From what we already know about loot 2.0 there's such a thing called Optimal loot. As described by MA in their dev blogs it can be measured by non-shrapnel to shrapnel ratio and plays a big role in obtaining valuable loot. Achieving optimal loot is more desirable as we usually want drops that have more markup than shrapnel. They also told us there are numerous factors related to that such as overkill, healing between mob attacks etc.
When we hunt a mob there are lots of things we cannot directly control. Like we cannot predict if we're gonna kill the next mob with 2 crits or he's gonna dodge 5 times in a row. Or we can suddenly crit on the last shot thus creating huge overkill.
Knowing that I decided to do some testing to compare loot between various individual mobs which we've killed in a different ways. For example the loot for all mobs which we've killed fast (in few shots) might be different compared to the case when mob have evaded multiple times. Or not.
The main hypothesis is: does the chance of obtaining valuable item from an individual mob depends on how we've killed that individual mob..
Spoiler for geeks
The script I used for analysis is written in R language. You may have a source code here https://pastebin.com/50NSg39W
The idea of method is to perform a 2x2 contingency test on how many times we've looted / not looted our desired item vs how many kills we've made in a special / non-special way. That kind of test is designed to work even on small sample sizes thus not requiring us to kill millions of mobs. Since it is 2x2 test all 2 factors must be boolean. So the "special" way of killing a mob must be also defined in a boolean manner.
1st test: all questions about rare pleak wing distribution were answered NO
Here are some preliminary results of the 2nd test.
For the 2nd test I used TT pistol, no amps,no pills, no Ares on 10hp Vixens
The results are quite different than in case of pleaks so at least this approach can be used as a method to determine if an item drops like a mission token or as natural loot.
As we can see the results of having good/bad loot from each single mob strongly depend on how the mob was killed.
In some extreme case we have 2.5x less chance of looting an item if mob has evaded twice.
This shows strong influence of RNG on the results.
This will require killing many mobs before evaluating your loot composition.
Also this will be an extra factor when evaluating loot patterns as the item may not drop even if its loot wave time simply because of bad RNG.
Update #3: I've killed 50% more mobs than it was in test 2 and combined the results.
That just lead to more overall confidence with highest being 99.92%.
When we hunt a mob there are lots of things we cannot directly control. Like we cannot predict if we're gonna kill the next mob with 2 crits or he's gonna dodge 5 times in a row. Or we can suddenly crit on the last shot thus creating huge overkill.
Knowing that I decided to do some testing to compare loot between various individual mobs which we've killed in a different ways. For example the loot for all mobs which we've killed fast (in few shots) might be different compared to the case when mob have evaded multiple times. Or not.
The main hypothesis is: does the chance of obtaining valuable item from an individual mob depends on how we've killed that individual mob..
Spoiler for geeks
The script I used for analysis is written in R language. You may have a source code here https://pastebin.com/50NSg39W
The idea of method is to perform a 2x2 contingency test on how many times we've looted / not looted our desired item vs how many kills we've made in a special / non-special way. That kind of test is designed to work even on small sample sizes thus not requiring us to kill millions of mobs. Since it is 2x2 test all 2 factors must be boolean. So the "special" way of killing a mob must be also defined in a boolean manner.
1st test: all questions about rare pleak wing distribution were answered NO
Here are some preliminary results of the 2nd test.
For the 2nd test I used TT pistol, no amps,no pills, no Ares on 10hp Vixens
The results are quite different than in case of pleaks so at least this approach can be used as a method to determine if an item drops like a mission token or as natural loot.
Do we have dirrerent chance of looting Vixen Andriod Gear if | Answer | How certain we are about that answer | Estimated multiplier to an average chance of looting the item |
We've killed the mob in exactly 3 shots | YES | 99.4% | 1.455 |
We've killed the mob in 5 or more shots | YES | 99.2% | 0.482 |
We've landed all shots on the mob | YES | 99.1% | 1.512 |
The mob evaded/dodged/jammed our attack at least once | YES | 99.1% | 0.66 |
The mob evaded/dodged/jammed our attack at least twice | YES | 99.1% | 0.387 |
As we can see the results of having good/bad loot from each single mob strongly depend on how the mob was killed.
In some extreme case we have 2.5x less chance of looting an item if mob has evaded twice.
This shows strong influence of RNG on the results.
This will require killing many mobs before evaluating your loot composition.
Also this will be an extra factor when evaluating loot patterns as the item may not drop even if its loot wave time simply because of bad RNG.
Update #3: I've killed 50% more mobs than it was in test 2 and combined the results.
That just lead to more overall confidence with highest being 99.92%.
Last edited: