The Division 2 Gearer Project – Probabilizer

The Division 2 Gearer Project is something I’ve been developing trying to use my Excel (and VBA) skills to find the best possible build in this game. For this to be possible I would need to:

  • Learn how to generate items the way the game does.
  • Create all possible items – this is, possibles models, brands, attributes and talents combinations.
    • Must consider combinations only achievable through recalibration.
  • Try ‘every’ possible loadout, created by combining one of the generated items for each slot, and find the best one.

Many decisions regarding subjective things – like what makes the best build – will have to be made and are gonna be explained as I go through them.

But, before completing the project I came with it’s first ‘product’, the Probabilizer (patent pending). Overwhelmed with the huge RNG the game drops have and not believing how nearing 10k items looted I haven’t found my desired backpack I stopped, after successfully creating all possible items (yellows/branded gear), and started calculating the odds of getting any of them, sharing my first results on Reddit.

For this 2nd and complete almost complete version of the Probabilizer, I considered the comment by user HerpDerpenberg on the Reddit post. It showed me how the game possibly generates items, which I explain with an example:

  1. When generating random gloves there are 10 possible gloves brands, so 10% is the probability of each brand being picked.
  2. If the rolled brand is Murakami, there are 2 possible models, ‘”Kote” Hand Guards’ and ‘”Yugake” Firearm Gloves’, so there is a 50% chance of either being rolled, let’s consider we rolled the “Yugake” model.
  3. Next, the first attribute is rolled, which can only be ‘Defensive’, and we have 4 possible defensive attributes on gloves (‘Armor Regen’, ‘Health’, ‘Health %’ and ‘Hazard Protection’), so there is a 25% possibility for each one of them. Let’s say we rolled ‘Health’.
  4. For the second attribute, we can have any of the 3 types, ‘Offensive’, ‘Defensive’ or ‘Utility’, so 1/3 chance for each. Let’s say we got ‘Defensive’ again.
  5. This time, there are only 3 blue attributes remaining, since ‘Health’ has already been rolled on this item and attributes can’t repeat, so 1/3 chance for each. Let’s say we rolled ‘Health %’ now.
  6. The last stat for this item is the talent slot, which can be: ‘Offensive Common’, ‘Defensive Common’, ‘Utility Common’, ‘Offensive Active’ or ‘Defensive Active’, so we have 20% chance of rolling each. Let’s say we rolled ‘Defensive Active’.
  7. The only possible ‘Defensive Active’ talent for gloves is ‘Clutch’.

So, to calculate the final probability of dropping this glove we get: 10% * 50% * 25% * 1/3 * 1/3 * 20% = 0.00027777%. This being the chance of dropping the item below from a branded gloves drop:

  • “Yugake” Firearm Gloves – Murakami Brand
  • 1st attribute ‘Health’
  • 2nd attribute ‘Health %’
  • Talent ‘Clutch’

Since a glove which rolled with ‘Health’ on 2nd attribute slot and ‘Health %’ on the 1st would be identical, for the purpose of this calculation, we can add the probability of this item to the first probability, and they are actually equal, so the probability of getting the item below is 2 * 0.00027777% = 0.00055556%.

  • “Yugake” Firearm Gloves – Murakami Brand
  • Attributes: ‘Health’ and ‘Health %’ (in any order)
  • Talent ‘Clutch’

When we divide 1 by a probability, the result is the number of attempts needed for a 63.23% chance of getting one (or more) of the desired result. Check this calculator.

If we divide 1 by our last result we get 1,800, which is an average number of gloves one needs to drop to have a good chance (63%) of dropping the gloves generated above. You can check on the game page in Uplay the total amount of items you have looted, I’m sitting in 9,070 with 356 hours played, if we say 1/10 of these are gloves, we get 907 gloves, so I’m only halfway and probably need more 356 hours to get these gloves.

Of course, if we consider recalibration we can get it faster, we only need “Yugake” gloves with either one of the attributes, ‘Health’ or ‘Health %’, and a defensive attribute on the other slot and ‘Clutch’ talent, or gloves with both desired attributes but with another ‘Active’ talent. I will have to cover recalibrations on my future steps of this Gearer Project, but for the Probabilizer it will stay out, I will try to reason why.

You must’ve noticed by now that we are not talking about stats (attributes and talents) values, I’m only considering their presence or not in the items, but any item can have good and bad rolls on the desired attributes. It is important to tell that every item in The Division 2 follows the same ‘budget’ formula. So if the item has a very good roll on one attribute it will have bad rolls on the others, since most of its budget was already spent on the good roll.

MIN/MAX is the term for minimizing weaknesses and maximizing strenghts when building a character/loadout.

This leads us to, in order to MIN/MAX, look for items with low native armor (which also consumes budget), exactly the talents we need and at least a good roll in one desired attribute and the other attribute being of the desired type. This way we can recalibrate a good rolled attribute over a low rolled one and increase the total budget of the item. If we recalibrate an item talents its budget stays the same.

The final result is a big Excel file (184MB original, 84 MB .ZIP, 52MB .RAR) that has every possible item listed (worksheets ‘Items’ and ‘Items Redux’) along with its drop chance. The second worksheet already groups and sums the probabilities of identical items. The probability shown is the chance of getting that specific item from any branded drop for the same slot. If you want the chance among any yellow gear drop – and considering each slot has the same chance of dropping – just divide the percentage by 6.

To use the spreadsheet just click the arrows in the heading titles (3rd row) and start filtering. Then check the probability in the last column. See the image below for this post example applied.

probabilizer-with-filters
click to open in new window

The chance is calculated using data from this worksheet by BestNadeThrower, which uses datamined data, updated by me, and can be seen on worksheets ‘Gear Talents’, ‘Gear Attributes’ and ‘Branded Gear’. ‘Salvaged’ and ‘Replica’ versions of items were removed along with green set items. It is important to note that Yaahl brand gear is calculated as having the same chance of dropping as any other brand, which is not what actually happens (Yaahl gear seems to only drop inside Dark Zones).

The links for you:

  • XLSX version 113MB – Microsoft Excel 2007+ compatible, without macros. Only ‘Items Redux’ spreadsheet. Zipped file 18MB.
  • XLSM version 184MB – Microsoft Excel 2007+ compatible, with macros. Both spreadsheets present. Zipped file 84MB.

Unfortunately, there is no Google Sheets version this time, since the loot pool now creates a table bigger than the maximum size supported.

Final thoughts

The RNG in this game is fucked up. In an unhealthy way. There are so many possible items variants that looking for 6 specific gear pieces to MIN/MAX a build is realistically impossible. And even when you get everything right you can still get very bad rolls and a lot of native armor when you are not looking for it.

This leads to senseless grinding. As you get nearer to the MIN/MAX build the probability of any drop being an upgrade keeps reducing (this is natural) but very soon, and not as close to MIN/MAX than it should, it gets so small that one can expect to play for weeks without finding an upgrade. This painfully kills the looter part of this Looter Shooter game.

I would like to leave some ideas of how the RNG could be reduced, making the grinding more satisfying. As it is not very polite to expose a problem and not propose a solution. For these, I am already considering the changes to crafted items that are coming in Title Update 5.

  • Raise the max material capacity by A LOT. We need to craft and recalibrate thousands of items to get an upgrade, don’t cap our resources so we can only craft 3 or 4.
  • Make every material needed to craft items receivable through deconstruction. Increase the value for the player of every item drop. Keep the farming wheel rolling through crafting and make the farming route clear.
  • Let us recalibrate two stats per item, but keep the recalibration budget the same. This would make the probability of any drop being an upgrade a lot higher and bring significance back to the looting of the game.

I hope you appreciated this post and that the spreadsheet can be useful for you. This is the first post of The Gearer Project and I expect to achieve much more. Even with all the problems I still love this game and plan to put a lot more hours into it. This post is just to help fellow agents, disseminate information and maybe show for someone making decisions how broken some things are in The Division 2.

The worksheet was created with the loot pool as in Title Update 4 and can be easily update when the loot pool changes. If this happens I will update the files and this post.

Edits

Edit 1

2-July-19

  • Added a little color to the ‘Items Redux’ worksheet, updated it in the .XLSX files (compressed and not) only.
  • Replaced the few remaining formulas among the data by its values, this should add speed for those getting long calculation times when filtering , updated it in the .XLSX files (compressed and not) only.

Leave a Reply