25 pieces off Chariot Bosses

[FailureMidgard]

Just the other day we got two. I was on quartermaster too as I am the only one left in my group that needs a zhaf boss drop and time was running short. I guess I'll have a emp pin -3 if I decide to lvl mnk.
Zhaf 5/6/boss runs can be a pain and mostly because our biggest factor on success is how often frogs use chorus. We can be 5-10 minutes ahead of our curve and then sorry kids, 6f madame decided she didn't want to use magic hammer and frog cheer tonight.

Failure, can you show how you go 97%?

edit: 0/16 after tonight

That was just after 5 runs.

(5/6)^10 = the chance to never see a specific 1 of those items. This is 16.15% chance to not see each item. Anyway, that's a 16% chance to not see Usu Body, also a 16% chance to not see Usu head, also a 16% chance to not see Ares Legs, etc.

I actually didn't do the math past that and just did 16*6 which is incorrect, but is more like the average # of pieces not seen yet for all groups. Basically, I did the chance that you havent seen 1 + the chance you havent seen 2 + the chance you havent seen 3 + chance you havent seen 4 + chance you havent seen 5. Which in reality, if I havent seen 2, I also havent seen 1.

16.15+13.54+11.35+9.23+8.11+6.72 = 65.1% chance that you haven't seen at least 1 drop after 5 runs, and of course, there are more possibilities that you haven't seen even 2 or 3 of them after 5 runs. (These #s are .1615*100, then .1615*(100-current sum of %)...repeat this for all 6 drops)

Point is, there's some murphy's law going on here...yes, the chance to not get the drop you want is low, but the chance to not get 1 of the 6 possible drops is pretty high compared to what people think. And of course, if you get lucky and get your drop early, you don't keep running and track the drops you still haven't gotten, so it is always the drop you need that pisses you off. Also, I dunno about some groups, but we were going for ~3-4 off the same chariot at once...of course one of them was going to take longer, it was nearly statistically impossible for one to not take longer.

[meson]

25th kill of BC. Still no slops.

heres our distribution of 25 kills:
Slops 0
Crackows 10
Gloves 12
Cuisses 11
Jinpachi 9
Togi 8

[Daiiawn]

1/1 on powderkeg, 1/19 on bodb's slops over the course of about 10 months -_- Jinpachi and Togi for all

[Balb]

^ lol this is close to meson, 'cept im pretty sure he was like 1/4 on Powderkeg, 0/21 on BC, he's had slops for 6+ months now. It's funny he mentioned all those cuisses drops, now that one of our guys finally got a pair of Deimos Cuisses to drop, now it hasnt dropped in like 5+ runs lol. Such a cockblock...

[chapstick]

1/1 on my Bodb slops, 0/lots on PK.

[Deadgye]

The only given is that the peice you need to complete somthing will NOT drop.

Random drops are random, unless you need it, then it never drops.

Truth.

Skadi always drops. Always.

LIES

35 Skadi body was insanely easy to get. I've done at least 20+ LAC and haven't seen a single 25 skadi body...

[rithridan]

35 Skadi body was insanely easy to get. I've done at least 20+ LAC and haven't seen a single 25 skadi body...

Haaaaaaaaaaate weather popping right when you pull that 8th gear when it shouldve been an easy pull. Freya's body isn't insanely easy to get because of the randomness thats called weather in Bhaflau. And even then when you do finally pop the mofo you get screwed because it either doesn't pop the damn NM for 20 minutes or you get a few and they don't even drop a 35. I've been 12/15 on skadi for ages now missing 35 head/body/hands and 15 body(will only get/kill when i get 35).

What a big surprise Bhaflau and SSR cockblocking like no tomorrow (any other almost finished set in our group is being cockblocked by SSR 35s).

[Gaea]

I think it's really hard to say with these kinds of mobs because you have a mob that would suggest an equal percentage value for all the drops with a 100% chance to drop two of them. If a mob can drop seven 25s (LBC) of something that should give you about a 14% chance of seeing a specific piece and about a 2% chance you'll actually see double of that piece (2 morrigan bodies from the same LBC for example).

Because 2 of the same piece can drop and there's up to two drops, we can definitely rule out that the drops are handled something like

Bodb's Crown OR Bodb's Robe OR Ea's Dastanas OR Freyr's Ledelsens
and
Phobos's Mask OR Phobos's Cuirass OR Tsukikazu Haidate

However a model like that would explain why the morrigan's robe would be more rare (since it's in a pool of greater number of items). However, that wouldn't account for an item having the ability to drop twice. So, let's try a different idea!

Since two of the drops are 100%, you have to try to guess if it's doing something like the following.

Let's say, like other drops in the game, each item in the pool here has a set rarity value. In this case we'll say about 1.5265% (4/256) because we want to try to give a somewhat equal chance for each item to drop. In order for this to work the system would have to have a predefined priority order for each item that gives one item a SLIGHTLY higher advantage than another.

Let's theorize that specific order is Bodb's Crown, Ea's Dastanas, Freyr's Ledelsens, Phobos's Mask, Phobos's Cuirass, Tsukikazu Haidate, Bodb's Robe.

I made the robe last because that's the piece you are having trouble with.

Now, You kill LBC, it's time to see what you get.
The first of your two guaranteed drops gets randomed for.
In thise case Bodb's Crown at 4/256. However the random came up with a value between 5-256 so NO DROP! Let's go to the next item in the list. Ea's Dastanas. Random number comes up 3! DROP! *YAY* The other items won't even be checked!

Second guaranteed drop. Same process.
Bodb's Crown MISS
Ea's Dastanas MISS
Freyr's Ledelsens MISS
Phobos's Mask MISS
Phobos's Cuirass MISS
Tsukikazu Haidate MISS
Bodb's Robe MISS

(Repeat because we missed and we're supposed to get a drop. Remember we only have 1.5% chance of getting a hit.
Bodb's Crown MISS
Ea's Dastanas MISS
Freyr's Ledelsens MISS
Phobos's Mask MISS
Phobos's Cuirass MISS
Tsukikazu Haidate HIT!
Bodb's Robe **Not checked again because we got a winner**

This is just an idea of why to explain why some items might be more common than other. It's simply just a design flaw that's trying to use existing game mechanics to generate a guaranteed two drops. However, its hard to detect because each item has such a low % of winning before another item, that it's likely that it goes through this random process a few times, statistically missing all 7 items several times and repeating the process until gets a hit. Stastically, you could go through the entire process 13 times and get nothing but misses and get a "guaranteed" hit on the 14th time because of "statistics". However, it could miss every item in 14 cycles, or it could hit the first item in the first cycle.

There would be no way for us to realistically guess which order these items are in even with a really, really really large data set. If anything we might be able to guess which two items would be first. In a huge dataset that proves the order I randomly created here, Double crowns would be the most common item in a huge data set.

This theory itself could be proved if we were able to prove that a specific chariot has a higher double drop value on a specific item for everyone across every server. However, at 4/256*4/256 you are talking about a dataset that can prove something within about 2/100ths of a single percentage point.

So, this sytem, while some what flawed by design, could suggest why your item actually is "technically" more rare. In two cycles of trying to get a hit, there will be 6 other items checked twice, before the one at the bottom (or near the bottom) of the list gets checked twice. That could be your item! QQ

[Therin]

Hey, I think it's just bad luck.

My Salvage group has had almost an equal distribution of all 25 pieces from every boss, except for Tsukikazu feet from LAC, and the first time we got those, we got double. So yeah, I'm going with luck and not some random arbitrary system like you just posted.

[Gaea]

Well, I'll agree that it's bad luck that you didn't get your piece. My system would need a dataset in the millions to prove one item has an advantage over the other albeit very slight.

However this is a logical way to handle the process of guaranteeing 2 drops out of a possible 7 while at the same time giving each piece what would be perceived by humans as a roughly equal chance for each piece.

For all we know which drops you are gonna get are determined at the point you enter and it goes by the freakin moon phase. I doubt it, but who knows. I'm just trying to get the point across that you can't give a drop a 100% value and a 14% value. It's gotta be either one or the other.

My suggestion just offers you how both could be done with a possible solution of why on a really large scale one item would technically be more common than the other.

But yeah, its really just bad luck. I see people here talking about getting a ton of jinpachis and no slops from battleclad chariot when I've gotten mostly togis and slops and didn't get my jinpachi til about the 10th kill. Just bad luck

[Thorny]

Or you could just roll once per drop slot and have each piece have an equal value in the range the roll represents. Shut up Poof.

[Amele]

Or you could just roll once per drop slot and have each piece have an equal value in the range the roll represents. Shut up Poof.

^ this.

why generate 14 numbers (wasting your precious precious clock cycles! omg) when you could generate 1?

[Gaea]

Or you could just roll once per drop slot and have each piece have an equal value in the range the roll represents. Shut up Poof.

Can you illustrate your point? I don't follow. How do you decide which two will be the 100% drops?

[Zhais]

Can you illustrate your point? I don't follow. How do you decide which two will be the 100% drops?

uh, wow. And this coming from someone who was babbling on about high-level computer stuff in another thread?


Create an array of 7 length, each with the name of a possible drop
generate a random integer from 0-6.
Take the loot array @ random integer for first drop.
Repeat for second drop.....


edit: or even to save your precious clock cycles! (thanks amele) you could just randomize a 6-bit number, take the first three bits for the first array index, the last three bits for the 2nd array index.

[Gaea]

uh, wow. And this coming from someone who was babbling on about high-level computer stuff in another thread?


Create an array of 7 length, each with the name of a possible drop
generate a random integer from 0-6.
Take the loot array @ random integer for first drop.
Repeat for second drop.....


edit: or even to save your precious clock cycles! (thanks amele) you could just randomize a 6-bit number, take the first three bits for the first array index, the last three bits for the 2nd array index.

I see that makes sense. It would provide a seemingly more "equal" chance between the items, but the reason I disagree with this being the method is because there are items in the game that share the same place holder that have unequal chances of being the item that drops.

In another thread we were discussng Mee Degi. Now if Mee Degi can only drop kote or knuckles, and knuckles is much more common, they are sharing the same place holder, but one item is a little bit more rare.

My solution shows how this could be expressed for those cases as well. And as far as my example with the chariots, I don't think calculating a drop rate for something is that big a deal. This calculation is done at a small fraction of a second. My argument is that with enemies that are killed by the thousands and tens of thousands on a daily basis you'd prolly want to make the system to be the smallest fraction as possible and maintain maximum flexibility.

Now, my example with the chariots does suggest that the drop rates between each of these items suggest some of the items have a slightly better chance at dropping than the others. However, the way this is accomplished is so minuscule that it would take a data se of millions of kills to start deducing anything resembling a pattern, so I don't think it's really that important.

[Therin]

This thread isn't titled "25 peices off Chariot Bosses and Mee Deggi" now is it.

[Amele]

In another thread we were discussng Mee Degi. Now if Mee Degi can only drop kote or knuckles, and knuckles is much more common, they are sharing the same place holder, but one item is a little bit more rare.

I was going to write you some assembly but it was taking too long.

create an array that holds item id's and relative rarity.

generate a random number between 0 and max (what max is doesn't really matter) compare to relative rarity.

if in bucket A, drop trash item. if in bucket B, drop awesome item.

if you want to save clock cycles, do this for all 10slots at once using a single 64 bit word. (gives you 6 bits per slot; or up to 64 unique items per slot, with a minimum resolution of 1.5625%)

[Gaea]

I was going to write you some assembly but it was taking too long.

create an array that holds item id's and relative rarity.

generate a random number between 0 and max (what max is doesn't really matter) compare to relative rarity.

if in bucket A, drop trash item. if in bucket B, drop awesome item.

if you want to save clock cycles, do this for all 10slots at once using a single 64 bit word. (gives you 6 bits per slot; or up to 64 unique items per slot, with a minimum resolution of 1.5625%)

I think I follow but I'd like an illustration on this too, or anything you can do to elaborate. I think one of the most important steps we can take toward trying to understand these things is understand the math that governs these underlying mechanics.

Also I'm not worried about processing efficiency, I was only using that to point out that people were suggesting more work was being done than needed whereas I feel there's no reason to have item drop rates to have to vary by less than 1/256. And even if it were the case, it would take an extremely large dataset for us to prove it anyways.

I'm trying to figure out if your proposition could give any kind of advantage to one item over another when two (or more) items of different rarity values occupy the same slot. When looking at sets of data (particularly BCNM drops) it sometimes seem that items with the same place holder that have virtually identical drop rates (that are pretty high) one still seems a little more common than the other. I assumed it was a phenomenon such as this.

I also I also thought of a way to rig the phenomenon in my model that causes drop rates at the bottom of the pool having a better chance of getting selected before the next cycle of checks by simply and very slightly raising its drop rate. I think if you did this kinda thing with the bottom part of the list it could compensate and stabilize the over all drop rate to a nearly identical, overall rate.

[Amele]

I think I follow but I'd like an illustration on this too, or anything you can do to elaborate. I think one of the most important steps we can take toward trying to understand these things is understand the math that governs these underlying mechanics.

illustrate it how? with an example? I basically gave the outline (in plain english) of how the algorithm would work. about all I could do now is sit down and spit out 10 lines of pseudo-code that would say the same thing...

[Hanyoko]

I agree, that was a very straight forward example.

Bucket #1 = "Inventory -1" - 254/256
Bucket #2 = "AwesomeSauce DX" - 2/256

Random! 0.53326432
Therefore you win! "Inventory -1"