Strategy to make a 91+ LP genie

Adroit - Lost City · March 2013

Fissile - Archosaur wrote: »

The (misguided) point he is making is that for every one 80/80 try with, you will have hundreds (or thousands) of 71+/80 genies you can try with. Of course, if you try with that many 71/80 genies you have a much greater chance of successfully reaching 91/100. This is backed up by Adroit's own simulation using the naive method (keep going til it is impossible to succeed).

But in doing that, you've also raised the spirit cost by 70% compared to the other strategy. Which is worth more, 10,000 mirages or 700 million spirit (2 months of daily PV, 2 tokens per day, with 4-6 runs per token + about 5-10 hp charms [1-2 charm ticks per run], since a wiz is going to have to take a beating to do runs that fast + any hp and mp food)?

yeah, I actually just realized I derped in my second simulation.. was continuing if currentLevel - currentLP < 9.. should have been < 10. I'm going to see if I can make this script multi-threaded, it takes this laptop several hours to make 10k 91+ genies.. and the processor usage sits around 5%.. would be nice if it didn't take quite as long lol (especially when I leave it on when I go to bed and forget to disable sleep mode.. so it still has to run when I get up in the morning

)

serbetel · March 2013

Adroit - Lost City wrote: »

yeah, I actually just realized I derped in my second simulation.. was continuing if currentLevel - currentLP < 9.. should have been < 10. I'm going to see if I can make this script multi-threaded, it takes this laptop several hours to make 10k 91+ genies.. and the processor usage sits around 5%.. would be nice if it didn't take quite as long lol (especially when I leave it on when I go to bed and forget to disable sleep mode.. so it still has to run when I get up in the morning )

Your making it unnessarily hard that s why it take that much time.

Just let the script produce all valid combinations
then set 80/80 in relation to the number of valid permutations.

If you want to find out the statistical chance
you have to calculate all invalid combinations too.

But there s nothing to simulate its all fixed combinations.

The chances at lowler level really dont matter
since all those lower level MUST reach into the harder level to hit the goal.
Thats why i repeat the word "symbol" that much.

LP per decade are nothing then a part of code wich has absolutely nothing to do with maths.

Adroit - Lost City · March 2013

serbetel wrote: »

Your making it unnessarily hard that s why it take that much time.

Just let the script produce all valid combinations
then set 80/80 in relation to the number of valid permutations.

If you want to find out the statistical chance
you have to calculate all invalid combinations too.

But there s nothing to simulate its all fixed combinations.

You're wrong. I'd explain why but judging by the rest of this thread I'd just be wasting my time. Please just stop posting in this thread, you aren't helping.

serbetel · March 2013

Adroit - Lost City wrote: »

You're wrong. I'd explain why but judging by the rest of this thread I'd just be wasting my time. Please just stop posting in this thread, you aren't helping.

It did help a huge amount of player s allready while your "simulation" is just an example of someone trying to learn stochastic.

Fissile - Archosaur · March 2013

Adroit - Lost City wrote: »

Hmmm, well this is taken from the docs of the random number generator I'm using (and I haven't added reseeding)

"If reseeding is never done by the script, the seed starts off as the low-order 32-bits of the 64-bit value that is the number of 100-nanosecond intervals since January 1, 1601. This value travels from 0 to 4294967295 every ~7.2 minutes."

Tbh I'm not really familiar enough with random number generators to know if this is sufficient or not, I should probably do a little more reading to find out. That does sound like a good idea to keep track of 71+/90 genies, I'll probably add that tomorrow or something. Thanks for the feedback

Given the 32-bit initialization it looks like it is probably a 32-bit implementation of a Linear Congruential (linear function) or something similar. There are 1,000,000,000 possible strategies for producing 91+ genies. If you set out to test them all (at millions of genies per strategy to get good statistics) you would go far beyond the usefulness of a 32 bit (based on its internal state) PRNG.

serbetel · March 2013

Fissile - Archosaur wrote: »

Given the 32-bit initialization it looks like it is probably a 32-bit implementation of a Linear Congruential (linear function) or something similar. There are 1,000,000,000 possible strategies for producing 91+ genies. If you set out to test them all (at millions of genies per strategy to get good statistics) you would go far beyond the usefulness of a 32 bit (based on its internal state) PRNG.

+1
Not to speak of precalculated rainbow tables wich might speed things up radical.
Decimal system is not rocket science.

Adroit - Lost City · March 2013

Fissile - Archosaur wrote: »

Given the 32-bit initialization it looks like it is probably a 32-bit implementation of a Linear Congruential (linear function) or something similar. There are 1,000,000,000 possible strategies for producing 91+ genies. If you set out to test them all (at millions of genies per strategy to get good statistics) you would go far beyond the usefulness of a 32 bit (based on its internal state) PRNG.

The docs say that it uses the Mersenne Twister random number generator, but I'm not sure if that changes anything. I'm not planning on trying out every strategy to produce 91+ genies, I was thinking I'd just guess a few strategies (I doubt I'd be patient enough to try more than 20 or so) and then just pick the best of my guesses. I'll do a little reading on this stuff tomorrow, kinda tired atm :P

soundslegit · March 2013

I think its as simple as this: calculate how many genies at the current level(divisible by ten ofc) could be made with the spirit needed to move the genie up 10 more levels, fractional amounts are fine. Multiply this by the percentage likelihood of a newly made genie having a better chance of reaching 91+ than the current genie. If the result is greater than 1 then decomp or sell current genie. If lower then level it up 10 more levels.

Vetrobor - Raging Tide · March 2013

You can bla-bla theory the whole day,
practice is what gets results.

Yes, theory is good preparation for practice,
but overdoing gives nothing good.

I don't know any way better than:

Make first genie, see if it has 20/20 or 30/30,
if not, convert it to EXP cube, then use cube to
start new genie and spare some SP on lower
Genie levels.

Some of Genies will get to 40/40, some to 50/50,
and so on. Numbers you determine, percentage
can gradually lower from 100% at low Genies
to 95% at lvl 60 genies, to 91% at lvl 90.

I think that after lvl 40 you need to switch Genie
to TRADE state to be able to decompose into XP
cube, and that takes 7 days of waiting (168 hours).

Also, cube and Genie level may not be too afar
or you get nerfed XP infused.

You will discover yourself when to infuse with SP
and when with cube.

~~~~~

BTW, at lvl 105 you can farm Warsong mobs for
Essences for more Genie XP. Simply join squads
and help people with Warsong BH's, and when
finish Snake and Cannon and they leave you
go on with Earth and Wood/Water paths.

You have to do something while waiting for TRADE
state to become active.

~~~~~

Yes, you know most of this, but others read this thread as well.

Another thing: one guy complained that his SP reached
max number reserved by database and after that went
to negative numbers and his SP was scr*ed.

I don't know if that number is around 1.1 billion
or bigger, but have in mind such thing, just in case...

serbetel · March 2013

To all those 20/20, 30/30 .. 80/80 fanatics:

+ Did you reach to "break" TT red drops yet ?
+ Did you reach to "break" lotterys like coral yet ?
+ Did you reach to "break" refinary yet ?

If you not full of red TT mats, allways get the 1st prize in lotto and made your +12 in first random try (no orbs),
then you might consider that genie system works exactly the same.. non predictable.

Vetrobor - Raging Tide · March 2013

serbetel wrote: »

To all those 20/20, 30/30 .. 80/80 fanatics:

+ Did you reach to "break" TT red drops yet ?
+ Did you reach to "break" lotterys like coral yet ?
+ Did you reach to "break" refinary yet ?

If you not full of red TT mats, allways get the 1st prize in lotto and made your +12 in first random try (no orbs),
then you might consider that genie system works exactly the same.. non predictable.

After 7 toons I finally made Cleric.

Got Zeal and after some time had 60/60.

Then got 61/70 and got so "disappointed". LOL

Now my genie is still 70/90 and I don't complain.

I will keep it as good Genie, because the worst case
will be 71/100 and the best 80/100, which is very good.

But some day at lvl 105 I might need to spend extra SP.
To think of even better Genie is the only thing to do with
"630 mill SP".

It may fail totally, but do you have better use of SP
when you finally learn every skill that can be learned?

See my previous post... hehe

serbetel · March 2013

No, the way you play is best bet imo.

b:victory

HrunsPanda - Archosaur · March 2013

serbetel wrote: »

To all those 20/20, 30/30 .. 80/80 fanatics:

+ Did you reach to "break" TT red drops yet ?
+ Did you reach to "break" lotterys like coral yet ?
+ Did you reach to "break" refinary yet ?

If you not full of red TT mats, allways get the 1st prize in lotto and made your +12 in first random try (no orbs),
then you might consider that genie system works exactly the same.. non predictable.

Non-predictable ..... Well no, but does it not matter to you if you have 10% chance or 1% for something you want ?

serbetel · March 2013

HrunsPanda - Archosaur wrote: »

Non-predictable ..... Well no, but does it not matter to you if you have 10% chance or 1% for something you want ?

It does but,
the assumption that full LP genie s are the better % chance
is simply wrong.

People missinterpret the numbers,
if it would be letters they would understand easier.

Imagine those LP sequences would be a code you want to break
and now you have the choice to throw in 1000 codes or just 1
what % chance would you expect then ?

The other aspect as i described earlier is the progressive amount of time and costs
if i would say "a level 80 genie cost twice the cost of a level 70"
ppl might think over again if all those "low number" genie s should be discardet..

Edit:
Final statement - The chance that the "Queen" will be 10/10/10/10.10/10/10/10
is like more then 99,9 % lower then the chance it will be ANY other valid combination.
(thats just approximation, most prly you can add some more 9s behind, im just to lazy to calculate something wich i allready know)

Decus - Dreamweaver · March 2013

serbetel wrote: »

Edit:
Final statement - The chance that the "Queen" will be 10/10/10/10.10/10/10/10
is like more then 99,9 % lower then the chance it will be ANY other valid combination.
(thats just approximation, most prly you can add some more 9s behind, im just to lazy to calculate something wich i allready know)

You'll forgive me as probability theory was never a branch I had any interest in, nor do I pretend to know much about it, but I have a really hard time understanding you. I am not sure if English is your native language or not, but it's difficult to interpret your posts and I'd like to actually understand what you're saying. Several others keep saying you're wrong, yet you seem quite convinced about your knowledge in the subject matter, and I can't really get around to passing judgement :P

Valirah - Sanctuary · March 2013

serbetel wrote: »

It does but,
the assumption that full LP genie s are the better % chance
is simply wrong.

No, it's not.

serbetel wrote: »

People missinterpret the numbers,

Yes, you're one of those people.

A 9/10 genie has a lower chance at being 91+ than a 10/10, I don't understand why you can't see this.

The 10/10 can still make it even if one of the next 9 rolls is a 1, the 9/10 can't. The 10/10 will make it on every set of rolls that the 9/10 will make it on in addition to a bunch of rolls that the 9/10 won't make it on. The 10/10 is clearly more likely to provide the desired result.

Adroit - Lost City · March 2013

I re-did my previous calculation that was flawed last night (while still possible), the new results for 10,000 91+ genies is:

Average LP value: 93
Average Mirage Cost: 19515
Average SP Cost: 2002m (with a standard deviation of 1198m SP)
Average # of 81-90 LP genies: 1.93
Proportion of these genies that were made with 630m or less spirit is: ~.263

I'm playing with my script atm, but I'll see if I can keep track of a few extra things and then try a few other simulations :P

Porkfruit - Lost City · March 2013

Instead of comparing these various builds and calculating their average spirit costs why not approach from the other end?

Decide on an avg amount of spirit you are willing to spend and build your levling rules around that?

Adroit - Lost City · March 2013

Porkfruit - Lost City wrote: »

Instead of comparing these various builds and calculating their average spirit costs why not approach from the other end?

Decide on an avg amount of spirit you are willing to spend and build your levling rules around that?

I would love to do that, just not sure how. I think the closest thing to that I've seen in this thread would be soundlegit's suggestion, which I'm thinking I'll try to simulate next.

HrunsPanda - Archosaur · March 2013

serbetel wrote: »

It does but,
the assumption that full LP genie s are the better % chance
is simply wrong.

People missinterpret the numbers,
if it would be letters they would understand easier.

Imagine those LP sequences would be a code you want to break
and now you have the choice to throw in 1000 codes or just 1
what % chance would you expect then ?

The other aspect as i described earlier is the progressive amount of time and costs
if i would say "a level 80 genie cost twice the cost of a level 70"
ppl might think over again if all those "low number" genie s should be discardet..

Edit:
Final statement - The chance that the "Queen" will be 10/10/10/10.10/10/10/10
is like more then 99,9 % lower then the chance it will be ANY other valid combination.
(thats just approximation, most prly you can add some more 9s behind, im just to lazy to calculate something wich i allready know)

It is hard to comprehend what you are saying. But what i do make out of it is lvl 80 genie is twice the cost of a lvl 70 and people thinking twice about discarding low level genies. That is contradictory. The exponential rise in Xp cost is exactly the reason to discard them. The early levels are so incredibly cheap that the only reason to accept anything less than perfect there is the size of your inventory filling up with discarded genies.

As i say, it is very hard to comprehend your post. I do have an idea what your line of thought is, but it is speculation: Did you mean to say something like the chance of your genie reaching 91 points by getting 1-10-10-10-10-10-10-10-10-10 points is equal to it gettig there by getting 10-10-10-10-10-10-10-10-10-1 ? If that is your excuse for not throwing away the level 1-10s, you are wrong because throwing away the 1-10s may reduce the 91/100s by a tiny bit, but it reduces the failures by a huge amount.

serbetel · March 2013

Maybe the OT should just change the thread title to
"Howto get a 91 LP Genie"

Prly then you understand that getting a 91 LP
is multiple 100 % easier then a 92, 93, 94 .. 100 LP Genie.

By unnessecarely trying for more then 91 LP
you drop the only factor wich you can have INFLUENCE on.

Sorry that i cannot explain better,
its the nature of stochastic =\

I simply set the limits for the formula
wich should allways be first step before even starting to create a statistical formula.

Asterelle and the OT didnt set such limits,
but set there "influence" to be a momentum.
(a non reproduceable timeframe of a stream of timeframes)

Those table are good when you dont know what is in Lucky coral
and after like 5000 tries you have a 99,9 % working table
wich shows every possible item.

The imagination that the amount of those items would matter in any way is simpley introducing another magic number
not making it easier since you cannot reproduce the results.
(Statistics try to avoid those random numbers since every new one can make the result exponentially less predictable)

The only factor wich you have influence is the minimum and maximum of LP
and its simply an illusion that a A.x.x.x x.x.x.x
would have a lower chance then a x.x.x.x x.x.x.A
or x.x.x.x A.x.x.x.

It s more like a Bingo game.

Adroit - Lost City · March 2013

So I tried another strategy (was out all day, didn't have time to modify anything else on my script) but I tried a strategy where until level 30, I only continued on perfect genies, and then after that used the 90% rule (must have 90% of its possible lucky points to move on). The thought was that it was cheap to level up lowbie genies, so may as well start off perfect and then be able to be less picky later on.. but this was not an effective strategy at all. I leveled 10,000 genies using this strategy to find:

Average LP value: 94
Average Mirage Cost: 151134 <- not a typo.. 151k mirages on average..
Average SP Cost: 2203m (with a standard deviation of 1279m SP)
Average # of 81-90 LP genies: 1.46
Proportion of these genies that were made with 630m or less spirit is: ~.239

Looking back, this does make some sense. The chance to get to 91+ from a 19/20 genie vs a 20/20 genie is fairly similar, I shouldn't have been tossing the 19/20s. Definitely not a good plan lol

HrunsPanda - Archosaur · March 2013

hmm ye, i always considered the mirage cost insignificant and only thought about the exp / sp. Thats wrong i guess b:thanks

So probably just go for 9 or 10 points on every round. And on the higher levels when you have gotten more often 9 than 10 points, you may open up some space for accepting some of those rare 8s and 7s. Or should we start accepting those early on as well ? (since when you take only 9s and 10s, you are gonna end up on average with something like 95/100)

I hope noone is advocating to accept 5/10 or 15/20 and worse, because that i am certain is not a good plan if you want to reach 91.

serbetel · March 2013

Time and slots (stash/bag) is the major factor

Adroit - Lost City · March 2013

serbetel wrote: »

Time and slots (stash/bag) is the major factor

Maybe time, but bag slots isn't an issue. Can just account stash level 40+ genies for a week and decomp them after. I'm not planning on doing this all in one day, I'd probably just level a few dozen genies a day when I'm bored and see if I get lucky. Not really in a rush for it, just wanna know the best way to try my luck :P

Tide_Surfer - Archosaur · March 2013

Adroit - Lost City wrote: »

Maybe time, but bag slots isn't an issue. Can just account stash level 40+ genies for a week and decomp them after. I'm not planning on doing this all in one day, I'd probably just level a few dozen genies a day when I'm bored and see if I get lucky. Not really in a rush for it, just wanna know the best way to try my luck :P

Just wondering but is that crit in your sig actually yours O_O

Adroit - Lost City · March 2013

Tide_Surfer - Archosaur wrote: »

Just wondering but is that crit in your sig actually yours O_O

Yup, was on yan the traitor a few weeks back. I'm fairly confident I could hit over 10m on the right mob (namely one that ice prison works on).. just need to be bored enough one day and get a few people to help debuff/amp the mob for me :P

serbetel · March 2013

HrunsPanda - Archosaur wrote: »

hmm ye, i always considered the mirage cost insignificant and only thought about the exp / sp. Thats wrong i guess b:thanks

So probably just go for 9 or 10 points on every round. And on the higher levels when you have gotten more often 9 than 10 points, you may open up some space for accepting some of those rare 8s and 7s. Or should we start accepting those early on as well ? (since when you take only 9s and 10s, you are gonna end up on average with something like 95/100)

I hope noone is advocating to accept 5/10 or 15/20 and worse, because that i am certain is not a good plan if you want to reach 91.

Now you are introducing a break-even-point for 91-95 LP genies.

The thing you leave out is that those "weak" genie s you are certain not to be good
fill your bag/stash anyway but the real factor is the time to ged rid of those.

It s no problem to wait cooldown for 200 or 400 over weeks,
but if you make it harder to get 91 LP by a factor of 1000 those weeks become years ..

Bhavyy - Raging Tide · March 2013

serbetel wrote: »

Now you are introducing a break-even-point for 91-95 LP genies.

The thing you leave out is that those "weak" genie s you are certain not to be good
fill your bag/stash anyway but the real factor is the time to ged rid of those.

It s no problem to wait cooldown for 200 or 400 over weeks,
but if you make it harder to get 91 LP by a factor of 1000 those weeks become years ..

What are you on about? I really don't understand anything you're trying to say here. I'm sure you have some sort of point on what you're saying, at least I hope you do. Could you just be a little better at explaining? Or perhaps someone could translate the randomness.

And I think the best way to do it would be to set a limit at 10 level intervals on what you'd settle for to go on or not go on to the next 10 levels.
E.g. My limits would be something like 28/30, 37/40, 46/50, 55/60, 64/70, 73/80, 82/90, 91/100.

Seems pretty basic I'm sure. I wonder if there's a way of calculating the best limits.

HrunsPanda - Archosaur · March 2013

Bhavyy - Raging Tide wrote: »

What are you on about? I really don't understand anything you're trying to say here. I'm sure you have some sort of point on what you're saying, at least I hope you do. Could you just be a little better at explaining? Or perhaps someone could translate the randomness.

And I think the best way to do it would be to set a limit at 10 level intervals on what you'd settle for to go on or not go on to the next 10 levels.
E.g. My limits would be something like 28/30, 37/40, 46/50, 55/60, 64/70, 73/80, 82/90, 91/100.

Seems pretty basic I'm sure. I wonder if there's a way of calculating the best limits.

Indeed, a limit at every interval is logical.

It is however not so easy to calculate because it depends on some values. We have a few resources that we can consider scarce. XP/SP, mirages, clicking time.

Now if you would consider only the mirage cost, but not the other 2, then obviously you'd level every genie until it has missed more than 9 points.

If however you'd consider mirages of no value, and only look at XP/SP and clicking time, then youd probably ditch your 9/10s, 19/20s and 29/30s up till a certain point.

So you are gonna have to assign each of these 3 resources a value that can be measured and compared towards eachother to fit into one formula. It will be different from person to person. Best we could do is make a spreadsheet where you can enter your values for XP and clicking time (in coins) and make it calculate on that. But i would honestly have no idea what to value my XP at.

So i think since there is not going to be a simple answer that is good for everyone. I guess we best just keep using the intuitive way and use those limits like you posted.

Someone may use 27 for a limit another 28 oh well.....

Strategy to make a 91+ LP genie

Comments

Categories