Hierarchy of Fortune Wheels

[Parlog]

What i see are interlocking fortune wheels with different ranges of loots in a strict, pyramidal hierarchy. You click a BP, drop a bomb or kill a mob and you turn the wheel that you are on. They are interlocked, so sometimes you enter a field where you exit one wheel and enter a new one. Sometimes you stay on your wheel like forever till you find the exit, but usually you gather spin, read loot potential, that will release you on a higher wheel in the hierarchy, which will result in better loots.

Dmg/sec helps to get up in the hierarchy of wheel, doing things out of the ordinary helps, beeing a good example helps MA teasing people in by giving you a teaser loot and thus highlighting you, because you are doing the things better than others do.

Doing the same things over and over again makes you drop in the hierarchy of wheels (especially if you do not do them better than others), but if you do them long enough you may be an example of endurance which again pushes you back up. Usually your way is a constant ups and downs between 2 wheels, one above the profit line one under it, resulting in good days and bad days. The average 2.8dmg/pec player is in the middle of the hierarchy which results in a tt return of 90%, spending more, spending longer and spending more economical than others are virtues that bring you up, but pushings things too far and thus making them all too common makes you boring and lame and thus will pull you down again. Fighting against windmills again lets you drop too, so watch out whom or what you challenge.

When you are close to leveling up your profession and/or ranking up a relevant skill the system let you drop to the bottom of the pyramide and gives you a loot curse, once you got the new rank or level, you will be back on a normal range of loots or will result in the long awaited hof. You can chip in and jump a lootcurse btw. its all money in MAs pockets if you chip. If you chip out big masses of skills (cause you are sick of your loots) you create a big mass of consumable TT, which again is a virtue and might result in a decent loot. If you chip in too much, you jumped too much bad loot periods and MA will make you suffer and you lose, what i earlier called spin, the chance to raise to a higher fortune wheel with better loots. (maybe levitation potential is a better word than spin)

There are certain things that are to avoid, for example using mainstream equipment. If everyone at your ligue is using Apis untiered, dont bother to use it untiered, though dmg-enhanced tier 2-3 might make you different and you gain. If everybody in your ligue is using XT tier 3, try tier 5 (more dmg/sec than average) or untiered(more endurance due to less tt loss). You might be surprised how this works out. But again dont do it all the time, you might get boring, might lose your advantage, because your setup gets common, and you fall to a wheel with worse loots than what you are used to.

I could go on and on with bromides but i will stop here, just one thing at the end: If you are a specialist, you need to be the best specialist in your ligue to see more upside wheels than bottom wheels, if you are a generalist, make sure you bring more flexibility than everybody else. This will help you deal with the system, but a whole other system is playing against other players and this is what you are doing 24/7 playing this game.

System busted!

 

[Corey 'Greyfox']

neat theory. Like all theories on loot it is well researched and wrong.

 

[ermik]

humans tend to look for patterns even when there are none.

a system designed to give loot to 10's of thousands similtainiously users CANT be built on complex patterns.

you buy loot from the system for a fixed price / hp .

no more , no less.

there are 2 seperate main systems.

1 that keeps track of your tt return in relation to hp killed and tt spent in the actual killing.

1 that distributes loot to various places , and if you are in a hot area and system 1 see you are low on return , you will be bumped up with a nice loot.

everything else would be to complicated , ofcourse there are sub functions to each main system but i doubt they do much other than maybe handling some variables of lesser concern than the main function.

cheers

ermik

 

[kleus]

humans tend to look for patterns even when there are none.

a system designed to give loot to 10's of thousands similtainiously users CANT be built on complex patterns.

you buy loot from the system for a fixed price / hp .

no more , no less.

there are 2 seperate main systems.

1 that keeps track of your tt return in relation to hp killed and tt spent in the actual killing.

1 that distributes loot to various places , and if you are in a hot area and system 1 see you are low on return , you will be bumped up with a nice loot.

everything else would be to complicated , ofcourse there are sub functions to each main system but i doubt they do much other than maybe handling some variables of lesser concern than the main function.

cheers

ermik

Half true.

That system designed to keep track of tt spent doesn't exist. The other one i'm not sure/don't have enough data, but it seems both unlikelly and unnecessary.

Loot follows a normal distribution, and it's average is equal to the cost of killing a creature playing eco, not counting the misses, armor or fap.

The problem is that the loot algorithm itself is made to make even the ocasional profit unprobable. So usually, to reach the 90%, you have to get a really high loot (which are rare, but they are necessary to complete the loot cycle). When you get that loot, it's not luck, it's just how the algrithm is programed.

To understand this last statement, picture this, lets imagine the cost of killing creature X is 1 ped. Arround half the time you won't get loot, when you do, it's most common value would be arround 0,7 - 1,1 ped, but some rare times you'll get 20 ped loot (Following this example you would get that loot after arround 50 kills or so), and it's this higher than average loot that balance the overrall loot, into reaching 90%.

Obviously, and as this is only an algorithm, it's possible to kill 2 creatures and they both be worth the 45 ped, or, on the other hand, it's possible to kill the 50 creatures without ever having the balancing loot. But if you kill 1000 of said creatures you'll end up with something really close to 90%.

It happens this way because if the loot was averaged, lets say, everytime you killed a creature you would get 2 ped, it would be easy to trick the system, and profit.

There are other variables that seem to be in the algorithm that can be taken into account, and it is in fact possible to get a really small profit (10% or so) from 50% of the hunts, you just have to know how the loot really works (i'm fairly sure that lot of people cracked it, but i'm yet to see it posted here, mainly people only come here to rant and whine).

Also, i don't mean, in anyway, that i really know how it works and i profit from it all the time, i'm just stating what i saw from the 3 or 4k kills of the same creature, while collecting data.

The trick into getting the 90% is to play eco, and making sure to kill enough of them to complete the cycle.

The existance of the cycle itself is proof that the algorithm is based on probabilities and it's not random has people usually say it is.

 

[mrproper]

You got something there, but either it's an incomplete understanding of the system by yourself, or you just can't explain it right so everyone else understands. I recommend some visual aids and rewording the whole thing. Plus, if you talk only about mob hunting, it becomes easier to explain

 

[mrproper]

That system designed to keep track of tt spent doesn't exist.

Seems to me that it exists. Practically the whole profit model requires TT tracking per participant, it is possible for it to exist, and I think it does exist. There is no proof that it does not exist, and there are plenty of circumstances where it has been shown to exist (such as sudden HOFs after bad loot streaks, larger loot when overspending on a mob kill, huge loot variance with uneco hunting, low loot variance with eco hunting, and many other examples, complete with charts, etc..)

 

[remontoire]

Loot follows a normal distribution, and it's average is equal to the cost of killing a creature playing eco, not counting the misses, armor or fap.

This is wrong. Loot curve is not normal distribution shaped.

The problem is that the loot algorithm itself is made to make even the ocasional profit unprobable. So usually, to reach the 90%, you have to get a really high loot (which are rare, but they are necessary to complete the loot cycle). When you get that loot, it's not luck, it's just how the algrithm is programed.

While there is a general assumption that the average that returns average to is 90%, it is not set in stone or proved, and some of the better documented cases of "loot aevrages to 90% TT have since gone on to achieve more.

What we know is that the no-risk return is 90%. No reason why return with risk could not be higher, or lower, dependning on circumstances.

 

[kleus]

You got something there, but either it's an incomplete understanding of the system by yourself, or you just can't explain it right so everyone else understands. I recommend some visual aids and rewording the whole thing. Plus, if you talk only about mob hunting, it becomes easier to explain

Yes i thought i was. At least what i stated was how the loot worked for the first 3 - 4k kills. But i made 1k kills last night that were totally off, disproving my theory.

Well, not disproving it completely, because i have more than enough data to backup my original theory, but somehow pointing to the existance of another algorithm that sometimes goes off and diminishes your return.

Maybe that's why they call loot dynamic. More than one algorithm that gets setted off from time to time, maybe there's some truth to what the OP of this thread said...

So this must mean that if there is one algorithm design to make you have losses, there's probably one designed to give you a profit. I'd have to collect more data.

 

[mrproper]

I'm glad you're using the scientific method on the data you collected yourself. Sometimes you're wrong and you don't even need others to tell you that. It happened to me a few times and I sympathize with the feeling.

Design your theoretical model, and make predictions about what your next X kills will look like. But please try to determine a baseline. For example, hunt for at least 25% of the test TT value, with a very eco weapon on small mobs. This will iron-out any imbalances that could get corrected during your effective testing of 100% TT value. Do another cool-down period of another 25% TT value so you detect any anomalities. Hopefully. If you get unusual data, like very long no-loots or out of the ordinary large loots, throw the whole test and start over.

Another idea would be to divide your main test into 3-4 parts, and use the data from each part individually testing your model, as well as the whole parts combined. If the model conforms to the sum of the parts and to the majority of the parts, you are very close to the desired model. Predicting a model with 95%+ confidence is a useful discovery as this model can be used with trust that it represents normal behavior of the system.

If you want, you can even go further, and add 2-3 more people into the experiment, each following your instructions on repeating the tests, keeping in mind the above.

 

[kleus]

I'm glad you're using the scientific method on the data you collected yourself. Sometimes you're wrong and you don't even need others to tell you that. It happened to me a few times and I sympathize with the feeling.

Design your theoretical model, and make predictions about what your next X kills will look like. But please try to determine a baseline. For example, hunt for at least 25% of the test TT value, with a very eco weapon on small mobs. This will iron-out any imbalances that could get corrected during your effective testing of 100% TT value. Do another cool-down period of another 25% TT value so you detect any anomalities. Hopefully. If you get unusual data, like very long no-loots or out of the ordinary large loots, throw the whole test and start over.

Another idea would be to divide your main test into 3-4 parts, and use the data from each part individually testing your model, as well as the whole parts combined. If the model conforms to the sum of the parts and to the majority of the parts, you are very close to the desired model. Predicting a model with 95%+ confidence is a useful discovery as this model can be used with trust that it represents normal behavior of the system.

If you want, you can even go further, and add 2-3 more people into the experiment, each following your instructions on repeating the tests, keeping in mind the above.

Those are very good tips, i will start that asap. I'm hunting shinkibas, i'll now start a 100 ped run on shinkiba young alone. 100 ped should be more than enough for at least 5 loot cycles. So i'll be doing it 25 ped each.

I'm hunting with an opallo with an a101 no armour, no fap. I'll get back to this thread with my findings.

*Edit* It seems that i don't have the necessary funding to do this test at the moment, so i'll have to wait for the next deposit (i'm being really carefull with deposits as to not overdeposit, this is just a game after all, and i'm playing with 250 ped a month).

Next month is really close though.

 

[kleus]

This is wrong. Loot curve is not normal distribution shaped.



While there is a general assumption that the average that returns average to is 90%, it is not set in stone or proved, and some of the better documented cases of "loot aevrages to 90% TT have since gone on to achieve more.

What we know is that the no-risk return is 90%. No reason why return with risk could not be higher, or lower, dependning on circumstances.

Sorry, i think i missed your post there. I'm just talking about the data i collected, haven't seen any other type of data in this foruns, though everyone seems to talk about the 90% tt profit, and it fits with what i found.

About it being a normal distribution, I haven't done the test yet, but it sure looks that way, i'll do the test when i collect more data.

 

[Serica]

Sorry, i think i missed your post there. I'm just talking about the data i collected, haven't seen any other type of data in this foruns, though everyone seems to talk about the 90% tt profit, and it fits with what i found.

About it being a normal distribution, I haven't done the test yet, but it sure looks that way, i'll do the test when i collect more data.

I've been keeping a fairly detailed blog here for the past few weeks.

Current return on TT is: 90.14%, over 8.9kped cycled.

Feel free to take a look.

 

[kleus]

I've been keeping a fairly detailed blog here for the past few weeks.

Current return on TT is: 90.14%, over 8.9kped cycled.

Feel free to take a look.

True, it is 90% tt profit. Are you hunting eco? If so, and if i may ask, what level of creatures are you hunting?