This post refers to a feature that will appear in the next website update, in a day or two.
Imaginary Strategies are a new feature for users who want complete control over how the optimizer ranks gear. We call them “imaginary” because you can make up anything you want, and make the optimizer do… anything you can imagine!
This is in contrast to our Adaptive Strategies and older Default Strategies, and also User Strategies created via our Gearing Strategy Simulation Batches, which are based entirely on simulation data. Imaginary Strategies are useful if you do not want to base your gear ranking on data, but rather on personal preference, anecdotal experience in-game, some math that you did on a napkin, a Spreadsheet of Doom – whatever floats your boat.
Most of what you need to know is directly on the page for creating an Imaginary Strategy in the Help section, reachable by choosing to Create a New Strategy from the gear optimizer, or editing an existing Imaginary Strategy via the edit link on it in Your Strategies list. This post has a few extra details, and can be used for discussion and asking questions.
Disclaimer: for the most part you are “on your own” with these strategies. We fully support the underlying optimization engine and will ensure that it correctly executes your strategy, but Team Robot will generally not provide much assistance with crafting your particular Imaginary strategies – it’s up to you to get it how you like it.
It takes a LOT of work to hand-craft a good gearing strategy. We don’t recommend doing this unless you really know what you are doing.
Even if you do know what you are doing, expect it to take a lot of trial and error.
If you are trying to match simulation results, one of our simulator-based strategies or our Adaptive strategies will almost certainly track closer to the simulation data. They use a lot of statistics to ensure that they match the data. The purpose of Imaginary strategies is to base gear ranking on something other than simulation data. Trying to manual tweak an Imaginary strategy to match your simulation data is almost guaranteed to be an exercise in frustration.
The user interface will not do a whole lot of error checking for you. If you put in something really wacky it will probably ignore it, and that’s about it. It is up to you to put in numbers that make sense.
If you choose to use caps, they must be entered as the amount from gear alone: add up the amount of haste shown on the tooltips for all of your items, enchants, and gems. Do not include consumables, do not include any rating multipliers (e.g. some specs get more haste per rating than others), and do not include any multipliers like the 2% from Sephuz’s Secret or temporary effects like Bloodlust.
The existence of actual meaningful cap in-game is very rare (though not impossible). They are more or less an arbitrary tool to try and get a single set of stat weights to generalize across a wider range, and effectively “shift” from one priority to another at a user-configurable point. Use them freely, but don’t get too hung up on how “right” they are – just tweak it until it ranks gear how you like it.
It is popular to “normalize” weights, usually to the primary stat. So you’ll see weights like Intellect = 1.0, Critical Strike = 0.9, etc. You can certainly use these, just keep in mind that you will have to enter values for special effects that are on the same “scale” or you will get very weird results.
We would recommend that you instead “normalize” all stat weights to DPS (or some other appropriate and meaningful metric). For example, if one point of Intellect gives 11.5 DPS, make it worth 11.5, not 1.0. Then you can enter much more meaningful values for special effects: if your 2-piece Tier 20 set bonus gives you 20,000 DPS, you can enter its value as 20000 rather than worrying about how to normalize it.
The stat summary and graph displays that you see on simulator-based strategies are not available for Imaginary strategies.
The “pantheon” trinkets from Antorus require some special handling. You will see two lines for them: one for the trinket, and one that says (Pantheon Effect) below it. The normal line should be the value of the personal trinket effect only. The (Pantheon Effect) line should be the value of the pantheon effect only. The pantheon effect value should be the maximum that you estimate it would be worth with a 30-man team who all have a variety of pantheon trinkets. We use statistics to automatically scale it down from there based on your Pantheon Trinket # setting in the filters/options on the optimizer.