How is it consistently ~95%TT returns over large data sets?

AckerZ

Provider
Joined
Jun 5, 2020
Posts
124
Dsiclaimer: I am new to understanding how RNGs/algorithms work.

The game is designed to look like the loots you get are random and to a degree they definitely are, however the game devs have stated clearly that over a large enough number of loot instances the TT return trends towards 95%-97%.

As someone with absolutely zero knowledge in RNG systems/Looting Algorithms would someone be able to give a laymans explanation of how such a seemingly paradoxical system can work?

Its random loot but over a large enough data set it isn't? Similar to slot machines that state they give a 97% return over a large enough data set how is this organised in the real world and entropia? Its a fascinating system whatever it is...
 

kingofaces

Old Alpha
Joined
Jun 9, 2013
Posts
701
Location
US
Avatar Name
Tony KingofAces Hans
Sampling statistics is the term you want to look up as well as the "law of large numbers".

Every population has a true average for various traits like height, etc. There's also normal variation around that average (i.e., standard deviation).

If you survey a few people what their height is, the average you get is likely going to be way off. As you keep sampling more people, your average is going to jump around, maybe well below the true average, and then way above. You don't know what the true average is, but as you get to larger and and larger sample sizes, the variation of that average is going to decrease.

Say you surveyed 3,000 people to get an average height though. Then repeated it again, again, etc. up to 100 times (300,000 people total). Each of the averages for those 3,000 are still going to be different and just due to randomness, somewhere around 5-10% of those averages are going to be very different than the true average height (sort of a false positive). In statistics speak whenever we present an average in scientific articles, we almost always include what is called a confidence interval saying that 95 out of 100 times, the average would be expected to fall within a certain range. So if one set had a slightly different average than the other, we'd say they are not significantly different because that difference was just due to chance and not some underlying factor in the population.
 
Last edited:

Duhword

Old
Joined
Nov 30, 2016
Posts
64
Location
USA
Avatar Name
John Duhword Boy
Got it, although i must say, that is probably misleading. There is also other official statements that say the average loot expected is 92% overall. I myself, returned 94%ish in 2019. That statement there is only for a 16day window and only for avatars cycling roughly 40k-100k per month in that small window.
 

Legends

Stalker
Joined
Dec 31, 2006
Posts
1,719
Location
Twin Peaks mall, 2nd floor
Society
Dirty Dingos
Avatar Name
Inherent Marxus Legends
It's a very complex topic, and very contentious. Almost anyone you talk to is bound to have their own opinion of how the loot system actually works.

One thing is for sure, there is a tremendous amount of randomity entered into the outcome, randomness on top of randomness. If you stop and think about it for a second, just all within one hunt, you're liable to get these factors:

- the amount of damage you deal with each attack is random
- the amount of damage the mob deals to you is random
- different maturities with different hp and dmg
- sometimes different mobs are mixed in the same spawn
- chance that your shot misses the mob
- or that the mob evades
- the mob's attack misses you
- you evade the mob's attack
- the mob critical hits you
- you critical hit the mob

All of this randomness stacks up in a way that the exact same mob that normally costs you 4 ped to kill could easily cost 6 or 7 ped given enough misses.

I believe the system is designed with extra randomness redundancy for a reason, i.e. to try to make it as unpredictable as possible.

I think that at the crafting machine, patterns can be seen a little bit better and it can 'almost' be seen what is going on, as every single click is exactly the same cost, the same input.

For example, one thing I have noticed is that the multipliers are actually quite ordinary numbers, i.e. x2.5, x5, x10, x15, x25, x50... but there is this +/-10% to it, another randomity factor that has been entered into it, just so that the actual loot amount will always be different instead of the exact same number every time. This makes it appear very dynamic, where in fact it's just an RNG on top of another RNG.

Add to that the possibility of looting rares and now you have an almost complete picture. Any activity can have 4 or 5 'random chance of' involved in a single loot event.

It has been asked more than once if MA had 'personal loot pools' programmed into it. Their official answer has always been 'No, no personal loot pools'. This could just be a matter of semantics, but then again we have seen people hit it big before, like 100k big, without ever depositing that much into the game, so it's likely they are telling the truth.
 
Last edited:

Ferial

Guardian
Joined
Jul 4, 2012
Posts
298
Location
Sweden
Avatar Name
Chris Ferial Book
Similar to slot machines that state they give a 97% return over a large enough data set how is this organised in the real world and entropia?
it´s all in the multipliers.
let´s say 50% of the loot instances are set to loot 50% of the wager cost, we now have 25% RTP.
let´s say next to those 50% we have 25% of the loot instances set to loot 75% of the cost to play. you know have 43,75%RTP
lets say next to those, you have 20% of the loot instances set to loot 100% of the wager cost, we now have 63.75% RTP
let´s say you have 2,5% return 250% of the wager cost. we now have 70% rtp
let´s say you have 1% return 5000% of the wagers cost. we now have 75% rtp.
let´s say you have 1% return 10000% of the wagers cost, we now have 85% rtp.
let´s say the remainder 0.5% return 20000% of the wagers cost we now have 95% rtp.

this way, MA never pays out more than 95% of their income, the average return is 95% over the entire playerbase. but the returns can vary quite vastly within the playerbase.
 

Alina

Elite
Joined
Mar 22, 2005
Posts
2,800
Location
Bucharest
Society
freeeeee
Avatar Name
Stoica Alina Alina
97-98% few mil peds with over 83 eff some rounds with 86-90 eff
 

AckerZ

Provider
Joined
Jun 5, 2020
Posts
124
Is there some system that safe guards against the improbable but possible outcome of multiple 20000% payouts? Or are they banking on it being highly unlikely aswell?
 

Svarog

Slayer
Joined
Dec 11, 2006
Posts
9,310
Is there some system that safe guards against the improbable but possible outcome of multiple 20000% payouts? Or are they banking on it being highly unlikely aswell?
They aren't risking anything. A tiny portion of decay never goes back to the lootpool, this is their revenue (which they share with the CLDs holders) and we fight for the rest. The max multiplier on loot is believed to be 10000x from the cost, but when you hit it, you harm only the fellow players, lowering the loot on this mob or this area or this planet (whichever way you slice it on your theories). It's the game of musical chairs that MA doesn't participate in.
 

kingofaces

Old Alpha
Joined
Jun 9, 2013
Posts
701
Location
US
Avatar Name
Tony KingofAces Hans
Is there some system that safe guards against the improbable but possible outcome of multiple 20000% payouts? Or are they banking on it being highly unlikely aswell?
You actually just asked about it earlier with respect to large sample size or “law of large numbers“ again. You aren’t realistically going to get that many big payouts, and everything will average out to whatever MA had it set at with some predictable variation around it.
 

GoNi

Slayer
Joined
Feb 2, 2007
Posts
7,564
Location
Calypso, Sakura City
Society
freelancer
Avatar Name
Gloria GoNi Nicestyle
They never say you can expect that on average, its just an information how it actually was in that period of time.

Beside that giving that average over all players, also includes the players that are way above and those way below.

To OP:

There is no gurantee to get 92% -- its still somehow random, can be worse or can be better.
Skill + Gear matters how good you will perform and where your stand at the end.

And its just the pure TT return. There is more than just TT, smart players go for markup!
 

Geo

Old Alpha
Joined
Jun 2, 2005
Posts
894
Location
Space
Society
Victory
Avatar Name
George Geo Clone
The way I look at it, the mobs give back, lets say 95%~ after a large enough sample size. With MA taking their 5%.
Players who hunt said mob will be above or below if hunting there during an up or down swing making the difference in % returned.
Your gear and skills determine what % of that % return you can obtain.
 

Jhereg

Stalker
Joined
Dec 13, 2005
Posts
1,581
Society
Rangers
Avatar Name
Feng Huan SecretAznMan Zho
I'll try to give your initial question a shot. I would recommend an online tutorial into probabilities and statistics if this area interests you. I find the subject both fun and useful.

Let take a coin toss for example. There are two different outcomes, and if the coin is fair, the outcome each have equal chance of happening. Therefore, you have:

Heads:Tails = 1:1 or another way of saying it is you have a 50% chance of the coin landing on heads and 50% chance of it landing on tails. Each coin toss is independent, that is, the previous coin toss does not affect the subsequent ones.

Now, we can assign values to the heads vs tails. Say, if the coin lands on heads, you get $1, but if the coin lands on tails, you get $0.
Let's say, you have to pay $0.50 for each coin toss.

Since the odds of head and tails are equal (50% each), out of many,many,many coin tosses, you expect to have 1 heads to every 1 tails.
So your expected return is 50%*$1 + 50%*$0 = $0.50, for every coin toss. (This is what some call the expected value, or EV).

Since the cost of each coin toss is the same as the expected return (both are $0.50), after many, many tosses, you would expect your input to equal your output. Therefore, you would have a return of 100%, meaning you would never lose money, and the house would never gain any money.

However, on single independent events, you could gain 2x your bet, or lose your entire bet.

In this simplistic version, how might you get something like 96% return? The simplest way to see this is to increase the cost of a coin toss.

If we increase the cost of the coin toss to $0.52, and keep the payout the same, we would end up $0.50 returned for every coin toss, and we pay $0.52 for every coin toss. This results in 0.50/0.52 = 96.2%. So on average, your return would be 96.2% of your input. The more coin tosses you make, the closer to this number you get.

Of course, MA's system have more possible outcomes, but the concept is the same. Each outcome has some probability, and you can calculate an expected return value based on the sum of the probability multiplied by the expected return value for each possible outcome. The overall return % approaches the expected number as the number of loot events become very large.

Your other question, how does MA ensure that they do not pay out more than the loot pool is a bit more complex, but here's how I would explain it simply. This is my guess of how their system works and NOT OFFICIAL! So take all this with a grain of salt and don't accept at face value disclaimer disclaimer :D:

1) Based on the probability of the largest multiplier, MA can estimate how many extremely large events can occur within a certain number of events. One 1000x multipler is very likely to occur in 1,000,000 events, but n=1,000,000 1000x multiplers is very unlikely to happen in 1,000,000 events, as a very extreme example.

2) MA just needs to make sure they have sufficient funding to cover these bigger payouts.

3) MA caps the largest multipliers. This prevents extremely large swings in the 'pool', or volatility, and ensures MA always has sufficient cash to cover.

I hope these help to answer your question.
 

AckerZ

Provider
Joined
Jun 5, 2020
Posts
124
They aren't risking anything. A tiny portion of decay never goes back to the lootpool, this is their revenue (which they share with the CLDs holders) and we fight for the rest. The max multiplier on loot is believed to be 10000x from the cost, but when you hit it, you harm only the fellow players, lowering the loot on this mob or this area or this planet (whichever way you slice it on your theories). It's the game of musical chairs that MA doesn't participate in.

This really answered my question, thanks svarog and others.
 
  • Like
Reactions: Geo

AckerZ

Provider
Joined
Jun 5, 2020
Posts
124
3) MA caps the largest multipliers. This prevents extremely large swings in the 'pool', or volatility, and ensures MA always has sufficient cash to cover.
I think this was the area that was throwing me off the most, it only really makes sense that MA prevents certain large loots from occurring too often, so as much as they say its random it would be random but aslong as there are funds from MA available for it to be so?

Obviously this is all pure speculation but this does make the most sense.
 

Alukat123

Stalker
Joined
Jun 20, 2017
Posts
1,696
Avatar Name
Son Alukat Goku
The max multiplier on loot is believed to be 10000x from the cost, but when you hit it, you harm only the fellow players, lowering the loot on this mob or this area or this planet (whichever way you slice it on your theories). It's the game of musical chairs that MA doesn't participate in.
something like that would be a silly thing to do.
Punishing many players, because one got lucky would be a good way to loose your playerbase
 

Svarog

Slayer
Joined
Dec 11, 2006
Posts
9,310
something like that would be a silly thing to do.
Punishing many players, because one got lucky would be a good way to loose your playerbase
But everyone takes turns being the statue and the pigeon, so it's fair in the long term :)
 

atomicstorm

Marauder
Joined
Aug 21, 2013
Posts
7,356
Location
Tennessee, USA
Society
Entropialoot.com
Avatar Name
MeLoveYou LongTime FiveDolla
Since the odds of head and tails are equal (50% each), out of many,many,many coin tosses, you expect to have 1 heads to every 1 tails.
So your expected return is 50%*$1 + 50%*$0 = $0.50, for every coin toss. (This is what some call the expected value, or EV).
This was always the fun part of gaming probability - because a coin flip is actually not a 50-50 proposition. Depending on the coin, it will depend on which side is up first because - for example, the quarter - the head is weighted heavier and has a slight edge. :) With roulette, the wear of the roulette table and the throw skill of the dealer actually can cause certain numbers to occur within a block far more frequently - which is why casinos have to change them out regularly.

Not really relevant to this, but fun fact nonetheless. :)

For everything else, do some simulations on law of large numbers. It is really all you ever really need to understand for games like this - or anything that has built in edges.
 

Menace

Elite
Joined
Sep 20, 2008
Posts
3,274
Location
Niagara Falls, Canada
Society
SECRET
Avatar Name
Rob Menace Klok
In order to give Menace 100k ATH Mind Ark has to take a 100 Ped from 1000. Then in order to give you a 5k HOF they have to take an additional 5 ped from that thousand. Now 1000 people are complaining about losing so that MA can boast about 95% return rate and one person can party like a rock star and make the 1000 losers want to gamble harder so they too can get 100k.
 

Menace

Elite
Joined
Sep 20, 2008
Posts
3,274
Location
Niagara Falls, Canada
Society
SECRET
Avatar Name
Rob Menace Klok
something like that would be a silly thing to do.
Punishing many players, because one got lucky would be a good way to loose your playerbase
Which is EXACTY what happened.
 
Top