How to make the cauldron work?

Share your ideas on how to improve Shop Heroes.
Forum rules
Please check other threads before posting a suggestion or providing feedback to make sure that there are no duplicate threads on the same subject. We value your feedback and suggestions and will prioritize those with the most demand, however, please be patient with us as we already have a very packed pipeline for new content delivery. Any feedback that has already been taken into account will result in the thread being locked as further discussions are no longer necessary.
Molly
Posts: 30
Joined: Wed Jan 18, 2017 7:41 pm

Re: How to make the cauldron work?

Postby Molly » Sat Apr 08, 2017 7:07 am

Sillvestra wrote:When fusing, I use a slightly different strategy.
2x good for greats
2x great for flawless
CF fusions for Epic


As you can see, I doing the same. But usually I get more green items then epics. Sometimes - much more.

Molly
Posts: 30
Joined: Wed Jan 18, 2017 7:41 pm

Re: How to make the cauldron work?

Postby Molly » Mon Apr 10, 2017 7:28 am

4 epics instead of 10 during 3 days!

http://dl4.joxi.net/drive/2017/04/10/0001/4033/114625/25/ebd99fb9f1.png

Its never ends. I repeat: I never got similar double amount of epics. Only RARE occasions of 3 epics in the row, leading with new bunch of greens or flawless.

And I still haven't any dwarfish/primal blueprints after spending 10k+ gems.

All what they can say: "We understand how frustrating it could be".
I'm at a dead end.

Molly
Posts: 30
Joined: Wed Jan 18, 2017 7:41 pm

Re: How to make the cauldron work?

Postby Molly » Sun May 21, 2017 8:22 am

For the first time I managed to capture an adequate number of epic items in 30-50 fusions (same items are fused)!

http://i.imgur.com/2t7yUmK.png


After the last series of failures, I generally stopped fuse according to this scheme. And I decided to check how things are now.

VeNoM0619
"Great" Mastery Member
Posts: 139
Joined: Wed Apr 19, 2017 2:07 am

Re: How to make the cauldron work?

Postby VeNoM0619 » Sun May 21, 2017 5:32 pm

Well good, looks like you are happier with the results :)

My overall experience with the cauldron have felt amazing, however making a post about it would seem like I'm bragging, therefor only the bad experiences will get reported on. I also know of the fallacy that good experiences are remembered less than bad ones, here is just one of the many articles on it: http://www.livescience.com/1827-bad-memories-stick-good.html


However... if you wish to dig into the RNG algo, I posted this elsewhere, and some assumptions on which variant of the algo need to be made as well since its a black box (until someone decompiles it).
Unity is using the Marsaglia's Xorshift 128 https://en.wikipedia.org/wiki/Xorshift (this was only mentioned in 1 post, by a non Unity member)


And as always: RNG isn't perfect, its pseudo.

Molly
Posts: 30
Joined: Wed Jan 18, 2017 7:41 pm

Re: How to make the cauldron work?

Postby Molly » Tue May 23, 2017 12:57 am

VeNoM0619 wrote:I also know of the fallacy that good experiences are remembered less than bad ones

This postulate is repeated almost in every message.

But there is a little more than just a feeling.

You notice the rare / unusual combinations well. And you find that they happen inadequately often. And, mostly negative.

Once (in Jan or Feb) I got angry because of the large amount of greens. Then I got a bunch of muskets. Threw them into the cauldron. And immediately took out five epics in a row! This happened only once throughout my game. And it was on muskets.

But it was easy for me to record this combination:

Molly wrote:4 epics instead of 10 during 3 days!
http://dl4.joxi.net/drive/2017/04/10/0001/4033/114625/25/ebd99fb9f1.png


4 epics and 14 of green+great!

Let's simulate these 35 fusions with 30% for epic, 50% for flawless, 15% for good and rest for great. Create text file "test.html" with the following content:

Code: Select all

<html>
<head><title>test</title></head>
<body>
<script>
      // chances (chance for great not need to be specified)
      
      var cGood = 0.15;
      var cFlaw = 0.50;
      var cEpic = 0.30;

      // amount of fusions
      
      var n = 35;

   // counters

   var good = 0, great = 0, flaw = 0, epic = 0;

   // lets go
   
   document.open();
   document.write("<pre>");
   document.write("=================\n");

   for(var i = 0; i < n; i++)
   {
      var r = Math.random();
      if(r < cGood)
      {
         good++;
         document.write("| g |   |   |   |\n");
      }
      else if(r < cGood + cFlaw)
      {
         flaw++;
         document.write("|   |   | f |   |\n");
      }
      else if(r < cGood + cFlaw + cEpic)
      {
         epic++;
         document.write("|   |   |   | e |\n");
      }
      else
      {
          great++;
         document.write("|   | g |   |   |\n");
      }
   }

   // totals
   
   document.write("=================\n");
   document.write("good:  " + good + " items\n");
   document.write("great: " + great + " items\n");
   document.write("flaw:  " + flaw + " items\n");
   document.write("epic:  " + epic + " items\n");
   document.write("</pre>");
   document.close();
</script>
</body>


Run it. Look at the result. Press F5. Look at the result. Press F5. Look at the result. And so on.

You will find
- how difficult it is to catch "only 4 or less epics" combination
- how difficult it is to catch "14 or more greens" combination
- and how IMPOSSIBRU! to catch "4 or less epics" AND "14 or more greens" (I did not succeed)

So simple RNG isn't perfect, but its fine in general.
And RNG in SH (seems) isn't uniform and (perhaps) even has undesirable side effects.
Anyway, I clearly see that RNG in SH is different.

Kim!
"Great" Mastery Member
Posts: 137
Joined: Sun Aug 07, 2016 11:01 pm

Re: How to make the cauldron work?

Postby Kim! » Tue May 23, 2017 3:03 am

- and how IMPOSSIBRU! to catch "4 or less epics" AND "14 or more greens" (I did not succeed)


Done:
Image

And for extra credit:
Image

I didn't edit these images or change the probabilities in your code, simply modified it to loop a couple times. While it's improbable to get such low results it's certainly not impossible. It's an RNG, it does what it wants. And that means sometimes it decides to pull awful results.

VeNoM0619
"Great" Mastery Member
Posts: 139
Joined: Wed Apr 19, 2017 2:07 am

Re: How to make the cauldron work?

Postby VeNoM0619 » Tue May 23, 2017 11:03 pm

I'm uncertain what RNG algo is used in webpages, so no idea on its startup/init routine or its period lengths (ie flaws). However, lets also point out in that data:

We had a 5% chance at great, and both rolls above had 0 out of those rolls when you expect 1-2 of them. The fact that many glanced over that (you missed that, didn't you? :P )and only focused on epic chances, just shows the bias we all have, correct? We only looked at epics. We didn't care about the odds matching up on everything else.

Unfortunately, RNG algos are usually tested with large period lengths, and the more you roll, the more the values start swaying towards the % you expect. 40 isn't enough... I'm talking thousands-millions. Testing the algos this way is part of the reason why everyone thought MT was great, they tested large amounts of data, but missed the broken startup routines it spit out or the dead zones. Minor flaws are constantly being found. The period lengths can also cause an "ebb and flow" system.

Will you get CC to write their own rand() function? Unfortunately no, because xorshift is *currently* the best pRNG known, there are others out there... but we can't task them to do that research.
Will they write a pool routine to help reduce period lengths and help match player odds expectations? No, no game ever has sadly.

I agree, it feels unfair, but RNG feels unfair in every game. (Although I remember playing a game that would say 30% but actually did like 35-40% on the better items, and nobody seemed to complain or felt wronged).


Return to “Suggestions & Feedback”



Who is online

Users browsing this forum: No registered users and 2 guests

©2015 Cloudcade, Inc. All Rights Reserved.