Feedback New relic drop table?

[Manganite]

It is an evidence that your data are of no significance to support your assumption.

Of course, it is not an evidence that there is nothing wrong with the ToR. I don't know if it works properly or not, how should I? I have only your data. I just checked, if your data really support your hypothesis. And they do not. Or in other words, they are not significant enough to justify to withdraw the assumption the ToR works like expected.


And I cannot comment on what ever once was wrong with he Arc. I have no experience with the Arc. But it seems, there was once an issue with the Arc and it is now solved. And again, why should they program the ToR like the not properly working version of the Arc and not like the one with no more issues? That doesn't make any sense. I would not assume they start programming the mechanics of each GB from scratch. I would assume they are all based on the same code.

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[DeletedUser7942]

Just my opinion but I think you're possibly making a mountain out of a molehill. Could there be a bug? Sure. Should Inno do some testing to be sure? Yeah sure why not. But should they based on what you're saying? I don't think so. You took a lot of time to gather some data which is always appreciated, at least you're not spouting conspiracy without anything to back it up like some have done regarding ToR. The reason why I say this is much like manganite is saying. It seems to me it's merely a matter of understanding probability theory regarding your take on this which is the issue at hand. I notice you use the word "expect" in your data post. An expected outcome, even if known as is the case with Relic drop %'s, will naturally have variability as your data shows and I'm sure you understand. Let us look again at how wide that variability can be according to your own data;

Level 70 ToR:
Total 20.31% - 31.25% avg 27.19% (Stated chance 31.5%)
Silver 35% - 64% avg 52.35% (Stated chance ?)
Gold 15.38% - 47% avg 29.93% (Stated chance 29.3%)
Jade 5.56% - 29.41% avg 17.72% (Stated chance 16.59%)

The stated chance for a relic to drop is 31.5% yet one week it was only 20.31% for you. By my reading your claim is not that this is a bug. Certainly you don't expect to get a perfect 31.5% chance for a relic every single week. But you instead, looking at the average of all weeks combined, in this case you came up with 27.19% over 20 weeks, you do the opposite and now claim this is a bug. I'm not sure why you do this. If it is that you can understand that statistically speaking you will have variability from week to week then certainly you can understand the "Law of large numbers" which states(according to wiki);

In probability theory, the law of large numbers (LLN) is a theorem that describes the result of performing the same experiment a large number of times. According to the law, the average of the results obtained from a large number of trials should be close to the expected value, and will tend to become closer as more trials are performed.
https://en.wikipedia.org/wiki/Law_of_large_numbers

Now I'm no statistician but even I can understand that even over a 20 week period I should not expect a perfect 31.5% outcome. However I know statistically speaking with the more trials I should expect to get close to that % over the long term. As mentioned 20 trial samples is simply an insignificant sample statistically speaking to draw any conclusion. A million samples? Yeah, that should get us pretty close.

Speaking of which look how close these are:

Gold 15.38% - 47% avg 29.93% (Stated chance 29.3%)
Jade 5.56% - 29.41% avg 17.72% (Stated chance 16.59%)

So how could the overall be bugged yet apparently those are not?

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[Manganite]

For me it is just a statistical problem. He has 20 samples for the number of relics from the ToR. From that samples he gets an estimate for the expectation value of 17.4 relics per week with a standard deviation of 2.98. The proposed drop rate of 0.315 gives an expectation value of 20.16. This is lower than 17.4+2.98 = 20.38, therefor pretty much compatible with his data. Because the only valid conclusion one can draw from his data is, that real value for the expectation value is somewhere in the interval 17.4 +/- 2.98.

 

[DeletedUser7942]

For me it is just a statistical problem. He has 20 samples for the number of relics from the ToR. From that samples he gets an estimate for the expectation value of 17.4 relics per week with a standard deviation of 2.98. The proposed drop rate of 0.315 gives an expectation value of 20.16. This is lower than 17.4+2.98 = 20.38, therefor pretty much compatible with his data. Because the only valid conclusion one can draw from his data is, that real value for the expectation value is somewhere in the interval 17.4 +/- 2.98.

I agree but with a bit more added in that it's not merely the small sample size that is the issue but also comparing the average from that to the stated percentage. For some reason he expects 31.5% on the average with a small sample size but apparently can accept variability week to week. I think the main thing he is missing is that over a large enough sample size will he then see the average closer to the 31.5%.

 

[Manganite]

I agree but with a bit more added in that it's not merely the small sample size that is the issue but also comparing the average from that to the stated percentage. For some reason he expects 31.5% on the average with a small sample size but apparently can accept variability week to week. I think the main thing he is missing is that over a large enough sample size will he then see the average closer to the 31.5%.

Yes, that might be true.

Of course, his combination of results his highly unlikely, but unlikely does not mean impossible. People tend to think, that events with a chance below 1% will never occur. But that is definitely not true. Every event with a chance larger than 0% will occur. It is just a question of the sample size. Here we ten thousands, maybe hundred thousands of active players. Therefor hundreds of players will be affected by such very unlikely events.

It is like a lottery. Of course, a lottery doesn't really make sense for an individual player. The chances are insanely small, but nevertheless, almost every week someone gets the jackpot. Just because millions are participating...

 

[Manganite]

Why do you think you lost anything? You admit now your sample size is too small to indicate anything yet you still think you lost something. I don't agree with manganite that talking with others about their results will do anything for you as any other sample will also be statistically insignificant. I mean I can see if it was like 20% lower every single week, still not out of the realm of probability, but that would give me pause for concern. You're so much closer than that with only 20 samples. How close do you have to be before you think you're not losing anything? + - .01%? Because you'll likely never see that in a given week and would take thousands if not millions of samples to get close to that for the average.

But one has to admit that his case is really on the edge. The a priori probability for his output is really low. Therefore, if you one would look at the output of lets say 5 other random guys, I would expect a higher output. But if all of them would scatter around his average output, it would be a strong point for his assumption.

It is not that you really need millions of trials to say something significant. E.g. the standard deviation of his data is about 3. To drop this to 2 he need roughly another 30 weeks with the same average output. The standard deviation goes down with the square root of the number of trials.

 

[DeletedUser7942]

It is not that you really need millions of trials to say something significant.

I'm just trying to drive the point home regarding "the law of large numbers". I'm just a layman but I get it, the more samples the closer to the expected value. It's not like you can't get near that expected value with few trials but the more the better. I mean in his case he's worried about a few % difference in the average. Look at that Wiki example, they had to roll the die over 600 times to start to get the expected value consistently over the average. It's like he's at the Orange Arrow and I'm saying get to the Blue Arrow:

 

[Manganite]

I'm just trying to drive the point home regarding "the law of large numbers". I'm just a layman but I get it, the more samples the closer to the expected value. It's not like you can't get near that expected value with few trials but the more the better. I mean in his case he's worried about a few % difference in the average. Look at that Wiki example, they had to roll the die over 600 times to start to get the expected value consistently over the average. It's like he's at the Orange Arrow and I'm saying get to the Blue Arrow:

View attachment 2657

Yes, in principle you're right. But it doesn't matter if one guy is counting for 5 years or 5 guys counting for 1 year.

 

[Manganite]

The ToR starts at a drop rate of 12%. And lets assume a maximum drop rate of 32% at level 71. That means on average an increase of 20%/70 = 0.29% per level. At the same time, the number of expected relics per week increases from 0.12*64 = 7.7 to 0.32*64 = 20.5. That's an increase of 12.8 or 0.18 per level. That means to change your expected drop rate by 1 you have to increase your ToR on average by 5.5 levels.

But at the same time your standard deviation increases from 2.6 to 3.7. That means on level one you can expect something between 5.1 to 10.3 relics per week, and on level 71 something between 16.8 and 24.2 relics. The question is now, by how many levels do you have to increase your ToR to expect an significant increase in the number of relics? To calculate that, you have to divide twice the standard deviation by the expected increase per level. For the lower levels that is 2*2.6/0.18 = 28.9, and for the higher levels that is 2*3.7/0.18 = 41!

What does that mean? You have to be at level 30 with your ToR so that it is really unlikely that someone with a level 1 ToR get's sometimes more relics than you. On the other hand, anyone with a ToR higher than level 30 has an increasing chance to get sometimes more relics than a someone with a level 71 ToR.

You would have to double the drop rate at level 71 to 64% to have more or less always an significant higher number of relics per week compared to someone 16 levels below you. Maybe even you aren't that greedy to realize that an average of 41 relics per week on level 71 would bring that game completely out of balance...

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[Manganite]

Everyone with common sense can see that the ToR is a very expensive GB. And everyone knows, that a GBs bonus increase is going down the higher the level of the GB is (Cape Canaveral is a great exception with it's constant 1 FP increase per level).

It is not anyone's fault if you can't calculate what you can realistically get back before you invest your FP... I mean, this game is all about numbers, and in principle you can calculate all the gains and costs in advance to optimize your investments. And I'm not 100% sure, but I would not be surprised if one would take everything into account, the best ToR is a Level 1 ToR. But I'm pretty sure that a very high level ToR is not the optimum, because it is not very likely that you will get back your invested FP back in any reasonable time.

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[DeletedUser5429]

Everyone with common sense can see that the ToR is a very expensive GB. And everyone knows, that a GBs bonus increase is going down the higher the level of the GB is (Cape Canaveral is a great exception with it's constant 1 FP increase per level).

Hi manganite,
Just a very small correction : some GBs see their bonus progression raise. At least for quite some levels.
Examples : Arctic Orangery (critical strike chance) and Seed Vault (chance of getting "something" when aiding another player).

 

[Manganite]

Hi manganite,
Just a very small correction : some GBs see their bonus progression raise. At least for quite some levels.
Examples : Arctic Orangery (critical strike chance) and Seed Vault (chance of getting "something" when aiding another player).

I would still count this as constant increase examples. I mean, there's progression in the 0.01% order of magnitude range.

The ToR is obviously an example for a GB where the bonus progression is lowered above level 10. You have not to go up to level 70 to notice that...

 

[DeletedUser5429]

I would still count this as constant increase examples. I mean, there's progression in the 0.01% order of magnitude range.

I fully agree with the ToR progression.
However this is not the case for the AO : the higher you get, the higher the progression rate.

AO critical strike progression :
102 => 103 : +0.32%
58 => 59 : +0.28%
19 => 20 : +0.16%
(source : http://forgeofempires.wikia.com/wiki/Arctic_Orangery)

 

[Manganite]

I fully agree with the ToR progression.
However this is not the case for the AO : the higher you get, the higher the progression rate.

AO critical strike progression :
102 => 103 : +0.32%
58 => 59 : +0.28%
19 => 20 : +0.16%
(source : http://forgeofempires.wikia.com/wiki/Arctic_Orangery)

It depends a bit, where you're looking:

4 => 5 : 0.23%
39 => 40 : 0.22%

For level 96 it's 29%, for level 1 it's 4.87%. That is an increase of 24.13%, or 0.25% on average per level. Anyway, like with the ToR you have to level up your AO by more and more levels to gain a significant advantage. For the AO you have to go to level 26 to double your chances, and then another 40 levels to double it again.

 

[DeletedUser5429]

It depends a bit, where you're looking:

4 => 5 : 0.23%
39 => 40 : 0.22%

For level 96 it's 29%, for level 1 it's 4.87%. That is an increase of 24.13%, or 0.25% on average per level. Anyway, like with the ToR you have to level up your AO by more and more levels to gain a significant advantage. For the AO you have to go to level 26 to double your chances, and then another 40 levels to double it again.

I was implicitly speaking of levels greater than the tenth. You cannot compare the first ten levels to what comes next (well, of course one can, but they are not reasonnably comparable).

As for doubling the chances, I don't know if that's a sound measure, at least not when taking the 1st ten levels into account.

(maybe we are a bit off topic )

 

[Manganite]

(maybe we are a bit off topic )

Yes, maybe ;-)

But it seems to be really difficult for some people to understand that you can not expect a very significant advantage if your chances are only growing by far less than 1% per level. And the noticeable gain is always higher on the lower levels compared to the higher levels, because the relative gain is decreasing.

 

[NormaJeane]

I have closed this thread temporarily because it seems to be turning in a condescending feud

I'll clean it up and will re-open it once this is done...

 

[NormaJeane]

What was done in the 'cleaning' process:
• some posts were deleted by the poster(s);
• some posts were deleted by me because they only referenced to another deleted post and did not contain additional information;
• some posts were edited, removing references to deleted posts;
• some posts were edited, removing some condescending language.

I did my utmost to maintain all essential observations and do hope that the posters feel I edited their posts correctly.
If not: feel free to contact me by personal message