Theory About How the Devs Increased Legendary Drop Rate

Since the devs increased the legendary drop rate, a few interesting things that I (and most of the people I run with) did not realize could happen have come to light.

Apparently, it is possible for a single boss to drop multiple legendaries for a single kill. In addition, it is possible for both of these drops to be the exact same legendary. It’s also patently obvious to pretty much everyone I run with that the loot tables are not balanced (e.g. you have a higher chance of getting X legendary than the rest of the loot table). Also, as far as I know, the double drops for each boss have also always shared a legendary with all other double drops from that boss (e.g. the double drops are either XX or XY but never YY).

Let me drop some numbers first. I have been farming Sentinel a lot trying to get a Stolen Edge of Arcvynorr. In over 100 completed runs (and a number of runs where the Old Sentinel glitched), I have gotten 7 Songs of Vigor (only 1 before the legendary increase), 1 Vigilant Power Scouter, and no Stolen Edges or Vigilance Links from the Old Sentinel; 2 Vyn’s Quivers (only 1 before the legendary increase) and 6 Chrono Keys from Vyn; and 8 Guardian Shieldbreakers (only 1 before the legendary increase) and 2 Symbiotic Gauntlet (both before legendary increase) from Arc.

Based on these numbers (admittedly a small sample size), before the legendary increase, the drop ratio I got for Sentinel was 1:0:0:0, 0:1 for Vyn, and 1:2 for Arc. While it’s not particularly robust, it’s reasonable to assume that there was either an equal (or at least close to equal) chance for each legendary to drop.

After the legendary increase, these ratios went to 6:1:0:0 for Old Sentinel, 6:1 for Vyn, and 7:0 for Arc. This suggests that something extreme was done, but only to one of the items in each loot table.

As such, my theory is this. On the back end, when a boss is killed, the game makes 2 checks in addition to all of the other normal loot stuff. One check is to determine if a legendary from the entire loot table drops while the other is simply to determine if the “common” legendary drops. It’s possible it’s been this way since release (if there have been cases of double drops before the legendary increase, that’s confirmation), but, based on my own (admittedly meager) data, it’s obvious to me that the devs didn’t globally increase the legendary drop rate: they simply increased the chance for the common legendary by about 5 times.

I’d be interested if anyone else has anything else to add to this (it would be really awesome for a developer to either confirm or deny this too!).

1 Like

Yes, there was a thread here about a double drop. Both legendary items were the same with the same stats.

1 Like

Having put quite a few nights into trying to understand the data behind boss drops, I don’t think there is a common legendary and a rare legendary slot. As far as I can tell, no legendaries are given higher priority, only whether they are using the Normal mode drop rate (which is set to zero for advanced only drops) or Advanced mode drop rate so 4 different conditional weights (ie NormalDropOnNormal, NormalDropOnAdvanced, AdvancedDropOnNormal, and AdvancedDropOnAdvanced) which all boss drops point to (meaning this isn’t set separately for each legendary). They do have the capacity to apply a custom scale multiplier per item, but none of the legendaries currently do this as far as i know.

What you are seeing is likely pure RNG in action. Depending on the RNG method, you can get something called RNG bias where the way it gets parsed I suppose you could say, it mathematically favours certain values more than others. I have a very basic understanding of it as I saw a talk on it once that was promoting newer RNG methods over older more commonly used (and flawed) methods that people still use because they are easier to implement and they work fast (very important in games)!

Personally I just think RNGJesus has decided you will be bestowed with certain legendaries more than others because that is the will of the gods in the same way I’ve got the Firmware Update 6x more than any other legendary while others have never ever seen it drop once.

While I have found drop slots data for chests and loot packs, this information for bosses has eluded me. :sleeping:

Just in the case of the Arc (since that’s the most “pure” data set I’ve got since there are no exceptions), assuming that there really is an equal chance to get any possible legendary, there’s a 3.125% chance of that happening.

From what I understand, the “fast” method of RNG is to use the current time (often just looking at the smallest fraction of a second that the computer tracks; it could also be another easy-to-access but difficult-to-replicate piece of data) and then applying an algorithm to it to get a result that fits within chosen parameters; it’s not random because it’s really just a function of the time, but it’s effectively so because, without computer assistance (to repeat the time precisely) and knowledge of the randomization function, it’s basically impossible for a human to replicate it. The “truly random” method of RNG relies upon the computer waiting for something truly random to occur (the event that I was told about was passing a large set of bits back, forth a massive number of times and using the statistically probabilistic data loss as the random factor) and then apply an algorithm to it. This is truly random because it’s relying on a random physical process to occur (there is a tiny iota of potential statistical bias if a given machine has specific bit storage that is of a higher likelihood for data loss), but it’s only really necessary for crypto where true randomness (generally in the creation and utilization of passwords and encryption keys) is absolutely required.

The “fast” method of RNG is good enough that I have never actually heard of a game actually generating a bias anywhere near what I have personally experienced. I might be willing to admit to being one of the “lucky” 3.125% that has had that experience with Guardian Arc, but when you factor in the similar results I experienced with the Old Sentinel and Vyn, it gets into the realm of the absurd (as in “this needs to be put into scientific notation because otherwise there are too many zeroes”).

We could more definitively answer the question with a significantly larger data set of people who have farmed a specific mission a large number of times before and after the legendary drop rate increase (I’m about 50% before and 50% afterwards, with the numbers provided). Of my friends, the only mission they’ve heavily farmed both before and after is Sentinel (and they’ve done so with me, sharing my loot results) so that data is unneeded.

Has anyone else been farming a specific mission repeatedly and tracking their loot results diligently? It would be great to have more data (I’ll keep tracking my own loot, but tracking the legendaries that drop can be pretty time consuming without some ability to data mine the results of all players, which is only really possible for the devs to do, because every piece of data is going to require, on average, an hour or so of work, which is about how often I get a legendary while grinding).

The bias i was talking about was more to do with applying something like a modulus to the random number to convert it into say an index in the item pool array. It has less to do with the actual RNG and more how it gets converted into a usable value. Because modulus counts up from zero to the modulus number and loops back around to zero, the lower numbers will have more chances than higher numbers to roll…or something like that. Not really my area, so I’m gonna stop right there.

RNG is a super complicated and tricky topic, and everyone is going to have a different opinion on it, because we all have our own measure of how lucky we are.

No amount of user sourced data (that you’d be able to get your hands on) is going to give you an accurate idea of whether or not there is a bias towards certain items. You would need thousands of results to even get close to being able to paint a picture of what’s actually going on. I don’t know thousands of people and you can bet at least half of those would be more interested in giving you their opinion rather than their results (as is the Internet :disappointed_relieved:). Only the Gearbox devs would have that information because they are collecting the raw (unopinionated) data from the whole player base…Which has led to a significant bump in the global boss legendary drop rate :sunglasses:

It’s totally possible there could be 2 or more drop slots with different drop rates applied to them, but the actual item that gets picked should always be weighted the same. It really is just RNG trolling you like it does everyone else. I have gotten over 20 Invective shotguns in Destiny which should have been something else, but that’s just not what the RNG gods had planned for me. All you can do is keep trying until you get the thing you want.

I like your theory. I think it is true

In reading theories like this I just can’t help but chuckle to myself because if this is true I picture the Devs at work at GearBox having a board meeting, “SOMEONE HAS FIGURED IT OUT!! WE HAVE TO CHANGE IT BEFORE IS IT CONFIRMED!!” so they stay “one step ahead” of its users hahahaa

Or vice versa this theory is totally incorrect and in that same board meeting they have it up on some projector screen and having a good laugh

I wish Carnut’s RNG post from the Borderlands section of the old Gearbox forums was still available. That thing was epic. I can’t recall any pertinent details, but it was super in-depth discussion of how RNG works and would be interesting to compare with this discussion.

Just realized that I should probably ping the devs to some extent if I really want to get a dev response of some kind to this. Any chance of that @jythri or @JoeKGBX?

Hallelujah! It’s not just me :slight_smile:

It’s possible, but I don’t think so. Since you’re arguing that when bosses drop a legendary, whether they drop their “A” legendary or “B” legendary is just random chance (50-50), let’s do a thought experiment on Vyn and Arc, who each have two possible legendaries they can drop. Going 7:0 with Arc would be like coin-tossing 7 heads in a row, a 0.78125% chance of happening (that’s zero-point, or less than 1% chance). After the unlikelihood of that, going 6:1 on Vyn would be even more unusual. It’s possible that it’s RNG, but those types of numbers with a sample size of 100 make it quite unlikely. Also, I experienced the same exact thing.

I, too, got a very disproportionate number of Sings of Vigor, Guardian Shieldbreakers, and Chrono Keys. And for that matter, I got a ton of Stable Executioners and Oaths of the Sustained on Hylis, with not a single Leechsteel Brooch or Vow of Vengeance, and the same again with Bunker Buster and Firmware Upgrade. In fact, I got two Bunker Busters from H3nchm4n on the same map once, and it was same H3nchm4n, mind - not just the same drop from both of his clones.

I think we all know that the RNG monster can be a tough adversary, and that we shouldn’t jump to conclusions, but I have a ton of data, scientifically recorded, that mirrors Kitru’s experience. He pointed me here from another thread, and the whole time I was reading his thread I couldn’t help but feel a little vindication and relief that it wasn’t just me - it was obvious we’re doing and seeing the same thing. I’m not saying that it’s proof-positive, but I’m at least 99% convinced now that not all legendaries have the same chance of dropping under all circumstances. Maybe team composition or difficulty or the phase of the moon has an impact, I dunno, but it’s something.

And to reiterate - we’re not talking about the chance of a legendary drop on a given run, we’re talking about the chance of a boss to drop a specific legendary from his pool after the game has decided that the boss will drop a legendary.

Ehh :sweat_smile:
I looked at the raw data files and that’s the impression I got. I too have had very mixed results with boss drops, but there could be a bunch of reasons for that including actual bugs in the way drops are determined.

Speaking of, I haven’t looked at the boss drop rates since the patch came out :smirk:

My “findings” for the exact same item : “stolen edge of impossibletorememberthespellingynor”.

I ran Sentinel/adv lots and lots of time during the lootpocalypse in order to get one and got none. Always got the song instead. An hour after the lootpocalypse ended I ran it once more and got the edge :flushed: !

So, at the time, it just seemed to me that the lootpocalypse code was a “superb”-“horizontal”-“untested” modification that just pushed everything in one direction, and the edge was pushed off and fell out of the table for a while…

That, or… It got “COBRA”-ed during lootpocalypse :laughing: !

On another note, yesterday I was running Saboteur/adv to just kill the first boss to get a Modernista and the first legendary that dropped, before I finally got one after a few runs, was a time-killer which I thought was supposed to only drop in normal mode. (I even ran it in normal in order right after to check if the legendaries hadn’t been swapped and got a second time-killer).

So as far as I’m concerned :

  • normal mode drops the normal item
  • advanced mode either drops the advanced item or a chance at the normal item as well or both…

Also, on tuesday (thus not during lootpocalypse) I had an algorithm/adv run where I got :

  • Bunker Buster (henchman adv.)
  • Firmware update 1.51C (Geoff norm.)
  • Shield Web interdictor (Geoff adv.)
  • Poor M-Pulse controller (emperor norm.)

(was going for Alamo-7 and all I got was these lousy…) :stuck_out_tongue_winking_eye:

Normal mode drops can drop on any mode.

All I really care about is I get to have fun shooting stuff (or punching, slashing and so on) and once and a while something nice drops. It’s icing on the cake for me. If the drop rates aren’t too low than I get to have a nicely iced cake. It sucks getting a cake with no icing. At the same time a cake with too much icing makes me feel ill. Wait… we were talking about drop rates and not cake, right? Cake is nice but I think I’d rather have a plate of nachos with salsa, cheese, sour cream and a few other nice toppings. That’s what I’m craving right now anyway.


I personally would like a packet of chips, but there are chicken schnitzels in the works so I’m not touching any snack food. :heart:

@Kitru @trevordmoore @shoogli I just had a quick look at the files in the PC version of the game and nothing seems to have changed, meaning either the new drop rate changes are still being hotfixed in at the title screen or they are just applying a global modifier the scales them all up evenly (who knows). This shall just have to remain a mystery for the time being. :wink:

(this isn’t exactly along the lines of the OP, but it’s of a similar theme and I didn’t feel like making a new thread for pretty much the same discussion. I’m still talking about the mechanics/probability of loot drops)

It’s been a while since I posted anything in here, but, in the interval, I have been collecting a lot of legendaries. As of this moment, I have exactly 278 (I’ve got 2 free shard generators, but I know I’ve got a pair of Heliophagic Goggles in the overflow), pretty much all of which were acquired on Advanced (only 1 drop was from normal mode). The list of them is as follows (I’m not listing lore legendaries or loot pack legendaries, which is why it totals to less than 278; should be around 250-260 total since I am too lazy to count, atm):

Henchman: 17 Bunker Buster, 7 Henchman’s Capacitor
Geoff: 22 Shield Web Interdictor, 12 Firmware Update
Galactic Emperor: 10 Poor M-Pulse Controller, 6 Hate Reflector
ISIC: 9 Alamo-7, 5 Hostile Takeover, 4 Pacifier, 3 Plasmite Transducer

Warlord Nix: 13 Veil Manipulator, 2 Sentinel Reset Switch
Conservator: 6 Long Goodbye, 3 Blissbeast Skull Plate, 3 Wolfskin, 2 Void Resonator

Jailer Hylis: 4 Oath of the Sustained, 4 Stable Executioner, 3 Vow of Vengeance, 3 Leechsteel Brooch

Bagranth: 6 Lorrian Skill Spike, 5 Vibro-Core Module, 4 Codex Fragment, 2 Veil Breaker Rounds

Guardian Arc: 12 Guardian Shieldbreaker, 6 Symbiotic Gauntlet
Guardian Vyn: 11 Vyn’s Quiver, 5 Chrono Key
Old Sentinel: 7 Song of Vigor, 4 Stolen Edge of Arcvynorr, 4 Vigilant Power Scouter, 2 Vigilance Link

Supervisor Antem: Bola’s Target Finder 2, Mossire’s Mukluks 2
Conservator: 3 Symbiotic Spores, 3 Vidania Root Poultice, 2 Intermittent Angel, 1 One-4-All Shield Array

Supervisor Varl: 12 Modernista, 5 Time Killer
Maligner: 7 Mini-Singularity Launcher, 2 Quartermastery Bin
Foreman Grall: 3 Boots of the Brute, 2 Premium Health Plan, 1 Mag Daemon, 1 Synthaline Tonic

Rendain: 2 Vow of Zealous Fury, 1 Heliophagic Goggles, 1 Helm of the Imperium, 1 Doomsday Key

While in some cases, there’s not particularly stirring evidence, in others, there’s really strong probabilistic evidence that there isn’t an equal chance of all loot dropping. I compared these results to @lowlines’ Normal/Advanced boss drops table with some interesting results (keep in mind, the only drop I have from normal mode is 1 VoZF; these are basically all Advanced runs).

(all ratios are Normal drops : Advanced drops)

Henchman: 7 : 17
Geoff: 12 : 22
Galactic Emperor: 10 : 6
ISIC: 7 : 14

Warlord Nix: 2 : 13
Conservator: 6 : 8

Jailer Hylis: 7 : 7

Bagranth: 6 : 11

Guardian Vyn: 6 : 12
Guardian Arc: 11 : 5
Old Sentinel: 6 : 11

Supervisor Antem: 2 : 2
Conservator: 6 : 3

Supervisor Varl: 5 : 12
Maligner: 2 : 7
Foreman Grall: 3 : 4

Rendain: (irrelevant since I got 1 VoZF from Normal mode; discounting that one, it would be 2 : 2 since I’ve gotten 1 of each on Advanced)

This raises some interesting questions. In most cases, there seems to be a roughly 1 : 2 ratio. In a few, it’s 2 : 1. In a few others, it’s basically even (some of which show a 2 : 1 ratio if you group them by highest/lowest drops rather than lowline’s categories; Hylis goes from 7:7 to 8:6, Grall goes from 3:4 to 5:2). Now, I’m not suggesting that I have a representative sample for all of the bosses, but the evidence for some of them is pretty strong (especially the mini-bosses that only have 2 drops), when I’m dealing with 20-30 total drops.

Now, I can’t remember if the Great Dataminer got the Normal/Advanced splits from the game files or whether it was constructed anecdotally. Depending upon where he got the information (and how reliable it is; he has stated that it could tweaked in the hotfixes since he doesn’t get to break those apart to sup upon their delicious informational goodness), there are a few ways to interpret this.

One way would be that lowlines’ categories are totally correct and each boss has its loot table tweaked individually and that there is no pattern whatsoever. Sometimes the Advanced drop is more common, sometimes the Normal drop is more common, and sometimes it’s a wash.

Another would be that lowlines’ categories are incorrect, which means that there’s a 1:2 drop rate on advanced favoring certain drops and that, because people have shown that it is possible to get the “advanced” drop on normal, it is simply very low chance of occurring rather than impossible (I recall that the table in question was made back before the Great Loot Drop Buff so probability being screwy is more probable).

Now, another possibility is just like what I suggested initially: there is a single drop for each boss that has a significantly higher drop chance than the others. For all of the 4 drop bosses that I have gotten lots of drops from (e.g. enough to argue that the population is of sufficient size to get a decent p), there’s one drop that is, statistically, about twice as likely as any of the others.

Regardless, it seems to me that it’s pretty apparent that not all drops are created equal, even moreso than it was when I first posted this.

1 Like

Also, how MUCH did they increase the drop rate by lmfao!?

I mean before I seriously think it must have something like 5% or less (as I was STILL finding legendaries from maps that I’d done like 50+ times), but now… I’d say 65%? Maybe higher?

The buff to the drop rate is ABSURD. I mean on maps with a lot of bosses (i.e. The Algorithm), I barely ever seem to walk away with anything less than at least ONE legendary at this point ffs lol.

I mean, I’m all for making legendaries less of a grind to obtain, but when I’m vendoring extras after almost every mission then I think that they may have gone slightly too far. XD

[quote=“Ulithium_Dragon, post:16, topic:1539669, full:true”]The buff to the drop rate is ABSURD. I mean on maps with a lot of bosses (i.e. The Algorithm), I barely ever seem to walk away with anything less than at least ONE legendary at this point ffs lol.

I think that’s somewhat intended, actually, since the legendaries cost the same regardless of their stats so you always want to get the max stat version (which has a roughly 50% chance of occurring, from what I could tell). Farming for legendaries isn’t so much about getting the legendary as it is getting one with max stats on it.

Algorithm is definitely a weird case where legendaries are concerned because Henchman and Geoff appear to have a chance to drop 2 legendaries, whereas I’ve only ever seen every other boss drop a single one. It’s also got 4 bosses that drop legendaries, which is the most of any mission.

My guess is that it’s a 50% chance of getting a legendary from any given boss. Algo has 4 bosses, so that’s about a 6.25% chance of not getting any legendaries (if you assume that Hench and Geoff have 2 chances, that’s a ~1.56% chance). Definitely makes farming the single boss missions a lot less rewarding, though.

Every boss has a chance to drop two legendaries. one from the Normal loot pool and one from the Advanced loot pool. Some bosses have more than one drop per loot pool if I remember correctly.

Checking here it appears that ISIC, the Varelsi Conservator (The Void’s Edge), Jailer Hylis, Bagranth the Gunhulk, The Old Sentinel, the Varelsi Conservator (The Experiment), Thrall Foreman Grall, and Rendain all have 2 legendaries per loot pool (Normal and Advanced).

I’m well aware. I actually referenced that same table in my most recent number-y post. That table, however, is based off of datamined information and does not necessarily reflect the state of the game, especially since I believe @lowlines couldn’t find anything changes to gear variables when the Lootpocalypse or drop rate buff occurred (the changes were implemented through hot fix, not a patch).

What I am attempting to surmise, however, is how the information I’ve gathered (which is probabilistic but obtained in practice) in game is related to what lowlines datamined (which is certain but theoretical). They do not seem to sync up and the way they don’t sync up does not have a readily apparent pattern, which suggests that the information lowlines presents doesn’t actually present the reality of the story mode loot tables, especially since there’s a lot of evidence for a “common” and “uncommon” assignments for legendaries that isn’t tied to lowlines’ datamined categorization.

Btw, the bosses with larger loot tables aren’t random; they’re all the final bosses of their story mission. Mini-bosses have 2 (potentially 1 Normal and 1 Advanced) and actual bosses have 4 (potentially 2 Normal and 2 Advanced).