Info: A snable experiment

[Wollongong]

I have taken my opalo, ammo, and a huge bunch of candybars up north, and started hunting.

AIM: to find out the TT returns on hunting snables

300 snables later, here is the results:

1) Number of animals with loot: 186 (62% of the animals killed, 38% of them are no-looters)

2) 61 animals had loot which is worth enough to pay for the ammo and decay used to kill it. This means 20,3% of all animals killed did at least break-even.

3) Total ammo used: 35,72 PED, decay 0,56 PED

4) Total looted: 26,12 PED

5) Loss of 9,60 PED over 300 snables, an average of 3.2 pec each.

6) Loss is 28,00441% of the cost.

So those saying that just hunting snables will result in profit... this undermines that statement. Eco-snabling is no more.

You lose, but you lose less rapidly.

 

[mastermesh]

tame em

tame some of those snables, sell the pets, and maybe that'll make the loss a little less. There are lots of whips I'm seeing pass through the auction at tt prices these days.

 

[Wollongong]

that was not the aim of the experiment. People say "hunt snables, it is profitable" while others disagree, but nobody actually did the research. Now, the research is done.

 

[Modified]

First, I'd say the 'experiment' is a bit small. To get a good sample I think you'd have to kill at least a 1000.

Secondly, the opalo (yes, even the opalo) has some overkill on mobs that have so little health.

Maybe try the TT handgun and a bigger sample?

Aside from that, i agree that snables aren't really that profitable.

Regards,

Mod

 

[Hijacker27]

The profit also comes from hoarding and then selling the pieces of loot that drop with markup, such as snable hides for example, and beautician supplies. And yes, try putting some more PED into them

 

[Wollongong]

First, I'd say the 'experiment' is a bit small. To get a good sample I think you'd have to kill at least a 1000.

Secondly, the opalo (yes, even the opalo) has some overkill on mobs that have so little health.

Maybe try the TT handgun and a bigger sample?

Aside from that, i agree that snables aren't really that profitable.

Regards,

Mod

Following the "central tendency" (no clue how it is properly translated), you can take an n=30 to draw conclusions about the population.

 

[Wolverine Queen]

Following the "central tendency" (no clue how it is properly translated), you can take an n=30 to draw conclusions about the population.

EU is dynamic, you should know that

Repeat that 300 snable test 10 times and the data starts to be near to enough

 

[ForrestPark]

Good test and clearly laid out results. +rep.

My comments:

- The return, 72%, whilst I am not saying it isn't 100% accurate, sounds lower than I would expect if you had killed 10,000 of them. I would expect this would be about 85%-90% in the longer run.

- 300 is quite a few mobs but not really enough. It sounds like you didn't kill enough of them to get a nice 10 ped loot which would have helped even things out.

- You don't seem to have taken into account any markup, although I can see you have clearly stated that in the test. The MU would not account for more than a ped anyway I would guess.

- You didn't appear to use an A101 amp, so I would hazard a guess that your efficiency could be marginally higher.

- I am assuming the Opalo is maxed (!) although it doesn't say that

It would be very interesting if someone else would do exactly the same test . . . .

 

[Wollongong]

Good test and clearly laid out results. +rep.

My comments:

- The return, 72%, whilst I am not saying it isn't 100% accurate, sounds lower than I would expect if you had killed 10,000 of them. I would expect this would be about 85%-90% in the longer run.

- 300 is quite a few mobs but not really enough. It sounds like you didn't kill enough of them to get a nice 10 ped loot which would have helped even things out.

- You don't seem to have taken into account any markup, although I can see you have clearly stated that in the test. The MU would not account for more than a ped anyway I would guess.

- You didn't appear to use an A101 amp, so I would hazard a guess that your efficiency could be marginally higher.

- I am assuming the Opalo is maxed (!) although it doesn't say that

It would be very interesting if someone else would do exactly the same test . . . .

1) Well, the purpose of a test is to show whether expectations are correct.... looking at all the statistics I have had in uni (quite a bit), my sample is big enough to be valid, so I dare say the expectancy was incorrect.

2) Sure, big loots help. I had 2 loot over 2 PED. 10 PED+ are extremly rare with snable. Sure, one or two might pop up over 3000 snables, but the effect on the total return would be negligable

3) No, the purpose of this test was to determine the return MA gives us. Markup is not relevant for that purpose. I could even argue that markup has no intrinsic value.

4) Possible

5) You assume correctly.

 

[~CHOPPER~]

And yes, try putting some more PED into them

Don't post this into Franks $15 a month thread

 

[LoTeK]

I did a test to find out the returns on mining.

I used 70probes(35ped) and got back 25ped in materials.

1) number of claims found 20

2) number of misses 50

conclusion: mining is not profitable any more


I thought i'd do a second test.


I used 70 probes(35ped) and got back 3,000ped in materials

1) number of claims found 20

2) number of misses 50.

Conclusion: mining is fantastic you win shitloads of ped


If i posted this and was serious about the results,everyone would laugh and rightly so.

 

[Hijacker27]

Don't post this into Franks $15 a month thread

He put $3.5 into them, not quite near $15

 

[Madhoax]

And what about the gain in skills? I'm hunting snables right now, not much loot, but skills are going up like crazy.

 

[Kosh]

Following the "central tendency" (no clue how it is properly translated), you can take an n=30 to draw conclusions about the population.

n=30 is just a rule of thumb (n=infinity is actually the formal one for the Central Limit Theorem), when the variance is big you would need a much bigger sample to get any significant results. Everything in EU has large variance, you need a much bigger sample to draw any conclusions.

 

[Wollongong]

n=30 is just a rule of thumb (n=infinity is actually the formal one for the Central Limit Theorem), when the variance is big you would need a much bigger sample to get any significant results. Everything in EU has large variance, you need a much bigger sample to draw any conclusions.

which is why I went with 300, instead of 30.

 

[exsolvo]

I Might be trying out a N=XXX * 10 or so this weekend if i iget the time and urge, on snables with opalo, or the TT pistol, no armor, no fapping (Self healing) just to see if i can confirm this theorem.

Ill get back to this thread with the results
-----------------
UPDATE
My journal of the experiment can be found here:
YASE Yet Another Snable Experiment

 

[Ime]

which is why I went with 300, instead of 30.

It's just not enough. I suppose you killed the 300 once in a row. You have to make several run on several different days.
You can't know how many times i lost big on a hunt and got a nice global on the 1st mob the day after. Time seems to adjust your return in your loot.
This test is irrevelant.

 

[Ime]

So those saying that just hunting snables will result in profit... this undermines that statement. Eco-snabling is no more.

3) No, the purpose of this test was to determine the return MA gives us. Markup is not relevant for that purpose. I could even argue that markup has no intrinsic value.

And the goal was to see if it could be profitable or to see what MA gives us ?
Because when someone says it's profitable, he takes the markup and % into the count.

 

[ano]

In one breath I want to stamp my feet and say wrong because you could have got uber lucky with a 2k snable and this would have meant (Assuming your cost to kill remained the same ~12 pec) you'd still be up after 16,000 no looters. 300 really isn't enough imo to justify generalised conclusions, particularly as you have so many unknowns that applying any statistics to it seems tricky at best, hand wavy and counter productive at worst.

In the other breath I can't help but agree with the conclusion. If any activity was found to be profitable in tt terms for all those who tried it, then the game would surely break. However, this does not exclude the possibility that there exists an activity for which a greater proportion of people profit than in others.

 

[Pauhe Gunor]

i was hunting snables other day and i come out at a 30 ped profit and that was just TT value

 

[wizzszz]

Now repeat that 49 times more and we might get some meaningful results...

 

[Wollongong]

It's just not enough. I suppose you killed the 300 once in a row. You have to make several run on several different days.
You can't know how many times i lost big on a hunt and got a nice global on the 1st mob the day after. Time seems to adjust your return in your loot.
This test is irrevelant.

You supposed wrong

 

[JohnCapital]

My "test":

Weapon used: Vumpoor Rx10 (L)
Cost to kill: 0.2396 peds /2 = 0.1198 (killed two males in one shot)

Result of 2nd mob


1,100 /2 (1st mob no-looted) = 550 per mob

So those saying that just hunting snables will result in profit... this undermines that statement. Eco-snabling is no more.

You lose, but you lose less rapidly.

No, you lost. I won.

 

[Karmic]

Wonderful, but the central limit theorem and pretty much all of the rest of your statistics only applies to normal distributions, which mob loot is not. In order to correctly apply it you need to take a sampling of the mean of several different runs in order to more closely approximate a normal distribution. So, do your experiment 49 more times or so and we'll have a better idea. if you take n=30 and do a distribution of 500 means (total of 15,000 snables) we'll have a good sample.

I just posted the results of a similar test in another thread btw.

 

[AL Killmore]

This "experiment" provides quite unsufficient data for any conclusions with real value.

Number of killed mobs - too low. any number below 10000 is too low (imho).
Time frame - too short. there are cycles for the loot. 2-3 weeks should be the minimum.
Eco setup - A101+finisher gun is optimal for any real snable test.
Maturity of the mobs - loot vary for the different maturities
Markup - it's common belief that long term TT return is about 80-95%. What is important for the player is the total amount of PED that he'll recieve for his hunting efforts. Not just the TT value. I would't mind regular 30% TT return if it was all ESI's

Other than that the experiment was just fine. You wasted some ammo, got some of it back and MA had their share too. I'd expect similar results from any random experiment like this one.

 

[Mordrell]

Sorry m8 but your experiment is just too incomplete.... seems like you had an average return of 75-75%??. That is pretty ok really it means if had hit just one 10 pedder you would have had your profit.

Hunting snable is just like hunting bigger mobs the rewards/losses are alot smaller. I hunt ALOT of snable/exa/combibo/daika/keberos & Berycled and whenever the pedcard runs low I just return to atlas haven and grind the snable/daika combo, or if they dont seem to drop much I tp east and do the same with the exa/kerberos spawn there.

Maybe do 100 runs like the one you did and then post results, it would be alot more informative

 

[Witte]

The only thing you can really test is the return without globals (or minis even). Since there is a good chance that this is dependent on the time and location you perform the test, there should be several tests on different times and locations to get an accurate figure for this.

But over all I think the conclusion is correct. When you hunt random snables with opalo you will lose in the long run. I don't know who is claiming otherwise, but I would be interested to see proof of that.

 

[Wollongong]

Wonderful, but the central limit theorem and pretty much all of the rest of your statistics only applies to normal distributions, which mob loot is not. In order to correctly apply it you need to take a sampling of the mean of several different runs in order to more closely approximate a normal distribution. So, do your experiment 49 more times or so and we'll have a better idea. if you take n=30 and do a distribution of 500 means (total of 15,000 snables) we'll have a good sample.

I just posted the results of a similar test in another thread btw.

The central theorem states that you can treat any population as standard normal divided with a sample large enough (with indeed, the rule of thumb being 30)

To all those saying: I want more, I ask one question...

do you have ANY clue how many patients are used to test whether a medicine works...?

No?

A lot less than the number of snables I killed.

@Al_Killmore
300 is statistically sufficient
Timeframe: It was done over about 5 hunts, divided over two weeks
Eco Setup: agreed, but I was testing opalo
Maturity: I kept track of the maturity. I have not found any significant differences.
Markup: It is a common belief....? I dunno, but I still feel research goes over "common belief". And markup... well... that was not what I was looking into. I was looking into the returns at tt value. Beside, markup is "hot air"... (remember how hair gel used to do 300%? Or sweat used to go for 0.7 pec?)

 

[Az Bloodz]

1. You can't compare EU with the real world nor can you compare mobs with humans.
Reason is quite simple, a Mob can drop a global, a mini just anything. So you have 1 mob giving you a larger percentage of your entire return. In medicine tests there usually is no single test subjects providing a larger % of date [i.e. risks] than the other ones. This is EU, this is different.

2. You need to kill more mobs.
Don't understand why you are so confident with your test. There are many blogs providing data showing that you need to cycle a large amount of PEDs to get any result. Not enough PEDs cycled won't give you any real average.

3. Everyone is telling you the same.
Yet you avoid a real discussing by repeating your flawed argumentation and comparison to the real world. It's your small flawed test versus any other bigger blog.

 

[Kosh]

The central theorem states that you can treat any population as standard normal divided with a sample large enough (with indeed, the rule of thumb being 30)

That is correct. Here the CLT isn't even needed because there was no hypothesis testing so no need to assume a normal distribution.
The law of large numbers that states the average converges to the mean is sufficient for your case (no need to assume anything about the distribution except for a finite mean).

I still do not think 300 is a big enough sample though. Tests for medicine are quite different, they usually use ANOVA techniques to explain variables that can only get a small number of values (toxic/non toxic and such). Also they are usually very expensive to conduct on huge samples.

Here you are checking a variable of lets say 100 mobs killed and checking the TT rate of return. Some runs you will get 30%, some runs you will get 500%. That is very big variance. Also this variable might be dependant on the time, previous runs, other hunters and who knows what else. So doing this 100 mob run 100 times would probably give you much more insightful results.