Login  |  Register

Author Topic: Statue probability  (Read 838 times)

0 Members and 3 Guests are viewing this topic.

Offline LilMiss

Re: Statue probability
« Reply #15 on: May 12, 2019, 12:43 AM »
Correct, no algorithm.  Just obeer action. I am tracking across multiple accounts to maximize the sample size, though I control all of the accounts and I do not omit any results, so as not to introduce bias into the study.


Again, I can’t follow everything and will read this tomorrow when I have time.

However, have you kept these stats separate or collated them?

And are they all maxed accounts or at different stages?
Phoenix Family

Offline LilMiss

Re: Statue probability
« Reply #16 on: May 12, 2019, 06:28 PM »
ok so,

Sorry, you are not allowed to see this part of the text. Please login or register.


This is the chart with info that is provided by SC


There is a little note on the Yessup's post that I hadn't noticed before.


Quote
[font=]  [/font][font=&amp][/font]
[font=&amp]My Statue has a different number than those in your table - is your table wrong?    [/size][/font]
The actual number may be off by +/- 1%. The way it works is that there is base boost % for all the statues, and each type is multiplied by the boost multiplier factor. For example, for Gunboat energy, it is 1.7x, while for troop damage, it is 1.3x.  As decimals are not used, the numbers are rounded to integers, so there is potential round off error. Interestingly, the rounding off is not very consistent in the actual numbers - sometimes, Supercell seems to use round up, while in most cases it uses round down. That's why you may find some statues off by 1%.




So, the table isn't an exact science.


I asked above whether you were collating the answers for your table. The reason is because the way that statues are designed by SC isn't as simple as what may first appear.  When you start the game your statues are pre-determined. That's why people statue peeked in the good ole days before the Statue Storage. Every so often SC mixes it up and re-rolls them. How SC decides I have no idea.


However, as statues are predetermined it could well be that the  sequence is correct, for each account, however, you can be lucky, or unlucky as to which part you are in.

Phoenix Family

Offline Bismarck

Re: Statue probability
« Reply #17 on: May 19, 2019, 04:20 AM »
Many mentioned my low sample size. At the time I posted this thread, I had 323 observations for idols, 28 for guardians, and 8 for MPs.  379 total.  That is large enough to be statistically significant in the case of idols. Nevertheless I am pushing forward and explanding the sample.

At present I have a total of 620: 495 idols, 86 guardians, and 39 MPs.
Angry Bismarck - TD32, TH 33, GBE 40, PSC 72, PSC 30, PSC 30, PSC 29, PSC 29, PSC 29, PSC 28
Leader, Angry Blizzard

Bismarck- TD 31, TD 13, TD 13, TD 13, TD 13, TD 13, TD 13, TN 35, TN 14, GBE 42
2 Legendary + 10 Diamond Crab Trophies

Offline Bismarck

Re: Statue probability
« Reply #18 on: May 19, 2019, 04:23 AM »
One of my accounts has 7 PSC, so no shortage of observations in waiting.


Sorry, you are not allowed to see this part of the text. Please login or register.
" alt="" class="bbc_img" />
Sorry, you are not allowed to see this part of the text. Please login or register.
Angry Bismarck - TD32, TH 33, GBE 40, PSC 72, PSC 30, PSC 30, PSC 29, PSC 29, PSC 29, PSC 28
Leader, Angry Blizzard

Bismarck- TD 31, TD 13, TD 13, TD 13, TD 13, TD 13, TD 13, TN 35, TN 14, GBE 42
2 Legendary + 10 Diamond Crab Trophies

Offline Bismarck

Re: Statue probability
« Reply #19 on: May 19, 2019, 04:28 AM »
I don't think my images are showing up and I have no idea which coding language this forum uses.  Fritznelly, you edited one of my posts to add images.  What code did you use for this?
Angry Bismarck - TD32, TH 33, GBE 40, PSC 72, PSC 30, PSC 30, PSC 29, PSC 29, PSC 29, PSC 28
Leader, Angry Blizzard

Bismarck- TD 31, TD 13, TD 13, TD 13, TD 13, TD 13, TD 13, TN 35, TN 14, GBE 42
2 Legendary + 10 Diamond Crab Trophies

Offline Bismarck

Re: Statue probability
« Reply #20 on: May 19, 2019, 04:34 AM »
Lil Miss, obsevations are recorded for 10 accounts but the vast majority come from 3, which are max.  Results are collated.  I am aware that build orders are predetermined as lists at account creation but I believe the contents of those lists are random.  Thus, aggregate data can be used.
Angry Bismarck - TD32, TH 33, GBE 40, PSC 72, PSC 30, PSC 30, PSC 29, PSC 29, PSC 29, PSC 28
Leader, Angry Blizzard

Bismarck- TD 31, TD 13, TD 13, TD 13, TD 13, TD 13, TD 13, TN 35, TN 14, GBE 42
2 Legendary + 10 Diamond Crab Trophies

Offline Bismarck

Re: Statue probability
« Reply #21 on: May 19, 2019, 06:27 AM »
There is a little note on the Yessup's post that I hadn't noticed before.

So, the table isn't an exact science.

It’s not just +/- 1%. I’ve recorded multiple instances in which statues were off by 2%. For example, Wood Production Masterpiece. The lowest tier is 30%, with the next tier being 35%. I have recorded one instance of a 32%. So I’m not sure how accurate these rounding theories are.
Angry Bismarck - TD32, TH 33, GBE 40, PSC 72, PSC 30, PSC 30, PSC 29, PSC 29, PSC 29, PSC 28
Leader, Angry Blizzard

Bismarck- TD 31, TD 13, TD 13, TD 13, TD 13, TD 13, TD 13, TN 35, TN 14, GBE 42
2 Legendary + 10 Diamond Crab Trophies

Offline Chaucer

Re: Statue probability
« Reply #22 on: May 20, 2019, 05:04 PM »
I guess I always interpreted the statue probability chart differently.
I say the chart only as a list representation of the base algorithm. So the chart for GBE MPs listed 10 selected values from the 20%-42% range, despite the actual possibility of getting any value within that range.

Perhaps my understanding is wrong, but it does appear that reality seems to be that statue values not in the chart are possible.

Offline Artist formerly known as Crows Nest

Re: Statue probability
« Reply #23 on: May 20, 2019, 10:37 PM »
Something doesn’t add up for me.

This is the guide/data that we are basing things on which I have just read through (all 26 pages)
Sorry, you are not allowed to see this part of the text. Please login or register.


There are several anomalies identified previously in this guide.
1. Why do the masterpieces add up to 100.7% whereas all others are 100%?
2. Why is the “rounding” so random? Does it really exist?
3. The origin of the data in the guide is from “.csv file from boom beach apv” created when installed (p22). is it still there?
4. Peeking implies that the RNG is predetermined just for statues, does that make sense and does it still work?

Basically this guide doesn’t hold together for me, it seems clumsy in that the process described is overly complicated for no reason.
No disrespect to the original poster of the guide but while it seems to cover/explain things there is a lot of unanswered issues that causes me to distrust it for the current game. Why would SC keep such a process unchanged if they are rewriting code?

As a first step I suggest asking someone like Fares to see if the statue info is still accessible in the game files/code and accurate to this guide.
« Last Edit: May 21, 2019, 05:46 PM by Artist formerly known as Crows Nest »
1️⃣👥💕🆕💥🏝

Offline fritzelly

  • Administrator
  • Administrator
  • Confirmed Know-It-All
  • *****
  • Posts: 4,416
  • Total likes: 88
  • Gender: Male
Re: Statue probability
« Reply #24 on: May 20, 2019, 11:15 PM »
RNG in computers is inherently not very random or is the fact that the same numbers keep coming up mean it is totally random - needs deep thought on that one

Where do you get the 100.7% from?
Legends - #9YUCCYQ2
Top 400 and climbing - DE and above

Offline Artist formerly known as Crows Nest

Re: Statue probability
« Reply #25 on: May 20, 2019, 11:20 PM »
Yes RNG is not really random but for statues there seems to be no random element beyond initial generation of...what? A text file of random numbers? A lot of people think that your statue RNG is generated at install or major update, that seems so clumsy and inefficient, would such a thing still exist in game with rewrites of code?

The probabilities for each idol and guardian add up to 100% but the masterpieces add up to 100.7%, why the difference?


My main point is that when data results don’t equate to expected assumptions then best to recheck the assumptions.
« Last Edit: May 20, 2019, 11:24 PM by Artist formerly known as Crows Nest »
1️⃣👥💕🆕💥🏝

Offline fritzelly

  • Administrator
  • Administrator
  • Confirmed Know-It-All
  • *****
  • Posts: 4,416
  • Total likes: 88
  • Gender: Male
Re: Statue probability
« Reply #26 on: May 20, 2019, 11:37 PM »
I'm guessing the percentage discrepancy comes from an algorithm to split the chances on a degrading scale and no one bothered to check if it still added up to 100% - maybe a fault of using stuff from CoC


I don't have time to work it out now but that would be my first guess, I know for sure they do use similar in other aspects of the game
Legends - #9YUCCYQ2
Top 400 and climbing - DE and above

Offline Chaucer

Re: Statue probability
« Reply #27 on: May 21, 2019, 03:39 PM »
I think the assumption that needs checking is that the chart is a perfect representation of all possibilities.Given the fact that one of my accounts has a 41% GBE MP (which doesnt appear on the chart).
As I said previously, seems a better explanation is that statues are generated by an algorithm which would more accurately be represented by plotting a curve.Pulling out select data points to populate a table will only be an approximation of reality.


As has been reported, the fact that players can statue peek implies that something is pre-determined. If SC is using a pseudorandom number generator to generate statues, all that would be needed is the seed. So when account is created a specific seed is stored with it.

Apparently sometimes that seed number is changed with updates.Much simpler to just store the seed, than a whole text file of results from that seed.

Offline Artist formerly known as Crows Nest

Re: Statue probability
« Reply #28 on: May 21, 2019, 04:54 PM »
41% gbe doesn’t directly appear on the chart but is implied by the +/- 1% variance stated on the guide. It is this rounding that I question as have others in the original thread.

Yes the seed determines the RNG series of numbers for generating statues and is shown by the peeking. We talked about the seeding during the magic account threads. I seem to recall you can peek more than one statue in the future or am I miss remembering that? Does peeking still work?
Sorry, you are not allowed to see this part of the text. Please login or register.


But I still feel that if the original information for the guide was found in the game files then a suitable person should check the game files again to see if anything has changed.
1️⃣👥💕🆕💥🏝

Offline bill-smith

Re: Statue probability
« Reply #29 on: May 21, 2019, 05:20 PM »
I'm guessing the percentage discrepancy comes from an algorithm to split the chances on a degrading scale and no one bothered to check if it still added up to 100% - maybe a fault of using stuff from CoC


I don't have time to work it out now but that would be my first guess, I know for sure they do use similar in other aspects of the game

Maybe so. I had noticed the individual probabilities added up to 100%. My thought was that it was just a data input error, and they decided to correct things by normalizing the probabilities to 100%, i.e. in this case, you'd divide each individual probability by 1.007.