Strategy to make a 91+ LP genie

ovenusarmanio · March 2013

serbetel wrote: »

Sorry but i m not using it - they do.

Not funny that insults are tolerated but so called "spamming" wich in fact is just answering questions and provocations makes a Mod take actions =\

It's that it's derailing the thread from it's original purpose. You're welcome to make your own thread about the topic if you wish to discuss it more.

@all

And please don't turn this into a flame war now. Just let's get back on track folks. Thank you. b:thanks

Adroit - Lost City · March 2013

I tried a simulation where I needed to have at least 70% of possible lucky points (and still possible) to continue, and the chance of making it went down slightly.

Average LP value: 93
Average Mirage Cost: 20619
Average SP Cost: 1187m (with a standard deviation of 858m SP)
Average # of 81-90 LP genies: 1.87
Proportion of these genies that were made with 630m or less spirit is: ~.412

It looks like the most efficient way to do this is keeping between 80% and 90% of possible lucky points, so I just tried specifying a few LP values to continue on
level 10 - 8+
level 20 - 17+
level 30 - 26+
level 40 - 34+
level 50 - 43+
level 60 - 52+
level 70 - 62+
level 80 - 71+
level 90 - 81+

Average LP value: 93
Average Mirage Cost: 23097
Average SP Cost: 971m (with a standard deviation of 766m SP)
Average # of 81-90 LP genies: 1.69
Proportion of these genies that were made with 630m or less spirit is: ~.470

This is the most efficient method that I've found so far, and probably getting pretty close to optimal. I'm a little bummed I've got less than a 50% chance to make one, but maybe using the leftover genie exp cubes will give me a slightly better than 50% chance to make my genie :P I'll probably try a couple more simulations and then start making genies in game.

Fissile - Archosaur · March 2013

I had a better result than this but I didn't write it down while I was playing with the inputs. (I believe it started 9 17 27 and ended 63 71 81, with ~56% under 630m spirit and average of 800m spirit.)

level 10 - 9+
level 20 - 17+
level 30 - 27+
level 40 - 35+
level 50 - 44+
level 60 - 53+
level 70 - 62+
level 80 - 71+
level 90 - 81+

Average LP value: (I don't calculate this)
Average Mirage Cost: 26454
Average SP Cost: 866m (with a standard deviation of 829m SP)
Average # of 81-90 LP genies: (I don't track this either)
Proportion of these genies that were made with 630m or less spirit is: 0.518

Fissile - Archosaur · March 2013

Just for fun I plugged your cutoffs into my code to make sure we get similar results. It seems one of us is calculating spirit wrong since there is a difference between what I get and what you posted. My value for the average number of mirages was only 3 less than yours so I think its safe to say we are both simulating the genie statistics the same.

I guess I will double-check my accounting of spirit.

Adroit - Lost City · March 2013

Fissile - Archosaur wrote: »

Just for fun I plugged your cutoffs into my code to make sure we get similar results. It seems one of us is calculating spirit wrong since there is a difference between what I get and what you posted. My value for the average number of mirages was only 3 less than yours so I think its safe to say we are both simulating the genie statistics the same.

I guess I will double-check my accounting of spirit.

I actually noticed the SP you sent me earlier was different from what I use, I just took my data from https://docs.google.com/spreadsheet/ccc?key=0AkkaqwR-6_tYcmhxZGp4V1lCanlXWDFRNWRrcnZDNlE#gid=0, I'm not sure how accurate those numbers are, but those are what I'm using. Thanks for the method suggestion, I'll play with those numbers a bit and see if I can bump it up any higher, or that may be what I end up using.

HrunsPanda - Archosaur · March 2013

both including infusion rate from a lvl 100 player ?

Adroit - Lost City · March 2013

HrunsPanda - Archosaur wrote: »

both including infusion rate from a lvl 100 player ?

I actually downloaded the spreadsheet, changed the player level to 105.. and looked at the far right column that says "Cumulative player spirit to level-up Genie". Just been using those values :P

Fissile - Archosaur · March 2013

b:surrender It appears my table of spirit values is correct but my calculation of infuse ratio was not. Based on what you linked there is a minimum infuse ratio of 10%. Also I had the wrong variable in my infuse ratio calculation and i was effectively using genie level + 1, where it should have been the genie level.

Adroit - Lost City · April 2013

Fissile - Archosaur wrote: »

b:surrender It appears my table of spirit values is correct but my calculation of infuse ratio was not. Based on what you linked there is a minimum infuse ratio of 10%. Also I had the wrong variable in my infuse ratio calculation and i was effectively using genie level + 1, where it should have been the genie level.

Ah yeah, btw I ran a simulation with the numbers you provided and came up with..

Average LP value: 93
Average Mirage Cost: 28281
Average SP Cost: 1009m (with a standard deviation of 787m SP)
Average # of 81-90 LP genies: 1.67
Proportion of these genies that were made with 630m or less spirit is: 0.454

It's a little odd that our numbers aren't that similar.. I would have thought that at least our mirage cost would have been near the same.. hmmm

HrunsPanda - Archosaur · April 2013

Spread sheet in production....
It can already tell that a 1/10 genie has a whopping 0.000032% chance of getting it to 91 while a 10/10 has an impressive 0.045% chance.
The 10/10 has a 1.13% chance to make it to 81.
Or a 8.76% chance making it to 71.

much more needed though.... still not sure if its ever gonna do all that i hope. Adding costs in SP and mirages to these chances will be peanuts. Addepting these chances however to a system where you dont proceed leveling all genies i still do not yet know how im gonna implement.

Proceed tomorrow...

soundslegit · April 2013

serbetel wrote: »

Stochastic is no easy topic and doesnt work like ppl think it does without learning it.

It starts with the chance a number out of [1..10] will be 10 is exactly 1:2
because the possible solutions are only TRUE and FALSE.

This statement is 100 % correct,
while the statement "the chance is 1:10" is non-proofable.

This is the single most insane thing I have ever seen on this forum.

Fissile - Archosaur · April 2013

Adroit - Lost City wrote: »

Ah yeah, btw I ran a simulation with the numbers you provided and came up with..

Average LP value: 93
Average Mirage Cost: 28281
Average SP Cost: 1009m (with a standard deviation of 787m SP)
Average # of 81-90 LP genies: 1.67
Proportion of these genies that were made with 630m or less spirit is: 0.454

It's a little odd that our numbers aren't that similar.. I would have thought that at least our mirage cost would have been near the same.. hmmm

I run your case again and I get 23586 mirage, 961m spirit, 47.1% below 630m spirit.

Rerunning mine with the corrected spirit calculation I get 27963, 973m, 45.9%. (Your numbers are within the standard error of mine.) I must have written it (the cutoff points) down wrong. b:embarrass

Adroit - Lost City · April 2013

Fissile - Archosaur wrote: »

I run your case again and I get 23586 mirage, 961m spirit, 47.1% below 630m spirit.

Rerunning mine with the corrected spirit calculation I get 27963, 973m, 45.9%. (Your numbers are within the standard error of mine.) I must have written it (the cutoff points) down wrong. b:embarrass

lmao, at least our simulations are fairly consistent then :P So have we found anything better than the first cutoff one I tried (got 47% chance)? I'll still play around with it a little, but I'm thinking that it is probably fairly close to what my final strategy will be. Thanks again for all the suggestions

@HrunsPanda, if you finish your spreadsheet, I'd love to look it over if you are willing to share :P

Azzazin - Dreamweaver · April 2013

Can one or both of you try and work out a way to figure out the average SP used getting genies above a certain cutoff? For example:

lvl 100 genie: 91+ LP, average sp cost ~1bil
lvl 90 genie: 81+ LP, average sp cost ~ ?
lvl 80 genie: 71+ LP, average sp cost ~ ?
lvl 70 genie: 62+ LP, average sp cost ~ ?

Basically, if you can figure this out, you can, as I said before, figure out the most efficient cutoff to get the highest number of candidate genies, then dish the genies out to friends with free SP to finish leveling them up.

I also envision being able to tabulate the total SP pool of you and your friends, then setting a cutoff for your own SP pool (example: use your 600mil sp to generate 50 candidate genies of 71+/80 LP, then finish them off with other people's SP).

Also, could you possibly calculate for us really quickly what the % chance is of your minimum cutoff genie for each lvl reaching 91+ LP? For example:

8/10 genie has % chance = A
17/20 genie has % chance = B
...
81/90 genie has % chance = X

If you get the gist of what I mean. This could be of real benefit to those of us who wish to obtain 91+ genies by purchasing genies from others, and weight odds against a proposed value of 91+ genie. (Example, if a particular genie has a 5% chance to reach 91+, and I value 91+ at 200mil, I would pay 1/20th of 200mil for the genie, or in this case, 10mil).

Azzazin

Adroit - Lost City · April 2013

@Azzazin - I'd be more than happy to try a few simulations to come up with lower LP, I'll just use the most efficient criteria I've found so far and run a few simulations overnight.

As far as determining the probability that a certain genie will hit 91+, I think you'd just need to spend some time playing with a spreadsheet to do this. Asterelle detailed a very straightforward way to do this (quoted below) and it sounds like HrunsPanda is working on putting the sheet together.. so you might be able to ask him if he finishes it (or you can make it with the method below).

Asterelle - Sanctuary wrote: »

It's actually just a direct calculation which isn't even all that complicated.
The probability of LP 1/10 up to LP 10/10 is straightforward:
P(LP1/10) = P(1)
P(LP2/10) = P(2)
etc...
The probability of LP 2/20 up to 20/20 is based off the terms from LP XX/10.
For example P(LP 4/20) = P(LP 1/10)*P(3) + P(LP 2/10)*P(2) + P(LP3/10)*P(1)
There are 3 terms here because there are 3 different ways to arrive at LP 4/20 (LP1/10 + 3, LP2/10 + 2, LP3/10 + 1).
In general you just use the values for P(??/YY) to calc P(??/YY+10).
Really this type of thing is very easy to setup in a spreadsheet.

The probability spread makes a kind of bell curve so I basically look at what percentile each value represents.
GRI(XX/YY) is 1 / sum (P(LP XX/YY) to P(LP YY/YY))
GRI of 100 would be something in the 99th percentile... GRI of 1000 is 99.9 percentile... etc.

Adroit - Lost City · April 2013

I did the lower level simulations like you suggested, I just used the most efficient criteria I had found in the past (8+,17+,26+,34+,43+,52+,62+,71+,81+).. but this may not be the most efficient way to make them, so keep that in mind :P

43+/50
Average LP value: 46
Average Mirage Cost: 264
Average SP Cost: 8m (with a standard deviation of 8m SP)

52+/60
Average LP value: 55
Average Mirage Cost: 591
Average SP Cost: 20m (with a standard deviation of 19m SP)

62+/70
Average LP value: 65
Average Mirage Cost:1458
Average SP Cost: 52m (with a standard deviation of 49m SP)

71+/80
Average LP value: 74
Average Mirage Cost: 3366
Average SP Cost: 126m (with a standard deviation of 121m SP)

81+/90
Average LP value: 84
Average Mirage Cost: 8745
Average SP Cost: 345m (with a standard deviation of 337m SP)

(and just for reference)
91+/100
Average LP value: 93
Average Mirage Cost: 23097
Average SP Cost: 971m (with a standard deviation of 766m SP)

Oups_Dead - Harshlands · April 2013

Adroit - Lost City wrote: »

I did the lower level simulations like you suggested, I just used the most efficient criteria I had found in the past (8+,17+,26+,34+,43+,52+,62+,71+,81+).. but this may not be the most efficient way to make them, so keep that in mind :P

43+/50
Average LP value: 46
Average Mirage Cost: 264
Average SP Cost: 8m (with a standard deviation of 8m SP)

52+/60
Average LP value: 55
Average Mirage Cost: 591
Average SP Cost: 20m (with a standard deviation of 19m SP)

62+/70
Average LP value: 65
Average Mirage Cost:1458
Average SP Cost: 52m (with a standard deviation of 49m SP)

71+/80
Average LP value: 74
Average Mirage Cost: 3366
Average SP Cost: 126m (with a standard deviation of 121m SP)

81+/90
Average LP value: 84
Average Mirage Cost: 8745
Average SP Cost: 345m (with a standard deviation of 337m SP)

(and just for reference)
91+/100
Average LP value: 93
Average Mirage Cost: 23097
Average SP Cost: 971m (with a standard deviation of 766m SP)

how did u calculate the standard deviation?

MagicEmpress - Lost City · April 2013

Adroit - Lost City wrote: »

After recently hitting 105, I now have ~630m sp sitting here.. and thought I'd try my luck at making a 91+ genie. The only thing holding me back is deciding on some criteria to determine if I want to continue investing sp in a genie. I originally was thinking that I'd keep working on it as long as the genie had at least 90% of its possible LP (i.e. a genie that was 18/20 would be worth continuing on), although I'm thinking this may not be the best way to approach the problem. Because the sp investment per level increases exponentially, perhaps it'd make sense to be more picky at lower levels and say it needs to be 19 or 20/20 to continue working on it because the sp cost is near negligible (although mirages/time isn't free either.. so meh)

I'd rather not just choose some arbitrary number, ideally there would be some way to calculate the most efficient way to do this.. but I'm not really sure how to do this. I'll take any ideas/opinions anyone has to offer, but obviously prefer something that isn't just arbitrarily chosen.

Thanks in advance,
Adroit

Is this an April Fools joke?
http://pwi-forum.perfectworld.com/image.php?u=3093928001&type=sigpic&dateline=1364035231
b:cry

Here's what I do (my sin has a 95/100 self made).
Save only:
8/10
17/20
26/30
35/40
43/50
51/60

Now you ask why 26/30?
It seems like 26/30 is not that good but if you roll a 9 or a 10 = 35-36/40 is good.
There is about a 35% chance to get a 9 or 10 so over 1/3 of those 26/30 genies would get it.
If you make a ton of genies that's a lot.

Now you ask why is 51/60 good?

If you get a genie to 51/60 there is a 50% chance to get 71/100 and 10% chance at 81/100 and a 3% chance at 91/100.
71 sell for 10 mil and 81 for 50 mil so every 51/60 genie has a 50% chance to be worth 10 mil if you take it to 100.
That's 5 million per genie if you have spirit to burn with a 10% chance at jackpot.
Sell enough of those and you can get 500 mil to buy a 91 if you have no luck making one yourself.

Here's what I mean.
If you can get 35 genies to 51/60 or better the 3% chance at 91 becomes even money.
3 - 4 should be 81 which means 150 mil - 200 mil when selling
and 15 others should be 71 so there's another 150 mil.

And don't let your spirit get to 1 billion or you lose it all (I hear). b:shocked

HrunsPanda - Archosaur · April 2013

any advise what site i should use to upload the file so you can download it from a link without needing to register, similar to photobucket ?
Then ill upload it tonight so people can give their tips and thoughts

Decus - Dreamweaver · April 2013

Zoho, I think?

Esnemyl - Dreamweaver · April 2013

My method:

1. use hundreds of mirages to create a couple of genies.
2. use all my spirit and watch as each one fails to go gold
3. try not to cry
4. lie down
5. cry a lot b:sad

FaceRolI - Sanctuary · April 2013

Esnemyl - Dreamweaver wrote: »

My method:

1. use hundreds of mirages to create a couple of genies.
2. use all my spirit and watch as each one fails to go gold
3. try not to cry
4. lie down
5. cry a lot b:sad

nonono thats not how this game works! it's designed to do what you dont want. so everytime that 30/30 genie goes 31/40 you are supposed to clap and when that 89/90 genie goes 90/100 you are supposed to burst out cheering! b:victory

CheetahWoods - Raging Tide · April 2013

soundslegit wrote: »

This is the single most insane thing I have ever seen on this forum.

what if those 2 dots between the 1 and 10 said "OR", as in [1 OR 10]
is this still insane in your opinion? this is what he really meant.

also, if you had been reading, the others already have sniffed out this error of only assuming 1 or 10 was possible

this is all moot , as it has been determined over 14 pages that monetary/time/other concerns were not part of this more pure theory solution.

as a result, fail genies need to be tossed due to practical considerations.

great thread. I finally was going to build a genie. these overall strategies fit with my gut feelings which have caused me to save 27+/30 genies and trash the lesser ones..

thx 4 everyone's effort and time, saved me a boatload.

Adroit - Lost City · April 2013

Oups_Dead - Harshlands wrote: »

how did u calculate the standard deviation?

I have a file for each method containing all of the 91+ genies created and it contains LP values, SP cost, mirages cost, and 81+ genies created. I go through the file once to calculate the mean SPcost, then go back through again and subtract each value by the mean, square this difference and sum the squared differences up. After I get to the end of the file, I divide the sum of the squared differences by one less than the total genies in file to determine the standard deviation. I did a small test with like 10 data points and used the StdDev() function in excel to verify that the results were exactly the same.

It is surprising how high the standard deviation is, but it does make some more sense when you realize how different the required SP values are. I'm including the first 10 data points I got from one of my trials just to give you an idea how all over the board the required SP is (first col is LP, 2nd col is #of genies tried to make it, 3rd col is SP cost, 4th col is # of 81-90 genies created using this method

92 1993 238749133 0
94 11017 1464660448 2
93 16706 1911820289 1
93 6084 800352793 2
95 481 97814488 0
92 7752 997454480 3
92 13822 1739635416 3
94 1369 212673138 0
95 834 130400669 0
95 1277 167056963 0

and for reference my script comes up with the following based on just the 10 above data points.

Average LP value: 94
Average Mirage Cost: 18402
Average SP Cost: 776m (with a standard deviation of 715m SP)
Average # of 81-90 LP genies: 1.10
Proportion of these genies that were made with 630m or less spirit is: 0.500

HrunsPanda - Archosaur · April 2013

Decus - Dreamweaver wrote: »

Zoho, I think?

Not really a convenient and simple one. So much BS and just 15 days trial.
You need an account to download anything.

Adroit - Lost City · April 2013

HrunsPanda - Archosaur wrote: »

Not really a convenient and simple one. So much BS and just 15 days trial.
I also doubt you can simply download it without having an account, but go try:

https://sheet.zoho.com/editor.do?nologo=true&doc=c2826da1f9894a5434d73e94a67a91abc8d045454b490407

link doesn't work

HrunsPanda - Archosaur · April 2013

Ok i made a hotmail account.

HrunsPanda@outlook.com
Password Archosaur

Its in the inbox.

It is however still very far from finished. Only does the chances and got some tables with numbers in it really. Tonight ill make some more functionality.

HrunsPanda - Archosaur · April 2013

Im updating the sheet with costs and found a most remarkable effect.

If you ditch any genie that is unable to get to 91/100, it costs you less on average to get a 91+ out of a 77/80 than out of a 78/80.

I have been thinking about this and checking my math for the last half hour. I am still a little perplexed actually, but it seems to be correct really. The reason is that the low and high scores have higher chances than the medium scores. The 77 hapens to create more genies that just not got to the 81/90 and thus cost no further investment and it happens that the remaining distrubution ends up costing slightly less even though you have to level more genies from 80-90 than for the 78.

Of course it is not totally realistic since you will never say that 91+ is absolutely your only goal and you are not gonna NPC that 80/90 genie that cannot get there anymore. It is actually a pretty valuable genie you made.

Maybe the value of these failures needs to be implemented as well to remove this oddity.

Adroit - Lost City · April 2013

HrunsPanda - Archosaur wrote: »

Im updating the sheet with costs and found a most remarkable effect.

If you ditch any genie that is unable to get to 91/100, it costs you less on average to get a 91+ out of a 77/80 than out of a 78/80.

I have been thinking about this and checking my math for the last half hour. I am still a little perplexed actually, but it seems to be correct really. Of course it is not totally realistic since you will never say that 91+ is absolutely your only goal and you are not gonna NPC that 80/90 genie that cannot get there anymore. It is actually a pretty valuable genie you made.

Maybe the value of these failures needs to be implemented as well to remove this oddity.

pretty sure you have an error in there..

HrunsPanda - Archosaur · April 2013

Ill upload it to my email what i have now. Im pritty sure its correct actually.

Uploaded. I have recalculated and my formulas manually. It works correct really. See edit in previous post for more explanation

The big difference in SP cost from 91-100 and 81-90 also plays a big role. This causes the amount of lvl 80 genies that make it into the succesfull lvl 90 selection to be less important than the distribution of them in this lvl 90 range and thus the amount of genies need to be leveled from 90 to 100

found something good for sharing

Strategy to make a 91+ LP genie

Comments

Categories