All Categories

fix drop with boss clan

fix drop with boss clan

Search
Comments
Dec 23, 2021, 17:1312/23/21
Dec 23, 2021, 17:19(edited)
11/05/20
64
Trips

So does Harle need to analyze your data and learn Russian over Xmas? :)

Actually, you don't need to learn Russian to watch this video! You are so smart and not aware of the subtitle function on YouTube?

You asked for data from me! I threw you a video where a person has a clear table for a drop from a clan boss for a year and how he has changed over the last month! Or is this still not enough for you and you will continue to deny that the drop from the boss clan has  changed?

Dec 23, 2021, 17:4212/23/21
11/05/20
64
dthorne04

Might I suggest reading back in the thread where you will find relevant information such as: "Looking into data takes time. More than that, this is the holidays and several people are on vacation." and that Quinn has in fact spoken with the people behind the RSL CB Drop Bot (and thus has an understanding of what's going on), whose bot you mentioned last week when you started this thread?

This is not my bot! It was created by people unknown to me! If you need a creator, you can find him here. https://www.reddit.com/r/RaidShadowLegends/comments/rfhklt/clan_boss_drop_changes/

Dec 23, 2021, 17:4812/23/21
12/19/19
5876
ЂǺРℳǺʎĔúĶǺ

Actually, you don't need to learn Russian to watch this video! You are so smart and not aware of the subtitle function on YouTube?

You asked for data from me! I threw you a video where a person has a clear table for a drop from a clan boss for a year and how he has changed over the last month! Or is this still not enough for you and you will continue to deny that the drop from the boss clan has  changed?

I got a sacred again today, nothing has changed for me.

 I watched 1 minute of the video saw a chest opened with a void shard then a lego book... lol.  Doesn't get much better than that.  I cannot extract the data from the video :)

If it is simply backwards looking data of one person who had a bad month.... see my request for REAL data earlier in this thread.  I can "prove" anything by cherry picking historical data.  You continue to push this with no real evidence, so yes I still deny CB drop has changed. 

Dec 23, 2021, 18:0512/23/21
11/29/20
391

I'm on a dry spell, two voids today, but epic books/ancients otherwise for the past month.  I still think it's just RNG, cause all my drops have been bad lately lol.

Dec 23, 2021, 18:0712/23/21
11/05/20
64
Trips

I got a sacred again today, nothing has changed for me.

 I watched 1 minute of the video saw a chest opened with a void shard then a lego book... lol.  Doesn't get much better than that.  I cannot extract the data from the video :)

If it is simply backwards looking data of one person who had a bad month.... see my request for REAL data earlier in this thread.  I can "prove" anything by cherry picking historical data.  You continue to push this with no real evidence, so yes I still deny CB drop has changed. 

In 2 weeks there will be a new report from the drop bot! Let's see what you say after)

Dec 23, 2021, 18:0912/23/21
12/19/19
5876
ЂǺРℳǺʎĔúĶǺ

In 2 weeks there will be a new report from the drop bot! Let's see what you say after)

Why not present the data daily to ensure no bias?  I don't need another garbage report

dthorne04Moderator
Dec 23, 2021, 18:2712/23/21
Dec 23, 2021, 18:28(edited)
12/30/20
5223
ЂǺРℳǺʎĔúĶǺ

This is not my bot! It was created by people unknown to me! If you need a creator, you can find him here. https://www.reddit.com/r/RaidShadowLegends/comments/rfhklt/clan_boss_drop_changes/

I am not sure if there is something lost in translation. So I will repeat myself, paraphrasing earlier things written in the thread, which you can go look at for yourself, too.  

I am aware it's not your bot, which is why I said that Quinn (the moderator you have been rude towards) has spoken to the two people (masha and g_o_r_l) that are behind that bot, and she knows more about it and the situation than you do. If you read back in the thread you would see that, as well as those two posting.

It will take time for to get sorted/figured due to the data involved and it being the holidays/Christmas.

With that said, please be patient, courteous and perhaps step away while things get sorted. Continuing to post angrily/rudely is helping no one.

harleQuinnModerator
Dec 23, 2021, 18:3712/23/21
02/24/19
7109
ЂǺРℳǺʎĔúĶǺ

Good evening Harle! Check out this video, please! The Russian-speaking blogger on the game Rae has collected statistics for the year from the clan boss! Look at the drop he had before! and this month, 0 sacred shards! You understand 0 !!!!


I may not "understand 0", but I do understand you are frustrated with your drops.

My question to you is, why is it important that every single poster on these forums agrees with you? I've already spoken to Plarium. Several other moderators have as well. You have been heard. You no longer have the need to come on here and be rude to me, or others.

Have a good day!

harleQuinnModerator
Dec 23, 2021, 18:4412/23/21
Dec 23, 2021, 18:46(edited)
02/24/19
7109

I should add on to my post for you, @wolverisha.

More than me already reporting this, and asking you to have patience, I have spoken to the authors of the bot. Masha is very nice and we continue to have an open dialogue.

In the future, I would like to avoid the condenscension you are leveling at me, so I want to make sure you feel heard. So like I asked earlier in the thread, what else would you like from me before you feel like you have been heard?

Dec 23, 2021, 18:5112/23/21
11/05/20
64
harleQuinn

I should add on to my post for you, @wolverisha.

More than me already reporting this, and asking you to have patience, I have spoken to the authors of the bot. Masha is very nice and we continue to have an open dialogue.

In the future, I would like to avoid the condenscension you are leveling at me, so I want to make sure you feel heard. So like I asked earlier in the thread, what else would you like from me before you feel like you have been heard?

I understood everything! No more questions! I apologize if you found my statements rude to yourself or anyone else, I didn't want that!

harleQuinnModerator
Dec 23, 2021, 18:5512/23/21
02/24/19
7109
ЂǺРℳǺʎĔúĶǺ

I understood everything! No more questions! I apologize if you found my statements rude to yourself or anyone else, I didn't want that!

If this is a bug, I want it fixed for you guys. 

Now lets dance and be happy.


i


i


Dec 23, 2021, 21:5012/23/21
03/18/20
306

ok 2 days in and another 2 sacreds 1 unm and 1 nm, yesterday both were unm cb. the rng has returned in my favour hopefully i dont jinx it now though lol 

Dec 24, 2021, 10:0012/24/21
07/17/20
20
Trips

I got a sacred again today, nothing has changed for me.

 I watched 1 minute of the video saw a chest opened with a void shard then a lego book... lol.  Doesn't get much better than that.  I cannot extract the data from the video :)

If it is simply backwards looking data of one person who had a bad month.... see my request for REAL data earlier in this thread.  I can "prove" anything by cherry picking historical data.  You continue to push this with no real evidence, so yes I still deny CB drop has changed. 

If you can prove anything as you say then I suggest you look at the RAW bot data filter it as you decide best and use this historical data to 'Prove' that the clan boss drops have not changed for a proportion of users.

Are there no statistical tests you can do on unknown data to test if it has been tampered with or contains errors? For example, test that it follows the expected distribution according to a model?  There is a model of the expected data according to data mined drop rates you can use for your model. You have hundreds of users recording continuous data for over 2 months, 2 drops UNM top chest.  I am surprised you can do nothing with such a data set.

You have the Raw Data, so you are not restricted by any of the previous reports filtering choices.

Most of the user's data follow the expected distribution from the data mining model.  You can confirm or disprove this using standard statistical tests.

Some users data shows a skewed distribution, which can be modelled instead by a slightly modified model with an altered drop rate.  This can be tested on a per-user basis and you can see if the same altered model fits all the other data or if there is disagreement. 

It seems the only way anyone will believe this or not is if they do it themselves.  If you have access to a statistical math package or anything even MatLab this work is easily done.  It can be done with a bit more difficulty on Excel.

If the users data statistically follows the model predictions or the altered model predictions with an acceptable score for whatever test you use then you can either.

1. Accept the results as showing something interesting

2. Disregard the results as being deliberately fabricated from the start.  This could easily be done by creating the data from the two distributions and an appropriate pseudo-random variation.

If you choose option 2 then you just refuse to accept the data no matter what.

But can you take the data and show through filtering yourself that you can show that all users follow the same distribution of all drops?  That would be really interesting and is something that I can not do, although learning such a technique would be useful when testing other authors work for me in general


Dec 24, 2021, 11:2912/24/21
06/20/19
2184

Yes you have raw data.  Yes you have an expected.  But your raw data has sampling issues.  To begin with, data must be sampled randomly to for the required tests to have any statistical meaning.  

Dec 24, 2021, 13:1612/24/21
Dec 24, 2021, 13:18(edited)
06/21/20
7
Angwil

Yes you have raw data.  Yes you have an expected.  But your raw data has sampling issues.  To begin with, data must be sampled randomly to for the required tests to have any statistical meaning.  

Okay, lets imagine that raw data has sampling issues. So if you try to get some distribution parameter from it, you will get biased value.  Thats right?

I have just 2 comments:

1)There is no evidence that we have biased data*. Until 17th it was no bias (comparing with datamined probs). Your arguments (I did not read all of them, sorry) are not relevant from my point of view. I think so beacause I know a little bit more about bot and people using it. If you don't belive me, your arguments can be relevent, but we have no accurate way to argue with them, cause they based on some potential cases.

For example, someone can suppose that huge group of people decided to trick us: they filled correct drop for a month or two. After in one moment they decided to start swapping values for epic book and ancient shard.

Is it true? - Afaik - no, I talked to a lot of people about their drop and how they put it to bot (motivation, how accurate are they, how regular and so on), and there is no evidence for such theory. Can I prove it in scientific way? - Definitly not, it is impossible. So do we have non zero probability that this theory is true? - Yes, theoretically it is possible. Should we focus on it and reject results based on such data? - You can decide for yourselft.

*If you would like to object, that it's ru-data, you will be right. But it means you suppose that plarium give different chances for different regions. It is much more serious case then we got. We have no data to discuss it seriously. And also we can continue talking about changes on ru-server.

2) If you get two samples using one sampling method, you get 2 samples with same bias. And its distrubutions must be equal each other in case of no external changes.

For example, you want to get average age of people in US.  And you decide to ask people in hospital about their age. After processng your data, you will get some number. This number is not correct (not equal to avg age of all people in chosen country) because you really have sampling issue. But if there is no changes in medical system, ecology and so on. And you decide to repeat you exp some time later, you will get same results with same bias. And if you don't, you can speak about some extenal changes, that changes distribution of your sample.


I talkted to few people from non ru-cummunity and understood: they really can't belive that people just fill their drop daily without any reason. I am not sure that I can explain it, but a lot of people in ru-community used to save their drop results, share it, discuss with clanmates. Actually we created this bot because a lot of people saved their drop to personal excel file and we decided that bot will be more comfortable for them. And they started use bot instead of excel, they do it every day and they do it for themselfs.

Also, one (or two:) year ago some ru-streamer asked his auditory to put theirs drop in shared google doc file. A lot of people did it. After one month he got same chances as datamined and same chances as we had (based on bot data until 17 nov). His data must be much more biased and unaccurate then ours, but he got unbiased results. Magic :)

Truly speaking, I don't see any reasons for future disussions until some one else spent h(is|er) time for analyzing data and bring us some conclusions that can be discussed.

Dec 24, 2021, 13:4712/24/21
06/20/19
2184

"Okay, lets imagine that raw data has sampling issues. So if you try to get some distribution parameter from it, you will get biased value.  Thats right?"

You can't say if that's right or not. That's the problem.

"1)There is no evidence that we have biased data*. "

That's true.  What evidence do you have that the data is not biased?

"Can I prove it in scientific way? - Definitly not, it is impossible. So do we have non zero probability that this theory is true? - Yes, theoretically it is possible. Should we focus on it and reject results based on such data? - You can decide for yourselft."

I don't reject the results, i just don't see them as 'knowledge" in a scientiic sense.  You seem to agree.  However, please realize that this post was presented as PROOF, not as likely, not as suggests, not as indicates, not as most likely.  When I read words like suggest (which is what this data does), it suggests to me room for further research.

"*If you would like to object, that it's ru-data, you will be right. "

No objections.  In fact I'd be appauled if plarium singled out any group. That would be wrong.  I think it more plausible that, if there is an issue and it 'appears' there is, that it would be a coding artifact rather than intentional.

"Can I prove it in scientific way? - Definitly not, it is impossible. So do we have non zero probability that this theory is true? - Yes, theoretically it is possible. Should we focus on it and reject results based on such data? - You can decide for yourselft."

I think you are missing the target here.  I see what you are saying.  If you'd have sampled age of hospitalized individuals before the pandemic, I suspect the average age would have been significanlty lower than during the pandemic.  Are these two different populations? Yes.  So I agree with you in principle.  But if you have no control in place this may not be obvious. The situation we have is a bit different.  First, the authors did not sample the population randomly, they surveyed the population and requested self-reported data.  Second, they have no controls in place.  Third, they performed no statistical tests.

I admit there is certainly something going on here and it needs further effort to address.  What I argue against (and I love to argue) is the claim that this study proves the theory.  I do not reject the theory, but I do not agree that it has been confirmed.

In any case I wish you a wonderful season.

Dec 24, 2021, 15:0712/24/21
06/21/20
7
Angwil

"Okay, lets imagine that raw data has sampling issues. So if you try to get some distribution parameter from it, you will get biased value.  Thats right?"

You can't say if that's right or not. That's the problem.

"1)There is no evidence that we have biased data*. "

That's true.  What evidence do you have that the data is not biased?

"Can I prove it in scientific way? - Definitly not, it is impossible. So do we have non zero probability that this theory is true? - Yes, theoretically it is possible. Should we focus on it and reject results based on such data? - You can decide for yourselft."

I don't reject the results, i just don't see them as 'knowledge" in a scientiic sense.  You seem to agree.  However, please realize that this post was presented as PROOF, not as likely, not as suggests, not as indicates, not as most likely.  When I read words like suggest (which is what this data does), it suggests to me room for further research.

"*If you would like to object, that it's ru-data, you will be right. "

No objections.  In fact I'd be appauled if plarium singled out any group. That would be wrong.  I think it more plausible that, if there is an issue and it 'appears' there is, that it would be a coding artifact rather than intentional.

"Can I prove it in scientific way? - Definitly not, it is impossible. So do we have non zero probability that this theory is true? - Yes, theoretically it is possible. Should we focus on it and reject results based on such data? - You can decide for yourselft."

I think you are missing the target here.  I see what you are saying.  If you'd have sampled age of hospitalized individuals before the pandemic, I suspect the average age would have been significanlty lower than during the pandemic.  Are these two different populations? Yes.  So I agree with you in principle.  But if you have no control in place this may not be obvious. The situation we have is a bit different.  First, the authors did not sample the population randomly, they surveyed the population and requested self-reported data.  Second, they have no controls in place.  Third, they performed no statistical tests.

I admit there is certainly something going on here and it needs further effort to address.  What I argue against (and I love to argue) is the claim that this study proves the theory.  I do not reject the theory, but I do not agree that it has been confirmed.

In any case I wish you a wonderful season.


*I'am the author of original post. And whole this situation makes me frustrated, some people started to rewrite my words with some misstakes, give not completely accurate answers for some reasonable questions and so on. Also general readers made some conclusions that I did not provide and so on.


>> Third, they performed no statistical tests.
We are not crazy, first of all we performed stat tests. We performed chow test for structural break at 17th nov (p-value about 1e-16) and chi-sq test (p-value ~1e-160).


>> First, the authors did not sample the population randomly, they surveyed the population and requested self-reported data.


Only plarium can sample population randomly and operate with fround truth data. We can just try to deal with self reported.


For example we made subsample with people who started filling drop before potential change and got same results for them as for whole dataset. We tried to find any changes in raw data (a lot of new users, average fill rate changes and so on), that can be sign of data corruption (covid in your example), but we did not find anything. (btw, I don't think that covid is reasonable example, but I understand what you mean).


Generally I have following point: you can select group of users (big enough) and collect their drop every day. It does not matter how did you sample them from general population if you assume that all people are equal (in our case - have same drop chances). If you see some prooved changes in theirs drop, you can claim that there is some changes in general population drop. I mean that if all people are equal all subsamples will have have same distribution as general population.


The only way when it is not true - people was not equal from the beginning. It can be because of many different reasons, for example - unannounced compensation system, when drop prob depends on your previous drop. But we have no reasons to think so, this theory was not prooved by any previous research. Personally I tried to find something like month guarantee for sacred shards (or sacred shard + leg book) (according to my feelings, you can't get less then 4-5 sacred shards from 2x top UNM chest per month) but failed, everything looked like fair random.


* I have to notice, that it is steel possible that we deal with some year-compensation (if you got to many sacred shards this year, you will get more epic books in last two month of the year???) but we don't have any data to speak about it. It is possible, but not very likely, I suppose.




Any way only plarium can say final word. And they assert that everything is okay.


I think this situation has two potential finals (if it is bug or a/b test):


1) plarium will rejet/fix it, we will see it in our data and report that now drop is as usual


2) plarium will accept changes, apply it for all users and we will see it in datamined data. (oh, I have so many questions to plarium, why they were generating drop on client side :D )


Alternative (I don't want to belive in it) - it is not a/b test or bug, plarium divided users by some rule (like they do with donate offers) and we will not see any changes in our data or datamined one. This situation is a dead end and we will never come to agreement.


So lets just wait...


Good luck :)

Dec 24, 2021, 20:2612/24/21
12/19/19
5876
g_o_r_l


*I'am the author of original post. And whole this situation makes me frustrated, some people started to rewrite my words with some misstakes, give not completely accurate answers for some reasonable questions and so on. Also general readers made some conclusions that I did not provide and so on.


>> Third, they performed no statistical tests.
We are not crazy, first of all we performed stat tests. We performed chow test for structural break at 17th nov (p-value about 1e-16) and chi-sq test (p-value ~1e-160).


>> First, the authors did not sample the population randomly, they surveyed the population and requested self-reported data.


Only plarium can sample population randomly and operate with fround truth data. We can just try to deal with self reported.


For example we made subsample with people who started filling drop before potential change and got same results for them as for whole dataset. We tried to find any changes in raw data (a lot of new users, average fill rate changes and so on), that can be sign of data corruption (covid in your example), but we did not find anything. (btw, I don't think that covid is reasonable example, but I understand what you mean).


Generally I have following point: you can select group of users (big enough) and collect their drop every day. It does not matter how did you sample them from general population if you assume that all people are equal (in our case - have same drop chances). If you see some prooved changes in theirs drop, you can claim that there is some changes in general population drop. I mean that if all people are equal all subsamples will have have same distribution as general population.


The only way when it is not true - people was not equal from the beginning. It can be because of many different reasons, for example - unannounced compensation system, when drop prob depends on your previous drop. But we have no reasons to think so, this theory was not prooved by any previous research. Personally I tried to find something like month guarantee for sacred shards (or sacred shard + leg book) (according to my feelings, you can't get less then 4-5 sacred shards from 2x top UNM chest per month) but failed, everything looked like fair random.


* I have to notice, that it is steel possible that we deal with some year-compensation (if you got to many sacred shards this year, you will get more epic books in last two month of the year???) but we don't have any data to speak about it. It is possible, but not very likely, I suppose.




Any way only plarium can say final word. And they assert that everything is okay.


I think this situation has two potential finals (if it is bug or a/b test):


1) plarium will rejet/fix it, we will see it in our data and report that now drop is as usual


2) plarium will accept changes, apply it for all users and we will see it in datamined data. (oh, I have so many questions to plarium, why they were generating drop on client side :D )


Alternative (I don't want to belive in it) - it is not a/b test or bug, plarium divided users by some rule (like they do with donate offers) and we will not see any changes in our data or datamined one. This situation is a dead end and we will never come to agreement.


So lets just wait...


Good luck :)

I think this is the most reasonable and logical post to date from those suggesting a problem may exist.  Unfortunately, I think (apologies if this sounds condescending) poor application of statistics and even worse methodology contributed to my dismissal of any potential issues.  

I stand by my assertation that the data is at best poor (self-reported, self-selected, incomplete, ....), but against my better judgement I downloaded the data and spent 30 minutes of crude analysis and cannot rule out a stench coming from Denmark.

The data would have to be manually manipulated to produce the change seen on Nov. 17.  I think Plarium A/B testing the impact of changing CB drop rates cannot be ruled out.     I also agree an annual limit is very unlikely.   As for differentiating drop rates for different segements (spending tier for example, geography, ....) similar to offers, I think that would be shady...


Dec 24, 2021, 21:2312/24/21
07/17/20
20
Angwil

Yes you have raw data.  Yes you have an expected.  But your raw data has sampling issues.  To begin with, data must be sampled randomly to for the required tests to have any statistical meaning.  

There was someone a few days ago just saying to send data from a single clan for 2 weeks and they could prove/disprove ...  Surely that is not randomly sampled?  

There are a large number of users reporting data for the full 2 months. Data is linked to an anonymous user profile so you can follow each user over 2 months. and see statistically if users stop reporting when they get a sacred... 

If the data was biased due to non-random sampling then I am sure you could show this in the first month of all users data.  The bias would show if you compared the histogram to the expected distribution from the data-mined model.   

Remember this initial experiment was to show users that their drop data did conform to the data mined model as in their first report.  The bimodal distribution came out as unexpected in the second month with many of the same user's data.  Not just an injection of new users data as I understand it.   So unless you think the users deliberately falsified data then how can some users continue to get normal results when a statistically significant change in the distribution for the next month.

Remember there is a mathematical model with probabilities coming from data mining which is generally accepted by the community (although no information was released enabling users to verify this).  This gives you the whole population statistics.

The vast majority of users report chest drops that agree with the predictions of the data-mined model for the first month.  Most users report the same for the second month.  A smaller but (significant depending on your test) report a changed drop rate distribution on the second month.

You can filter the raw data using your own skills and professional experience.  Only take UNM chests, reject users who give an impossible data value such as 6 saceds (max 2 from 2 chests). If you think users stopped giving data when they got a sacred drop then remove users who conform to this pattern.

Any professional or university student would have access to mathematical analysis software (eg MatLab stats package) which can very quickly perform various hypothesis tests.

1 That the results were all produced by the data mined mathematical mode for the 1st and 2nd months.

2. That some users drop probabilities changed in the second month to form the bimodal distribution that can be represented by the data-mind probabilities and a single second set of probabilities.

If someone could show a mistake in the data or the analysis then it would be great, problem solved.  Surely there is more that can be done than just dismissing it?

Dec 24, 2021, 21:3212/24/21
06/20/19
2184

I believe that your thinking is the best GIVEN the data available.  Your post is very reasonable. What I like best about it is you're looking forward to further data analysis (as I've been saying) in which you provide 3 alternatives and state we will see it in continued data collection.  

I apologize for your frustrations regarding any mistatements regarding your posts.  I generally agree with Trips posts.  Please realize I am unaffected by this and that in itself represents a bias.  But I do encourage you to continue to investigate.

I think beyond the data and anaylyis, what do you want to happen?

I would like to think, if we assume all data/analysis demonstrates the veracity of the results, that a coding error/artifact is the reason.  I'm no programer so have no idea how they couldn't just copy the code from one server to another so how is this an issue?

I hate to think it's an A/B test.  I hate to think this is based on spending/geography, etc.  I hate to think this is year end compensation (in fact if rng is working there should be no reason for this).  But i can't rule out any of these.

Again, the quetion is "where does the current state of knowledge lead you?  What are your next steps?"


The topic is locked. You cannot post comments.