the Challenge Mode statistic analysis based on scaled gear? Thanks on advance for answer!

March 16, 2016

I'd love to implement CM scaling in ShadowCraft. The engine should mostly support it since all procs are implemented under scaling rules currently but the UI would need to be modified to support CM scaling rules. Right now Tamen and I are quite busy trying to get things ready for Legion, if someone wants to help out and implement CM scaling we'd love it but it probably isn't going to happen unless we get more people helping with ShC.

Liked by: Duze

Do you think that the talent/playstyle combinatoric space for legion is shaping to be too big for a modeling tool whereas a simulation tool only contends with the talents/action and not with the combinatorics? Is the ShC model needed to get reasonable results for legion too complex to be manageable?

December 10, 2015

If anything I'd say the opposite. The large search space, even if we confine ourselves just to talents and artifacts may cause problems for heavily simulation driven approaches. Yes simulation driven approaches can handle the variety of talents much more easily however APL development is still a primary bottleneck for talent evaluation. Some combinations may be handled sanely with the same APL but will the handling by anywhere close to optimal? This is hard to know without putting non-trivial effort into rotation space exploration for that talent combination.

I expect the size of the combinatoric space will benefit model driven approaches since it makes doing a full talent space (and artifact space) search feasible on a reasonable personal computer. While people can produce monolithic talent ranking lists these have known weaknesses and everyone agrees ideally players should evaluate their individual gear profile.

The large search space also may lead to benefits for communities having both analytical model and simulation tools. Analytical models can often suggest bounds on talent behavior which can be used to motivate rotation exploration. Writing an APL to nail the DfA interval at as close to 20 seconds as possible is challenging however a model can suggest that bound very easily showing the potential of that rotation.

In my admittedly biased view, the large combinatoric space in Legion will be well served by analytical models.

What are some resources for people like myself to get into TC?

December 10, 2015

Broadly speaking there are three components to theorycrafting.
1) In-game mechanic testing. This doesn't require any special coding skills, just a knowledge of game mechanics and an ability to construct experiments.

2) Model building and validation. Given our knowledge of mechanics we would like to construct models that can help provide information about the game. This tends to be the most coding and math intensive part of theorycrafting although it is important not to overestimate the math complexity of many mechanics.

3) Design space Exploration. This involves using the models to find better, and hopefully optimal, gear, talent, and rotation choices for a spec. This typically requires experience and game knowledge to suggest potentially fruitful directions to explore.

The most important resource to start theorycrafting, no matter what part you want to contribute to, is experienced theorycrafters in your class/spec community. They can help direct you to potentially interesting open problems in the community, ideas they may explored and found fruitless, and generally knowledge that isn't well documented. Obviously you don't have to listen to the experienced theorycrafters in your community but it will probably help.

Some useful resources for each of the three components:
1) Theck wrote a series of posts on WoW experimental design that is required reading if you want to do mechanic testing.
http://www.sacredduty.net/tag/tc101/

Warcraft Logs is often important for data collection and analysis. I don't know of any very good guides to it but if you want to do mechanic testing you should familiarize yourself with WCL and its very powerful feature set.
https://www.warcraftlogs.com/

2) For those interesting in model building the most useful resources are existing models. Below are a number of good spreadsheet examples to look at. Note that spreadsheets aren't the only form of model but they tend to be the most readable in my experience.
Binkenstein's LavaCalcs (7.0): https://docs.google.com/spreadsheets/d/1hSpFEjvcjbZ184PCOqCFN7CyfzX0a7ZWQxiGGB1I-FM
Rotund's Windwalker Spreadsheet (6.1): https://docs.google.com/spreadsheets/d/1ISW2o_al1tjwdovymFC75KUm_pVryc6bVjrhk_9_sLE
Hamlet's HealerCalcs (6.1/6.2 PTR): http://iam.yellingontheinternet.com/healercalcs/
I'd planned to link Aldriana's combat and assassination spreadsheets from 3.3 since those are what I learned from but the links appear to be dead.

Also shameless plug, I wrote up a short example of simple theorycrafting model in this blog post: http://ravenholdt.net/index.php/blog/21-roguechat/106-theorycrafting-and-simple-math

3) Since a large amount of design exploration today is done in SimulationCraft the obvious first place to look is the SimC documentation: https://github.com/simulationcraft/simc/wiki/StartersGuide
As part of the series of posts linked above Theck also wrote a very good introduction to SimC: http://www.sacredduty.net/2014/07/21/tc101-testing-simulationcraft/

Is arhicmonde trinket really that good for combat? If i play a bit with shadowcraft i see a steep drop in its value with 2 adds and more (the loss actually increases with the number of adds). At 3 adds the Archimonde trinket rates at the same level as BRF NM other trinket for me or is it a SD bug?

July 15, 2015

Yup, looks like there was a bug in the ShC code. I was applying the eviscerate multiplier after the damage distribution had already been copied for blade flurry. I'll have a new build out later tonight that should fix that.

With the tier 18 2pc sub bonus, is it better to incorporate vanish into your opening rotation to boost burst damage? or is it better to hold until after FW drops from dance?

July 13, 2015

2pc only rotations for sub are probably underexplored. A sad side effect of theorycrafting primarily in BiS gearsets.

With 2pc only whether its better to synchronize vanishes with dance or not is an open question. With 2pc and 4pc you should follow the same kind of rules we follow for AR with combat. Use vanish ASAP, never hold it through a finisher so you don't waste any of the cooldown reduction from 4pc.

Why did you stopped seriously raiding with horseman?

July 13, 2015

Horsemen died. Exactly why is a long story but not exactly a unique one looking at the fates of some of the other MoP low hour guilds.

With the recently buffed class trinket, do you expect sub or assassination to be higher in terms of single target dps with the t18 4pc + the class trinket, assuming BiS gear for both specs?

July 13, 2015

Assuming T18 4pc and/or archimonde trinket sub tends to be ahead of assn. in the interim assn can be even or slightly better, especially with T17 2pc/T18 2pc and no archimonde trinket. There is also an important skill consideration, assn is a much less punishing spec than sub, especially at with T18 gear and/or archimonde trinket. For players who struggle with sub the gap between the specs is probably small enough to support playing assn.

Fierydemise

Nothing here yet