So i came in here adimently objecting to the original Posters claims, mainly because i do not agree with the methods he calculated the armor probabilities (and still do not). however, I then took my own probabilites of how to calculate armor probabilities. (which is much higher, as I divided them by class, and each class has its own armor sold each week. I then extended this, and created a data set where each piece of armor was given an ID number: [u][b]Warlock[/b][/u] 1-Apotheosis veil 2-light beyond nemosis 3-skull of dire ahamkara 4-sunbreakers 5-heart of praxic fire 6-voidfang vestments [u][b]Titan[/b][/u] 1-insurmountable skullfort 2-helm of inmost light 3-helm of saint-14 4-no backup plans 5-crest of alpha lupi 6-armametarian [u][b]Hunter[/b][/u] 1-Ach. Symbiote 2-knucklehead radar 3-mask of the third man 4-young ahamkara spine 5-crest of alpha lupi 6-lucky raspberry Then I made a data set for the first 13 weeks, which was when the DLC released more options week-----------------warlock---------------------------titan-----------------------------hunter 1------------------------------4----------------------------------4-----------------------------------3 2------------------------------4----------------------------------2-----------------------------------6 3------------------------------1----------------------------------5-----------------------------------2 4------------------------------4----------------------------------3-----------------------------------1 5------------------------------6----------------------------------2-----------------------------------6 6------------------------------2----------------------------------6-----------------------------------1 7------------------------------6----------------------------------6-----------------------------------5 8------------------------------6----------------------------------6-----------------------------------6 9------------------------------4----------------------------------5-----------------------------------3 10----------------------------6----------------------------------6-----------------------------------5 11----------------------------6----------------------------------4-----------------------------------4 12----------------------------6----------------------------------6-----------------------------------5 13----------------------------2----------------------------------2-----------------------------------4 I then created similar datasets for Post-DLC xur, and likewise for weapons. I then took this data and submitted it to a battery of the following tests for randomness: Frequency Test: Monobit Frequency Test: Block Runs Test Test for the Longest Runs of Ones in a Block Binary Matrix Rank Test Discrete Fourier Transform (Spectral Test) Non-Overlapping Template Matching Test Overlapping Template Matching Test Maurer's Universal Statistical Test Linear Complexity Test Serial Test Approximate Entropy Test Cumulative Sums Test Random Excursions Test Random Excursions Variant Test the results are quite extensive, and are in the form of Stata Output. If anyone would like to see the output, I can send you the Stata file upon request. But the overall conclusion was that the armor failed the majority of the test (when the standard error is ignored) wheras, the weapons seem to follow a pattern that is still indicitive of randomness. From all of the armor sets, the hunters failed the fewest amount of tests. Hoewver, with this in mind, the sample sizes are far to small, so the standard error on the tests were so large that no statistically conclusive evidence can be drawn. however, the current data does give a reason to at least raise suspicion. So I apologize to the OP for my original attack, it seems that you do have something with your claims (though i still disagree with your probability calculations for armor). and thank you for making a topic that provoked me to conduct my study, I love data analysis, and i had a blast testing this. Edit: and right after i did this, I watched the video where Deej admited that Xur isn't random... sooooo.... yeah

pretty sure deej confirmed he wasn't RNG but i guess its ok.

except the Deej confirmed that xur is rng

No, he isn't. They said,he isn't.

I have a few issues with the math here. Assuming his distribution is correct (and appears so to me)... The chance of a a specific weapon being sold during any given week is as he said... 7.14%. The chance of ANY weapon being offered is 100%. I know that sounds stupid to point out, but it's important for part 2. The chance of the a specific weapon being sold two week in a row is about 0.51%, or once every 200 tries (e.g. just under four years). BUT, the chance of having a case where some weapon was sold two times in a row (that is, where it doesn't matter what weapon it is, but that whatever weapon sold last week is sold this week) is.... 7.14%. This isn't all that improbable, and should happen about 1/14 weeks in the long run. You can do the similar check for armor. The chance of some armor being sold for multiple weeks in a row would be (I haven't counted armors, but using the OP's 26 count...): Same unspecified piece of armor two weeks in a row -- 3.85% (1/26 weeks) Same unspecified piece of armor three weeks in a row -- 0.148% (1/676 weeks, or 13 years) Same unspecified piece of armor four weeks in a row -- 0,0043% (1/23323 weeks, or 448 years) So, yeah.. it's unlikely to happen BEFORE the fact. But, as people have noted.. once it happens on the second week, the chance of an armor repeating for a third week is only 3.85%. If it does happen, then it's 3.85% to happen for a fourth week.

Without knowing things about statistics, to see xur sellings only things from the dlc twice (three?) times is just a business strategy to make people buy the extension (if some haven't bought it yet) So either way : - they manually chose the stuff => it's not random - there are weight on exotics to make some more easy to get even if the choice is random. Gjallahorn weight 1, dragons breath weight 99. Xur has 99/100 to pick dragon's breath => it's still "random"

this argument is so flawed. let me go into a few reasons why First of all, let me hit on the armor point you mentioned. Your calculations have suggested that 1 piece of armor is sold each week, but this is not the case. 3 pieces of armor all sold each week, and they are divided into class types. second, ou have assumed that RNG means all outcomes are evenly weighted, which has never been claimed, and may or may not be the case. third, you realize, no matter what probabilities you calculate (even if your calculations were not seriously off base), every possible permutation should yield the same probability (unless weighted). so if we assume unwieghted items. No matter what outcome we recieve, any other possible outcome would have the EXACT SAME probability of occuring... I dont care if he sold a gjallarhorn ever single week so far, that would have the exact same chance of happening as what we have experienced. the odds of every possible outcome are small, yet identical. so do not go throwing around numbers acting like you know all about statistics, when you obviously do not have a proper education on the nature of probabilities.

.071 percent chance of being sold? Something went [i]very[/i] wrong with you math about right here

Flawed. Consecutive doesn't stack. The % stays the same each week. If it's 50/50, it stays 50/50 throughout. One not dropping doesn't increase the chance the next week.

I thought ps had 2 exclusives.