How to analyze loot

[florin]

so much math....need aspirine

[Phaser]

Originally Posted by florin

so much math....need aspirine

Im whit a cold got some extras boxes of aspirine if u need them.

[falkao]

Originally Posted by florin

so much math....need aspirine

unfortunately it is needed here and I know it might be hard to follow. Furthermore, I'm looking for one or two persons that are able to verify my results. Noodles already signalled this. Maybe somebody else?

[Jimmy B]

Originally Posted by falkao

unfortunately it is needed here and I know it might be hard to follow. Furthermore, I'm looking for one or two persons that are able to verify my results. Noodles already signalled this. Maybe somebody else?

I would if I could Some of the stats is beyond me I'm afraid.

[Doer]

I've had several experiences lately that suggest that specific mob "species" can have up and down periods within the same geographical region.
For example, the day i hit 8 warrior globals in an area with a lot of gen 6-9 drones (about same HP as warriors) without a single drone global.

A similar thing happened again last night when i hit a mining bot global, ran around for half an hour killing drones looking for another mining bot (grr, what an annoying mob to hunt), and then globaled on the next mining bot i found.

This would be different from when an area or avatar is "hot", hitting good loots on everything for a period of time.

True, the second example is possibly coincidental, but thinking about less hunted mobs and how you can go a whole hunt on them without them getting "hot", whereas with common mobs you can see them go up and down a few times, it seems to support some kind of cyclic overlay to the randomizer.

Maybe, a sine wave? I imagine the data could be extracted from the tracker and analyzed to support or disprove that idea, but considering it has missed about half my globals, i'm not sure how conclusive that would be.

[Jimmy B]

Originally Posted by Doer

I imagine the data could be extracted from the tracker and analyzed to support or disprove that idea, but considering it has missed about half my globals, i'm not sure how conclusive that would be.

Is it still missing them now? It wasn't seeing quite a few of mine either at one point, but the last week or so its got them all.

[Doer]

Originally Posted by Jimmy B

Is it still missing them now? It wasn't seeing quite a few of mine either at one point, but the last week or so its got them all.

http://www.entropiatracker.com/index...PlayerID=12968

It only has three of the eight warrior globals, missed my two miner bot globals last night, and just generally sucks.

Oh yeah, and it also missed my trooper HOF.

Not useful yet, IMO.

[MAS Flottilla]

Originally Posted by falkao

unfortunately it is needed here and I know it might be hard to follow. Furthermore, I'm looking for one or two persons that are able to verify my results. Noodles already signalled this. Maybe somebody else?

I truly appreciate your effort, even if some questions may arise... (one for all: how can we consider high mark up items in the equation? I mean, if a trox drops a jag part worth 5000 peds it's considered as a 60 ped global by your model... correct me if i'm wrong)

I had over time a much more empirical approach to the loot problem, and my stathistics maths is rusty... dating back to 20 years ago and more, so i'll only drop in some empirical results without pretending to be of much help, but they are nevertheless true.

I had three 1million ammo tests on troxes during past few months and some 1 million ammo tests in feffs and drones, so perhaps on an empirical basis my average return means something... well, it always varies between -7,5% and -15%, including decays ofc, and calculated on each million ammo spent on a single mob without moving from that zone/mob. It's never been higher and never lower...
That's what I call my "normalized" return, that is excluding uber items and hofs > 1000 peds. If I throw in the equation these last, in the long, long run I can fairly say I'm even and something more than that.
So in my opinion, game is far from being defined a gamble (definitely that's what u also tried to explain, with formal maths models)... so yes I agree with your conclusion (even if I don't understand every passage eheheh) and certainly on my empirical experience it's true. My mean return in these tests was even higher than the one u state here (around 90% excluding uber hofs) but that prolly varies depending on skills etc etc.

[Jimmy B]

Originally Posted by Doer

http://www.entropiatracker.com/index...PlayerID=12968

It only has three of the eight warrior globals, missed my two miner bot globals last night, and just generally sucks.

Oh yeah, and it also missed my trooper HOF.

Not useful yet, IMO.

Ah, that's not so great then. Hopefully it'll get more reliable with time.

Gratz on the mighty Sabakuma Mature global btw

[falkao]

Originally Posted by MAS Flottilla

I mean, if a trox drops a jag part worth 5000 peds it's considered as a 60 ped global by your model... correct me if i'm wrong)

I did not include any markup in the model. This would be missleading. So the model is based on retured TT. Markup can be introduced as mean markup or on a per mob basis if we know the distribution of special items.

Originally Posted by MAS Flottilla

...average return means something... well, it always varies between -7,5% and -15%.

My mean return in these tests was even higher than the one u state here (around 90% excluding uber hofs) but that prolly varies depending on skills etc etc.

That's a preciuos input. My model is very conservative, (i.e. I use the lower limits of the estimated class probabilities). A loot for class >= C3 is atm 2.5%. I'm sure it will be higher in the end, but due to limited data I don't want to publish results that are too high. So my given 78% return on Ambu Y means, the return is at least 78%.

Buk told me, that in his data using TT-value without markup, he has a return of about 96%. So we will see where we end up with a larger dataset.