How to analyze loot

[falkao]

Let's try to formulate a possible relation between drop of globals and area or ava.

Loot value has a now more or less known distribution and it follows a random process. Waittimes between globals however do show a bias, i.e. some grouping. Moreover, it seems that globals tend to be higher when return was quite low before.


Hypothesis

Ava effect: To get a global an Ava has to spend some peds before. The waittime till the first global or the next one will follow a random distribution. However, the probability of getting one in the next kill will depend also on the previous outcomes, i.e.

P(Gn = 1| Gn-1=xn-1, ...,Gn-t=xn-t)


In a pure random situation without any relation to prevoius kills, the following must be valid:

P(Gn = 1| Gn-1=xn-1, ...,Gn-t=xn-t) = P(Gn = 1)

this assumption is already falsified in analyzing waittimes and global grouping. Hence we can conclude that there is something.

Now the question is if

P(Gn = 1| Gn-1=xn-1, ...,Gn-t=xn-t; Ava = AV; Area = AR)

depends also on Area and Ava.

Atm we don't have any valid data about it.


Let's do some further reasoning. LA's get an income via tax. It would be strange if this income is purely random. It might be reasonable to assume, that there is only income if peds have been spent in this area. Therefore an Area effect might be plausible. To make a further example: assume an Ava did spend about 1k PED in area A without any reasonable return. He changes then to area B and gets a hof. Area B would have an income and area A not.

Sure everything can be random, so that in mean these things should get balanced. However, I doubt that this is the case.

The same Area reasoning could be applied to the Ava. If one Ava gets a hof and makes an overall profit, assume a hof in the first ever kill, this would imbalance the whole system. Therefore an Ava effect is also plausible.

So what should we test.

In my Hogg run I have seen mainly the Ava effect. I will therefore extend it and try to get as many kills without a global as possible. In the last run I had the situation, that I didn't global in 45 kills. After a break I've got a hof in the second kill. So if I'm able to collect non globalling Hogs (say about 30-40) I'll change area and see what happens. If my assumption is right, I should start a new loot series. If I get a global quite soon that my assumption might be wrong.

To test both effect it will be necessary to repeat such runs. But I have a lot of time in front of me. So let's see how long it takes.

[kalanen]

How about universal "uptime" and grouping? It should be easy to dig out using existing global data.

For example how long "holes" occur between 6am and 10am (MA time). And those mobs that global at silent times, tend to have higher peaks during "uptime".

[falkao]

Originally Posted by kalanen

How about universal "uptime" and grouping? It should be easy to dig out using existing global data.

For example how long "holes" occur between 6am and 10am (MA time). And those mobs that global at silent times, tend to have higher peaks during "uptime".

should the holes be analyzed per mob?

Silent times are times where we have a lower number of globals recorded, and you mean that their loot value should be higher?

[Kolobok]

Originally Posted by falkao

Loot value has a now more or less known distribution and it follows a random process. Waittimes between globals however do show a bias, i.e. some grouping. Moreover, it seems that globals tend to be higher when return was quite low before.

This is true. But I'm afraid that with your next hypothesis you are going the wrong direction. You are trying to check whether previous spends affect the future payouts. Without this assumption it can be seen as just time-dependent probabilities per avatar. In practical sense, it means the following (over-simplifying, of course): go find a good mob, wait for good time, get three globals, go back home. It works no matter how much I've spent before.

What I learned from my mentors since my first days in EU is that there can be a good mob and a bad mob to hunt for today. If a mob is looting bad, change the mob. If it doesn't loot well anymore, stop hunting. If hunting in general is bad for today, go mining. If everything is bad, have a cup of tea and go read EF. If one persists with a bad hunting strategy, one can spend literally 16 hours of continuous hunting with low return. With a hot mob globals come soon. When a hot mob stops looting well, stop the hunt and return home with profit.

Remember one of VUs when MA made a mistake with extra high mining returns, and people started to spam the global chat with hofs. Those of them who were quick enough to realize what was going on, started to drop bombs where they stood and get hofs and towers. This lasted for three minutes then MA shut down the servers.

All of this is an indication that the probabilities of loot slots are dynamic, time-dependent, controlled by MA manually or semi-automatically with regards to some "global loot pool". These cycles can be seen as a superposition of several periodic functions, individual for an avatar, think of them as biorhythms. I can't predict precisely when next hot time comes (some people claim they can). But it is very possible to see that you are in the hot time, because it has very different loot pattern. It is also more or less clear when this hot time finishes, and it is better to go home with a big bag full of loot.

[kalanen]

Originally Posted by falkao

should the holes be analyzed per mob?

Silent times are times where we have a lower number of globals recorded, and you mean that their loot value should be higher?

By example:

During first 30 mins, there's 2 50 ped hogg globies. Next 15 minutes holds 2 50 ped feff globies and one 100 ped hogg.

Second period is half of first period and loot values are doubled. Feffs (having half of hoggs hp) loots are doubled, which made them above 50 ped global limit.

Something like this

[falkao]

Originally Posted by Kolobok

This is true. But I'm afraid that with your next hypothesis you are going the wrong direction. You are trying to check whether previous spends affect the future payouts. Without this assumption it can be seen as just time-dependent probabilities per avatar. ...

that's exactly the point.

I had the same impression till we found the grouping. Let me explain


If everything is purely random, then the prob. to get a global in the next kill is the same for all kills. If you're interested in successive kills, then it is a simple binomial prob. An example:

Assume the prob. to get a global is .01. The prob to get at least one in 10 kills is then 1-(1-.01)^10 = .1 and to get at least 2 in 10 kills is .004. What we have seen is, that there are groups of globals and it happens rather frequently that you'll get in 10 kills 2-3 Globals followed by periods of no globals. So I'cant expect one global with a prob of .1 in 10 kills but I have periods where this prob is much lower and periods where it is much higher.

This implies that the prob to get a global in the next kill depends on some factors. It's still random but now a conditional probability. What I`m trying to understand is what the conditions are. Is it area, the mob, a loot balancing within ava or whatever.

Originally Posted by Kolobok

,, With a hot mob globals come soon. When a hot mob stops looting well, stop the hunt and return home with profit.

I doubt on this. I don't think you will ever get more back as spent. Sure if you look only the last x kills, then you might have a profit, but overall its still lower as peds spent.

Originally Posted by Kolobok

Remember one of VUs when MA made a mistake with extra high mining returns, and people started to spam the global chat with hofs. Those of them who were quick enough to realize what was going on, started to drop bombs where they stood and get hofs and towers. This lasted for three minutes then MA shut down the servers.

Or when we had the asteroid impact. The system can be tweaked by parameters, and when a human tweaks something then instead of setting .1 he might set 1. So instead in getting something with prob .1 you'll get it always. Your example shows only that errors might happen and that MA tweaks loot.

Originally Posted by Kolobok

All of this is an indication that the probabilities of loot slots are dynamic, time-dependent, controlled by MA manually or semi-automatically with regards to some "global loot pool". These cycles can be seen as a superposition of several periodic functions, individual for an avatar, think of them as biorhythms. I can't predict precisely when next hot time comes (some people claim they can). But it is very possible to see that you are in the hot time, because it has very different loot pattern. It is also more or less clear when this hot time finishes, and it is better to go home with a big bag full of loot.

What I try to understand is what kind of game I'm playing. I hate gambling and therefore I want to be sure that EU is not gambling, although it looks like that. Another reason is to understand what my expected return is, what I can improve to have a better one and how all this fits to RCE. Are my investments save, can I trust MA etc etc.

The strange taste I have with that is, why we have to discover it and why it isn't explained somewhere.

[falkao]

Originally Posted by kalanen

By example:

During first 30 mins, there's 2 50 ped hogg globies. Next 15 minutes holds 2 50 ped feff globies and one 100 ped hogg.

Second period is half of first period and loot values are doubled. Feffs (having half of hoggs hp) loots are doubled, which made them above 50 ped global limit.

Something like this

ok, I'll try to check the data I have. Didn't consider amounts yet.

[-=rAndom=-]

Originally Posted by falkao

The strange taste I have with that is, why we have to discover it and why it isn't explained somewhere.

Most likely because if we were told we would play differently than MA wants us to p(l)ay.

[Kolobok]

Originally Posted by falkao

I doubt on this. I don't think you will ever get more back as spent. Sure if you look only the last x kills, then you might have a profit, but overall its still lower as peds spent.

It is clear that the overall payout can't be higher than peds spent. However some people are more successful than others, and knowing these people I doubt it is a matter of pure luck.

The examples of non-depositors show that profiting is possible, including myself. But I'm not a good example. Among my soc mates there are people who made their initial money from sweating and never deposited. One of them got an ATH yesterday after an amazing series of mining hofs.

[falkao]

Originally Posted by Kolobok

It is clear that the overall payout can't be higher than peds spent. However some people are more successful than others, and knowing these people I doubt it is a matter of pure luck.

That's the question, do we have an overall random payout that in sum can't be greater than peds spent, or is the payout balanced to the avatar. The first one would be very gambling like, the second for sure not.

Originally Posted by Kolobok

The examples of non-depositors show that profiting is possible, including myself. But I'm not a good example. Among my soc mates there are people who made their initial money from sweating and never deposited. One of them got an ATH yesterday after an amazing series of mining hofs.

I do not exclude those cases. When I talk about profiting, I do not include markup. That's another matter.