Feedback for Musco, You Asked For It (In Regards to Drop Rates)

[brazilboy]

You're math is incorrect.

 

Even if you account for 2 chances at a saber per pack, 3 packs would be 6 chances. If the saber drop rate were 1%, that would only be 5.85199% chance of seeing a saber in those three packs. Even a 30 pack hypercrate (60 "chances") only equates to a 45.28434% chance of seeing a saber if the drop rate is 1%.

 

Actually this is exactly my math. 99%^60=54.72% -- this is the NOTCHANCE ... Therefore, 1 - 54.72% = 45.28%, which is the number you cite and is the chance. [i mistakenly typed "packs" when I meant hypercrates, but I corrected it.]

 

I think the actual drop rate is lower than 1% because with a 1% hypothetical drop rate on the saber and two "chances" per pack, or 480 "chances" in 240 packs, there is a 99.19667% chance of seeing at least one saber. Notice, though, that even with 240 packs (480 "chances"), there is still no guarantee of seeing a saber.

 

Again, this is exactly my math. 99%^480 = 0.8% ... therefore 1 - .8% = 99.2% chance of getting at least 1 saber, which is your number above.

 

The only thing I didn't assume was 2 chances per pack. I thought it was only one chance.

 

I agree I think the drop rate is lower like 0.05%.

Edited February 13, 2016 by brazilboy

[Ratajack]

Actually this is exactly my math. 99%^60=54.72% -- this is the NOTCHANCE ... Therefore, 1 - 54.72% = 45.28%, which is the number you cite and is the chance. [i mistakenly typed "packs" when I meant hypercrates, but I corrected it.]

 

 

 

Again, this is exactly my math. 99%^480 = 0.8% ... therefore 1 - .8% = 99.2% chance of getting at least 1 saber, which is your number above.

 

The only thing I didn't assume was 2 chances per pack. I thought it was only one chance.

 

I agree I think the drop rate is lower like 0.05%.

 

Changing "3 packs" to "3 hypercrates" makes a world of difference in the math, doesn't it?

[AbsolutGrndZero]

I hate the 'chance cubes' idea. New packs should only contain the stuff from that pack. If we want older stuff, we can get the Grand Packs... bring those back too. Seriously, I'm usually a "Bioware cheerleader" but even this has me saying *** BIOFAIL.

[Dyhanna]

I bought 2 hypercrates:

 

1 sidecar mount

1 marquis mount

2 Uxibeast

Lots or armor but no complete sets (except for 2 on chance cubes but nothing cool)

No hot tub

No lightsabers

Various more common things including some mounts & pets but nothing worth a lot

I did get about half of the decos but again, not one hot tub in 2 crates?

Chance cubes were wholly unfulfilling

 

Normally when I buy 2 hypercrates I get at least one of everything. This is highly unusual based on my experience.

 

So basically I spent $76 and didn't get what I wanted.

 

I understand it's a "gamble" but this drop rate seems really low. And the chance cubes are way too much "chance" for the money. Opening the packs is fun, but not so fun when you don't get what you want in a hypercrate. $35 should be enough money to give me an excellent chance to get me what I want and $76 should be a sure thing.

 

Suggestions:

 

The chance cubes are a good idea in theory. They should rotate shipments and those shipments should be announced beforehand - "this pack's chance cubes contain items from shipment X" or "shipments X & Y".

 

Super rare items should have a higher drop rate when we're spending this kind of money. Or perhaps be released for direct CM purchase at a later date. Or both.

 

I've been a CE sub since pre-order with 3 CE's. I currently maintain 4 subs for myself and my family.

 

Love the game. "Wanna get in the hot tub! Gonna make me sweat, ah!" - Eddie Murphy

[Dyhanna]

I hate the 'chance cubes' idea. New packs should only contain the stuff from that pack. If we want older stuff, we can get the Grand Packs... bring those back too. Seriously, I'm usually a "Bioware cheerleader" but even this has me saying *** BIOFAIL.

 

Kinda ditto. I like the chance cubes but I think they should be extra items and not replace items as found in the previous system. Also, as I stated above, the cubes should be restricted to one, two or max 3, specific shipments.

 

I too am usually a BW cheerleader. My glass is always half (or 3/4) full.

[roadez]

I like the chance cubes and I think it's something that is fun that was added to the crates.

[HiveOracle]

Personaly, i do like the chance cubes, although I have to agree they contain just to much companion customizations and crystals for comfort 8happy with the amount of dies though, since in my experience most armors need one anyway, so we need a lot of them ).

But I also agree that there should be a limit to the amount of chance cubes in a crate. Either adjust the amoutn of chance cube drops downward, or just make it a fixed '1 cube 1 pack item' thing. That way you'll end up with a decent mix imo.

And looking at the price of an honestly not all that exceptional saber on the GTN and direct sales, it indicates it's stupendously rare considering that at least 95% of the playerbase are probably not even intending to get one if they got to spend more than ahundred k or so.

 

On a side note: Give us Weapon outfit design slots to be able to adjust how the weapon looks. can't be all that hard after all, and woudl make both crystals and a lot of the mor ecommon weapons a lot better/more worthy since we had a use for mroe than 1, maximum 2 per character.

[jediloki]

The new Anarchist crates are a total disappointment.

 

The newest crates should be for the newest items. Nothing else.

 

Want to sell a hyper crate of chance cubes for people to take chances of getting some of the older useful items...sure, make it a separate hyper crate pack. But remove them from the main pack completely.

 

OR, as an alternative, have a THIRD pack slot that is ONLY for the chance cube to possibly drop. You allocate the first two slots for the new pack items, and then there is a chance on the third slot (not guaranteed drop) that you get a chance cube with the older items. If the RNG says no chance cube in third slot, nothing additional drops.

[Thagrynor]

That's OK, he didn't do the math either. He just counted stuff. Let's go back to basic statistics and probabilities. 244 samples is *nothing* when assessing distributions like this. The individual items, especially "what's the probability of item X dropping" when X is a cube-only item among thousands, have a lower probability per sample-point than 1/244.

 

So, obviously, by an inversion of the pigeon-hole principle, it is clear that there will some items (thousands, even) which are not received in a 244-chance sample. There aren't enough chances to get them all. Duplicates, on the other hand, are startlingly common. Someone mentioned 2000 different items in total -> the math behind the "birthday paradox" says that the probability of a duplicate (i.e. that at least one item appears more than once) exceeds 50% when the number of chances is around 1.2 times the square root of the number of items. (This "paradox" is illusory, and named for the fact that if you have groups of people, you'll find that a group of just 23 people has a > 50% chance of at least one shared birthday.) 244 is way far more than 1.2 times the square root of 2000, so the chance of duplicates is correspondingly higher than 50%.

 

Then we move on to sample sizes and "margin of error", and we conclude that 244 is too small a number of samples to say much of anything for sure. Consider all those political polls that asked 1500 people and have a margin of error quoted at 3-4 percent. That margin of error is a function of the total population (i.e. all the samples taken by everyone), the sample size (the 244 figure), the probabilities of each possible outcome, and a measure of how confident we want to be that the true probabilities will be inside the calculated margin of error.

 

It's rare for political pollers to publish their confidence levels, but it's usually 95% or 99% - the higher it is, the wider the margin of error will be. They do publish their sample sizes, and they are normally much higher than just 244. The higher the sample size, the narrower the margin of error will be. 244 samples is fairly low, and will produce a correspondingly wide margin of error.

 

There are enough different possible items that we would need to make tens or even hundreds of thousands of samples to get a decent idea of what the probabilities, and yet the birthday paradox speaks of getting duplicates at 50+% probability with a number of samples in the mid fifties (assuming 2000 different items).

 

The main problem with your entire argument is that you assume there is an equal chance of each of those 2000 items dropping. In actuality, the packs have a chance to drop 1 of about 25 or so items, one of those items being a cube that has an independent chance of being about 2000 items. The problem is in getting one of the 25 items, not one of the 2000 items. So, in reality, using your own model, 1.2 * sqt(25) should come to exactly 7. 7 packs should give me a better than 50% chance of obtaining any 1 given item, assuming equal probability of dropping. And this also has the problem of assuming there is not a weighted probability, but rather a uniform probability of any 1 item dropping.

[Ratajack]

The main problem with your entire argument is that you assume there is an equal chance of each of those 2000 items dropping. In actuality, the packs have a chance to drop 1 of about 25 or so items, one of those items being a cube that has an independent chance of being about 2000 items. The problem is in getting one of the 25 items, not one of the 2000 items. So, in reality, using your own model, 1.2 * sqt(25) should come to exactly 7. 7 packs should give me a better than 50% chance of obtaining any 1 given item, assuming equal probability of dropping. And this also has the problem of assuming there is not a weighted probability, but rather a uniform probability of any 1 item dropping.

 

No. Statistics does NOT assume an equal chance of each item dropping.

 

Each item has a given drop rate, although not all items have the same drop rate. That drop rate is exactly the same for each possible slot in every pack you open. If an item has a 1% drop rate, every pack you open will have a 1% chance of containing that item in each of the two possible slots, whether that be the first pack or the 600th pack.

 

The same holds true for an item with a 75% drop rate, such as the grand chance cubes seem to have. Each pack you open will have a 75% chance of containing a grand chance cube in each of the two possible slots.

 

The grand chance cubes have their own drop table with the items on that drop table having their own drop rates.

[Kirani_Borel]

I posted this in a different thread, but I think it is more appropriate here.

 

I agree the new pack is complete garbage. I bought the set of 30. When I opened them I got about 40 of the chance packs. I did not get a single piece of armor from any of them. Almost everything I got was useless: emotes, titles (the same ones multiple times), pets, a few of the very cheap weapons and mounts. I will never buy another pack if this is the trash they provide. I spent over 5,000 cartel coins for this! I doubt all of it combined will sell for 50,000 credits at the broker.

[ZanyaCross]

I really hope this blows up in their faces. They've been tweeking the packs since 4.0 to make the items people want harder and harder to get in the hopes they will buy more and more packs, and it's been working-- for awhile.

 

Now it looks like that bubble is finally going to burst. My only complaint is that it took this long, but the whales wanted their shinies in the previous packs and kept telling EA and Bioware that they would spend more and more, so here we are now.

[ZanyaCross]

I posted this in a different thread, but I think it is more appropriate here.

 

I agree the new pack is complete garbage. I bought the set of 30. When I opened them I got about 40 of the chance packs. I did not get a single piece of armor from any of them. Almost everything I got was useless: emotes, titles (the same ones multiple times), pets, a few of the very cheap weapons and mounts. I will never buy another pack if this is the trash they provide. I spent over 5,000 cartel coins for this! I doubt all of it combined will sell for 50,000 credits at the broker.

 

Someone on Dulfy posted a screencap of them getting a 1c chair from a coob. I laughed when I saw that, but it was really sad.

Edited February 14, 2016 by ZanyaCross

[AcousticGod]

Ummm...perhaps someone needs to tell Bioware they are DOING IT WRONG.

 

You don't get people hooked if those people don't get the "payoff" for their comp coins.

 

It's like a liquor company trying to get you hooked on booze that only gets you drunk 0.09% of the time.

 

These guys at Bioware really don't get it. Perhaps hiring some folks from Jack Daniels or Smirnoff might show them how to get people hooked on something.

 

(Yes, I am well aware of how politically incorrect that statement was.)

 

as a half puerto rican half irish man, i approve this message XD hahaha

[SteveTheCynic]

The main problem with your entire argument is that you assume there is an equal chance of each of those 2000 items dropping. In actuality, the packs have a chance to drop 1 of about 25 or so items, one of those items being a cube that has an independent chance of being about 2000 items. The problem is in getting one of the 25 items, not one of the 2000 items. So, in reality, using your own model, 1.2 * sqt(25) should come to exactly 7. 7 packs should give me a better than 50% chance of obtaining any 1 given item, assuming equal probability of dropping. And this also has the problem of assuming there is not a weighted probability, but rather a uniform probability of any 1 item dropping.

The 1.2 times square root thing is the chance of having at least one *duplicate* - in the origin of the name of the birthday paradox, that means the chance that two people in the group have the same birthday. AND, it is 1.2 times the square root of the number of trials (the number of people in the group), not 1.2 times the square root of the reciprocal of the probability.

 

And yes, the 1.2 times square root thing is based on all outcomes being equally likely. Violating that assumption complicates the calculation immensely, but (a) doesn't make it impossible, and (b) doesn't change the result that the probability of duplicates becomes close to 100% much sooner than people think.

 

You are trying to calculate the "expected value" of the number of trials required to get just one of the desired item, also known as the "Number of trials to first success". Following a tangly piece of mathematics, that turns out to be 1/p, where p is the probability on one trial. In the context of the Anarchist's pack, opening one pack gives two "per-slot" trials whose probability of success can be calculated easily, for a gold item, as:

 

P(the gold item I want) = P(get it directly) + P(get it via a cube)

 

P(get it directly) = P(get any gold item) / the number of different AP gold items). Assumes all gold items are equally likely.

P(get it via a cube) = P(get a cube)*P(get any gold item) / the number of different all-packs gold items.

 

P(get any gold item) is about 5%, I think, and certainly was for older packs.

P(get a cube) is about 85%

The number of all-packs (meaning counted across all packs) gold items? Not sure, but it's large, especially as many armour sets come in three boxes, and so count for this analysis as three different items. (Intuitively, this is why people end up with relatively large amounts of Skotia boxes and fewer of equal-rarity weapons - there three times as many kinds of Skotia box as there are of Silver Weapon Three.) That tends to make P(get it via a cube) low, probably low enough that we can ignore it.

Similarly, the number of different AP (Anarchist's Pack) gold items is something we'd have to count, and while it is much smaller than the number of all-packs gold items, it still serves to make the probability of ==> that gold item low.

 

However, the key thing to take away from this analysis is that "expected value" refers to what is in essence the average result, averaged over all attempts. For a correctly-balanced six-sided die, the probability of rolling a 1 is 1/6, so the expected number of trials to first success (where "success" means rolling a 1) is 1/(1/6), or 6. This doesn't mean that your first 1 comes on the sixth roll every time you begin a series of rolls. What it means is that if you average the number of rolls required over all series that you try, you get 6 rolls. On some series, you get the 1 on the first, or the second, or..., and on some it might even take 600 rolls, or a million.

 

In the context, that means that opening 120 packs and getting (or not getting) a particular new-pack item is far from sufficient to say much of anything about the drop rate, as I said previously. It just isn't enough packs.

 

(For reference, I got an Unstable sabre somewhere close to pack 20 of crate 4. There's one - not mine - listed on the GTN of my server, for 23.5 million. Foolishly, I bought the crate in question on a Trooper. )