Something is really off with Keys drop rate

So I run for keys daily and noticed a weird thing - sometimes I can literally do ~200 runs one after another with 0 to very few keys. And sometimes I just freshly login and they drop like crazy, double from summoner, one from nih in a single run and then almost every first/second run there is at least one

Key, then I take a break and next day again it's either many keys, or nothing for hours. Please don't tell me it's RNG, I know math and this is not how probability theory and normal distribution work. I also don't believe in conspiracy and my main theory behind that is their random generator uses a really predictable source of randomness or there is a bug in how it's coded. What do you think? Does it happen to others?

For me it's pretty clear that there are times where it doesn't matter how much you farm, you don't get anything valuable.

Then suddenly you pop the nth run and you start getting a noticeable better loot, even if you don't get anything useful you can clearly see the difference.
It keeps going this way for a while until you start getting only whites and blues, once again.

It's like if there is some mechanism that flags and unflags your account making it able to access to certain type of loot.

Yup, farming keys I had the same issue. Sometimes I had to give up and wait till the next day. Some others I will farm 2 3x3 sets back to back. Its just RNG

I'm skeptical, why not do the math then? What's ~200 really? What's "0 to few" really? My worst streak was like 30 runs and that's still like a sadly solid 7% chance of it happening on p8. If you play enough you'll get that and worse, as well as the really good streaks.

Its just the RNG gods man. I did about 50-60 full key runs of all 3 and only got 1 single Hate key. Then i do 20

Nihlathak runs and get 4 D keys. /shrug

Bounder wrote: 3 years ago

I'm skeptical, why not do the math then? What's ~200 really? What's "0 to few" really? My worst streak was like 30 runs and that's still like a sadly solid 7% chance of it happening on p8. If you play enough you'll get that and worse, as well as the really good streaks.

Your math is correct for 30 runs. And that's true, I didn't count my runs/drops, I would do that. It was obvious enough for me, that drop rates are off, because 8% drop rate of the

Key is not that small, it's actually pretty high, (1 - 0.92^8) = 0.51 = 51% that at least 1

Key would drop within only 8 runs. 8 runs is like what? 10 minutes of my time. I got easily 3-4 hours without keys. My concern here also is a distribution, that after hours after no drops I suddenly get godlike drops, it's like if now I would flip 10 tails in a row, go eat and after lunch would flip 10 heads. In the end it was 50/50 heads and tails, but the distribution is very improbable.

kaikossades wrote: 3 years ago

Bounder wrote: 3 years ago

I'm skeptical, why not do the math then? What's ~200 really? What's "0 to few" really? My worst streak was like 30 runs and that's still like a sadly solid 7% chance of it happening on p8. If you play enough you'll get that and worse, as well as the really good streaks.

Your math is correct for 30 runs. And that's true, I didn't count my runs/drops, I would do that. It was obvious enough for me, that drop rates are off, because 8% drop rate of the

Key is not that small, it's actually pretty high, (1 - 0.92^8) = 0.51 = 51% that at least 1

Key would drop within only 8 runs. 8 runs is like what? 10 minutes of my time. I got easily 3-4 hours without keys. My concern here also is a distribution, that after hours after no drops I suddenly get godlike drops, it's like if now I would flip 10 tails in a row, go eat and after lunch would flip 10 heads. In the end it was 50/50 heads and tails, but the distribution is very improbable.

That's the thing, those little numbers matter a lot. No keys in 150 runs is almost impossible. But if I assume "few" = 5, then now you're looking at sum from x=1 to 5 [(150 choose x)(.08)^x(.92)^(150-x)] = 1%, suddenly not so crazy right? [hopefully I still remember math]

If you're consistently seeing that then a lot immediately that's weird, but we'd also have to check how much is a lot really over the same amount of runs. A lot in like 20 runs after which you'll go trading for other keys is not as convincing that something is strange. Hell even "normal"/slightly unlikely drops feel like a lot to me after a bad run.

Everyone is guessing here.
Unless you had access to the source code you don't have a way to guarantee anything you are saying is actually true.
I just wanted to mention it because sometimes people sound ridiculous stating things they don't have a way to know.

MadBarbosa wrote: 3 years ago

Everyone is guessing here.
Unless you had access to the source code you don't have a way to guarantee anything you are saying is actually true.
I just wanted to mention it because sometimes people sound ridiculous stating things they don't have a way to know.

Most drop rates known today probably come from the code. Not the source, but the binary that skilled enough people can read. Of course there could technically be a difference in bnet since drops are server based there, but if the offline math checks out for online then drop rates are pretty well understood.

edit: not even binary (not that that's impossible), looks like the rates are in some game text file? So yeah if blizz isn't trolling (or has some bug, which is also possible) most stories people tell about their "crazy" drops/lack thereof tend to fall within expectations of RNG if the math's done properly. The most common one I hear is "well I did 5k meph runs" and well yeah...5000 monsters is very little.

kaikossades wrote: 3 years ago

Bounder wrote: 3 years ago

I'm skeptical, why not do the math then? What's ~200 really? What's "0 to few" really? My worst streak was like 30 runs and that's still like a sadly solid 7% chance of it happening on p8. If you play enough you'll get that and worse, as well as the really good streaks.

it's like if now I would flip 10 tails in a row, go eat and after lunch would flip 10 heads. In the end it was 50/50 heads and tails, but the distribution is very improbable.

The odds of this happening are around 1 in a million.

Surely the coin is flawed, and also it was changed for a different flawed coin during the lunch, right?

I mean, 1 out of 1 million, nothing with such low probability should ever happen... Well, other than lottery I guess. Which is won every single day, despite having similarly low odds.

How does it happen? Because many players roll on that 1 out of 1 million odds. And one of them gets it.

Imagine if 10,000 players flips these 20 coins, every day for a month; Suddenly, there's 25% chance that one of them gets that exact sequence you described.

The thing with probability, is that people don't see the big picture.

I've played many games in which people said similar things; Poker being the main one (literally 75% of the players think they're unlucky), but also games like Hearthstone when people say things like "I lost 20 games in a row that shouldn't be possible!"; Sure that's very unlikely if you're the only one playing, but when 1 million players play 10, 20, 30 games every day, someone is bound to lose 20 in a row quite regularly.

So, to come back to the OP's situation:

If streaks like that happen once: Shit happens, when thousands of people do the same thing over and over again. The only "weird" thing is that it happened to you.

If you think it happens all the time: Start recording your stats. People who find "anomalies" in RNG systems never have any track records, only vague estimates.

If you truly believe there's something off with the system, start recording your stats and it should be very easy to prove it.

But odds are, recording your stats will make you see that there's nothing wrong with the system, it's just normal lucky streaks followed by unlucky streaks that happen naturally with RNG.

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Everyone is guessing here.
Unless you had access to the source code you don't have a way to guarantee anything you are saying is actually true.
I just wanted to mention it because sometimes people sound ridiculous stating things they don't have a way to know.

Most drop rates known today probably come from the code. Not the source, but the binary that skilled enough people can read. Of course there could technically be a difference in bnet since drops are server based there, but if the offline math checks out for online then drop rates are pretty well understood.

Not sure what you mean, source and code are probably the same thing, context aside.

And you can try to decompile binaries and make yourself an idea about what it does, but honestly, I don't see anyone here with the needed skills, or even the energies.
Fuck, I've been working with systems for the last 2 decades and I wouldn't spent a minute of my time in something like that.

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Everyone is guessing here.
Unless you had access to the source code you don't have a way to guarantee anything you are saying is actually true.
I just wanted to mention it because sometimes people sound ridiculous stating things they don't have a way to know.

Most drop rates known today probably come from the code. Not the source, but the binary that skilled enough people can read. Of course there could technically be a difference in bnet since drops are server based there, but if the offline math checks out for online then drop rates are pretty well understood.

Not sure what you mean, source and code are probably the same thing, context aside.

And you can try to decompile binaries and make yourself an idea about what it does, but honestly, I don't see anyone here with the needed skills, or even the energies.
Fuck, I've been working with systems for the last 2 decades and I wouldn't spent a minute of my time in something like that.

How do you think people come up with d2 mods? Yes, people have the skills to do this. But I agree it's impressive! I do have a CS background and it would take me a while to get into that type of thing properly. But I do know enough to know it's feasible.

I'm pretty new to

Key farming, and spent a few days last week farming Countess. Took me about ~60 runs to get my 3 keys.

But now this week, I'm running Summoner. And GFDI, I am on my 4th day in a row, about 5-6 hours per night, getting in 5-10 runs per hour, and NO KEYS from him. It's painful and doesn't make any sense. Probably @ 100 runs now without a single

Key drop from him. Starting to think he doesn't drop them anymore!!

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Everyone is guessing here.
Unless you had access to the source code you don't have a way to guarantee anything you are saying is actually true.
I just wanted to mention it because sometimes people sound ridiculous stating things they don't have a way to know.

Most drop rates known today probably come from the code. Not the source, but the binary that skilled enough people can read. Of course there could technically be a difference in bnet since drops are server based there, but if the offline math checks out for online then drop rates are pretty well understood.

Not sure what you mean, source and code are probably the same thing, context aside.

And you can try to decompile binaries and make yourself an idea about what it does, but honestly, I don't see anyone here with the needed skills, or even the energies.
Fuck, I've been working with systems for the last 2 decades and I wouldn't spent a minute of my time in something like that.

How do you think people come up with d2 mods? Yes, people have the skills to do this. But I agree it's impressive! I do have a CS background and it would take me a while to get into that type of thing properly. But I do know enough to know it's feasible.

Sure, there are people with enough skills and time for it, but if you base your opinion in something you read about what other people did, you are still far from being sure about anything.
That's my point.

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Everyone is guessing here.
Unless you had access to the source code you don't have a way to guarantee anything you are saying is actually true.
I just wanted to mention it because sometimes people sound ridiculous stating things they don't have a way to know.

Most drop rates known today probably come from the code. Not the source, but the binary that skilled enough people can read. Of course there could technically be a difference in bnet since drops are server based there, but if the offline math checks out for online then drop rates are pretty well understood.

Not sure what you mean, source and code are probably the same thing, context aside.

And you can try to decompile binaries and make yourself an idea about what it does, but honestly, I don't see anyone here with the needed skills, or even the energies.
Fuck, I've been working with systems for the last 2 decades and I wouldn't spent a minute of my time in something like that.

How do you think people come up with d2 mods? Yes, people have the skills to do this. But I agree it's impressive! I do have a CS background and it would take me a while to get into that type of thing properly. But I do know enough to know it's feasible.

Sure, there are people with enough skills and time for it, but if you base your opinion in something you read about what other people did, you are still far from being sure about anything.
That's my point.

Now that's getting a bit ridiculous. Just because I don't do it myself it doesn't mean I'm so far from understanding it to reasonably believe it. Try spending a few minutes on a d2 modders forum, there's actually a lot of info and it makes sense from what I've been reading. What is your actual point? That we don't truly know d2 drop rates? So all the people who figured out the game files are wrong? Or you're saying the server does something special? If the latter I'll entertain the possibility. Otherwise, just no.

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

Most drop rates known today probably come from the code. Not the source, but the binary that skilled enough people can read. Of course there could technically be a difference in bnet since drops are server based there, but if the offline math checks out for online then drop rates are pretty well understood.

Not sure what you mean, source and code are probably the same thing, context aside.

And you can try to decompile binaries and make yourself an idea about what it does, but honestly, I don't see anyone here with the needed skills, or even the energies.
Fuck, I've been working with systems for the last 2 decades and I wouldn't spent a minute of my time in something like that.

How do you think people come up with d2 mods? Yes, people have the skills to do this. But I agree it's impressive! I do have a CS background and it would take me a while to get into that type of thing properly. But I do know enough to know it's feasible.

Sure, there are people with enough skills and time for it, but if you base your opinion in something you read about what other people did, you are still far from being sure about anything.
That's my point.

Now that's getting a bit ridiculous. Just because I don't do it myself it doesn't mean I'm so far from understanding it to reasonably believe it. Try spending a few minutes on a d2 modders forum, there's actually a lot of info and it makes sense from what I've been reading. What is your actual point? That we don't truly know d2 drop rates? So all the people who figured out the game files are wrong? Or you're saying the server does something special? If the latter I'll entertain the possibility. Otherwise, just no.

I clearly stated my point already, twice at least, sorry if you don't get it.
It's not surprising thought.

MadBarbosa wrote: 3 years ago

Everyone is guessing here.

No one is guessing, they are just giving their personal experiences about the range. We already know the average drop rates of every item in the game as well as the keys. But just because a

Key has a 1:14 chance of dropping doesn't mean you're guaranteed to get one per 14 runs, so people assume the drop rates are much lower when they are simply unlucky.

Silospen's Item Finder: https://dropcalc.silospen.com/item.php

MadBarbosa wrote: 3 years ago

For me it's pretty clear that there are times where it doesn't matter how much you farm, you don't get anything valuable.

Then suddenly you pop the nth run and you start getting a noticeable better loot, even if you don't get anything useful you can clearly see the difference.
It keeps going this way for a while until you start getting only whites and blues, once again.

It's like if there is some mechanism that flags and unflags your account making it able to access to certain type of loot.

I swear this is true! I get nothing of value for a couple weeks, then a bunch of stuff within an hour, then nothing for weeks, then 2 HR's in the same run, etc. It's all ups and downs. I've played a ton (10 hours a day for the first couple weeks), and I never found Andy's Visage. Found 3 in the last 2 days playing just a couple hours. In addition, it's always at the start of my play session when I get good loot.

A youtube video on randomness. TLDW People's concept of random is closer to homogenous. In a large sample set (the set of all keys not dropping before a

Key drops on bnet, because chance to drop is smaller than no drop) people discount how frequently long streaks occur by more than half. The problem isn't the keys not dropping; rather, the problem is your concept of randomness.

Spoiler

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Bounder wrote: 3 years ago

MadBarbosa wrote: 3 years ago

Not sure what you mean, source and code are probably the same thing, context aside.

And you can try to decompile binaries and make yourself an idea about what it does, but honestly, I don't see anyone here with the needed skills, or even the energies.
Fuck, I've been working with systems for the last 2 decades and I wouldn't spent a minute of my time in something like that.

How do you think people come up with d2 mods? Yes, people have the skills to do this. But I agree it's impressive! I do have a CS background and it would take me a while to get into that type of thing properly. But I do know enough to know it's feasible.

Sure, there are people with enough skills and time for it, but if you base your opinion in something you read about what other people did, you are still far from being sure about anything.
That's my point.

Now that's getting a bit ridiculous. Just because I don't do it myself it doesn't mean I'm so far from understanding it to reasonably believe it. Try spending a few minutes on a d2 modders forum, there's actually a lot of info and it makes sense from what I've been reading. What is your actual point? That we don't truly know d2 drop rates? So all the people who figured out the game files are wrong? Or you're saying the server does something special? If the latter I'll entertain the possibility. Otherwise, just no.

I clearly stated my point already, twice at least, sorry if you don't get it.
It's not surprising thought.

What was your point? That everyone should build their own nuclear reactor to believe nuclear powerplants are real? Or perhaps we should everyone just take part of moon-landing personally?

Good god, Diablo's code is not some government secret, it has been extracted some million times during 20 years. Those calculators are not guess-work based on someones 200 runs.

You made no point that has any sense to it.

While at it, lets calculate the odds of dropping 2*

Lo runes in single Chaos run on player1 game.