Understanding the RNG from a mathematical perspective
|
Many things in PoE are driven by a RNG, this we all know.
However the only thing people ever say about it is "RNG is RNG" which is not particularly useful and is glossing over important mathematical concepts related to statistics. Things like expected value, variance, standard deviation, convergence, law of large numbers etc. Every RNG related thing in the game is bound to a specific expected range. When you kill a mob and the RNG's fire, they are only random to a point. I mean after all how many wisdom scrolls drop vs. exalted orbs? It's not random chance that you seem more scrolls, the RNG is setup such that the scrolls drop rate is signficantly higher. So let's say hypothetically the odds of a chaos orb dropping off any given mob are 1 in 10,000 (I'm totally just making this number up for the example). This does NOT mean you will get a chaos orb every 10,000 drops. What it means is that each individual monster you kill there is a 1:10,0000 chance you will get a chaos orb. So after 10,000 kills your expected number of drops is 1. A lot of people might have 1, but some people might have 0, others might have 5. That's related to sample size. Over a small sample size the actual numbers can vary dramatically from the expected numbers. The longer you play, the more your actual numbers will converge to the expected numbers, same story for everyone else. Given that the drop rates for some of the rare currencies are pretty low, a large sample size is, well, large. This is why things like the Orb of Chance/Fusing community logs are useful. Given a large enough sample size one can reverse engineer the actual rate. It would be useful if GGG posted the currency drop rates so we'd have an idea of what we can reasonably expect in terms of drops. AKA if a chaos orb was indeed 1:10,000 and I could kill 2,000 mobs an hour I would know that on average I will get a chaos orb drop about every 5 hours (over the long run). Last edited by Zaqwert#4447 on Mar 19, 2013, 1:17:53 PM
|
|
|
very informative post. there seems to be alot of people here who have zero understanding of what probability is and how it works, maybe this will clear some of it up
| |
|
So you could say RNG is a game's "luck" system.
We are playing the lottery. [Suggestion] How to end Desync: [url]http://www.pathofexile.com/forum/view-thread/335390[/url]
|
|
|
+1
IGN: Iluminado
|
|
|
Nice post OP. To be clear, your stated chances for anything to happen are always mathematically more accurate only in the long term, with "long term" a somewhat arbitrary measure but usually considered to be after a minimum of 30 runs. (30 in often considered to be a statistically minimum sample size) So if your chance of rolling a unique with a chance orb is 1 in 500, this means that in the short term you may roll that unique in your first go or not until your 3000'th go, but over a long series of rolls (around 15,000 in this case) you would start to average out at 1 unique per 500 chance orbs used (assuming that 1:500 was a statistically valid ratio). I am still over simplifying but this is not a million miles away. . |
|
"exactly chances of rolling a perfect or near perfect item are veeeery slim therefor; rolling the perfect items = unmatched endorphine and serotonine release thats why RNG is king however from a large numbers of posts on these forums its easy to tell that alot of players dont like or/and understand rng and what the charm is. they are used to a more linear form of progression, and thereby they get turned off by the cruel nature of rng however once youre addicted to rng you really dont want it any other way, and i hope that poe stays pretty much all rng. so next time you blow 700 fusings trying to get a 6L and are about to go make a huge rage post, read some on probability and rng, and try to remember the lucky breaks you must have caught somewhere along the way. Last edited by fsg#3483 on Mar 19, 2013, 1:45:10 PM
| |
|
RNG - Raging Nuance God
With all intelligence, insanity becomes the backdrop of the good man, and will sooner become what usurps the good man.
If you tell a truth that people don't want to hear, the truth will imprison you. What's the point of knowing the world, if the world never talks back and others don't listen? |
|
"interesting. probability is kind of the black sheep of the math family, since from a deterministic point of view there really are no such thing as probability or randomness | |
|
Indeed.
Now calculate the odds of crafting a Belt that would be useful to a CI character. Now you see the flaw. Belts are Melee favoured items. The only attribute enhancement on a belt is Strength. No Dex, No Int. You can't get Evasion on a belt. So, from the 17 available mods on a belt, only 4 of them are useful to a CI user (as in excitedly demanded). Of those 4 mods, you then have a multiplex of level dependent stats. So the chances of a CI user ever finding a decent CI belt are radically different to those of a HP physical damage user. Why does this effect opinions on RNG? Because is creates obscured understanding of how effective RNG is. One character is having to look so much further than another. And this extends to the RNG of everything in the game. Because the game is in Beta, and is amended and altered on the fly, the balance of every item/build is radically different to the next and requires completely different RNG to be on a level RNG playingfield. The issue about the orb drop-rate is a completely different topic, and one that has been around for a long time. The main problems occurring the more orbs become currency than crafting items. In Closed Beta people tended to use their orbs more, because in Closed Beta trading was a pain in the neck, with no trade screens and no trade forums, and because everything got wiped on a regular basis anyway. Orbs were used in a 'might as well give it a shot' kind of way. In Open Beta, the orbs are more-so currency than ever. So people use them with a mindset of is it worth it, as in, what else could I buy with it. So, to then apply full RNG to currency is to randomly select which players will be poorer than others from a gameplay perspective while at the same time permitting many to become crazy rich from non-gameplay non-RNG aspects. Ergo: Currency drop rates don't actually matter, however they are RNG'd, to the majority, but to the person who simply runs through the levels, they are appalling. What really matters with the RNG in the game is how different builds/items get screwed by an uneven RNG, even though the RNG base stats (whatever they are) are the same for everything. Another example of this being the socket colors on items effecting the RNG of various builds, for example. |
|
|
If i understand the way magic drops work and while you are talking RNG.... you only have a 20% chance max to get a rare drop no matter what your rarity is correct? The purpose of rarity is to get white drops to be blue correct? The more blue drops you have, the better chance to get the 1 in 5 rare if i understand the mechanics behind magic/rare/unique drops. Please school me if im wrong as i would like to know the mechanics behind the drops.
Best results as Ranger rank 29 (1 Hr Lethal TI Solo (S02E148)) rank 40 (2 Hr Fixed Seed Solo (S02E249)) rank 49 (1 Hour BLAMT Solo (S02E258)) Last edited by Zazzaro703#3484 on Mar 19, 2013, 2:06:41 PM
|
|
