Are you freaking kidding me?

"YaeVizsla;c-1907275" wrote:

"ZAP;c-1907195" wrote:
Hundreds is not even close to figuring out averages. Hundreds of thousands is getting warmer, but will still create some outliers

Not... exactly. Or rather... no. Not how that works.

Poking around at binomial calculators online, since it's been years since I've taken statistics, my 92 drops in 390 tries gives me a 99% confidence that the drop rate for finishers is somewhere in approximately the 15-30% range. As I make more attempts, that range will narrow.

Hmmm, you said the exact same thing I said with a bunch of “stuff” in between.

What I said was absolutely true and you even confirmed it in the above bolded sentence.

It’s no different than percentages in poker hands. Those percentages are derived from millions of trials, not a handful. The more trials, the more you narrow the range, we’re saying the same thing, aren’t we?

"ZAP;c-1907836" wrote:
Hmmm, you said the exact same thing I said with a bunch of “stuff” in between.

What I said was absolutely true and you even confirmed it in the above bolded sentence.

It’s no different than percentages in poker hands. Those percentages are derived from millions of trials, not a handful. The more trials, the more you narrow the range, we’re saying the same thing, aren’t we?

No, we did not say the same thing. For reasons you are doubling down on by going up into the millions.

Your original post:

"Hundreds is not even close to figuring out averages. Hundreds of thousands is getting warmer, but will still create some outliers"

This is an extremely disingenuous take.

You set an impossible goal. You set a nigh unattainable standard to achieve the impossible goal. Then you concern troll a vanishingly implausible potential failing even if you achieve the nigh unattainable standard.

That's not an accurate assessment of statistics. That's rhetoric to dismiss statistics as an accessible tool.

Fact of the matter is smaller data sets are valid ways to draw conclusions with appropriate caveats. Being able to say with 99% confidence that the drop rate on finishers is between 15 and 30 percent is meaningful, actionable data. It does not require millions of trials. And while a narrower range would be lovely and will come in time, you only ever need so much precision and an arbitrary degree of precision for all practical purposes is attainable LONG before the hundreds of thousand of trials you initially stated.

Ftp players who farm meta characters like cls rey revan dr obtain them around the same time the meta is changing. The one exception may be hans falcon (which couldve been had in apr for ftp players if you immediately stopped what you were doing to farm bh ships when they were made available back in nov/dec, if I remember correctly).

I assume, as a ftp player who finally 7 starred my falcon, the meta will be changing shortly. Stay tuned? Or am I way behind already

I'm FtP and I got 5* Falcon on its original pass. It was incredibly useful even then. 7*d it on the 2nd pass. It was still meta and has remained meta.

Likewise, I got Darth Revan on the 2nd pass, though it took until 3rd pass for me to get JKR. I'm all ready to go for Malak 2nd pass and Padme 2nd pass (I actually could have 5*'d her the first time, I was told, but I didn't feel like bashing my head against the wall when I still had to devote gear to my DR-acquisition squad so I wouldn't be able to get her to useful gear levels anyway). My Brood Alpha is 40+/65 right now.

I'm not saying that I have as focussed a roster as the PtW folks. Nor do I get anything on the first pass (unless its available at 5*, like Falcon). But when I do get things, they're still plenty useful and the meta hasn't completely changed by the time I get them.

"YaeVizsla;c-1907275" wrote:

"ZAP;c-1907195" wrote:
Hundreds is not even close to figuring out averages. Hundreds of thousands is getting warmer, but will still create some outliers

Not... exactly. Or rather... no. Not how that works.

You cannot prove randomness. It's random. There will always be variance. There will always be outliers. You do not prove odds. You develop confidence intervals.

Given the context- rolling dice for events in the tens of percent- hundreds is a statistically significant set. Poking around at binomial calculators online, since it's been years since I've taken statistics, my 92 drops in 390 tries gives me a 99% confidence that the drop rate for finishers is somewhere in approximately the 15-30% range. As I make more attempts, that range will narrow.

Now, 16-30% is a pretty broad range. But it is still significant data, and lets us draw some meaningful conclusions. It tells us that the 10% estimate is very likely pessimism talking, not a reasonable estimate of the drop rate. It tells us the 20% estimate others have used seems fairly reasonable and perhaps even conservative.

At a thousand trials, that 99% confidence interval would be down around a 5% band. At twenty five hundred, it would a 2% band. And while you don't know what the odds are at that point, being able to say with 99% confidence that you're at this rate +/-1% is close enough for government work.

If you’re aggregating data, mine were hovering at around a 20% drop rate after 3 days (roughly 250 attempts for 50 pieces).

I also said the thread title when I went 8/10 today.

"Dirty_Litle_Smuggler;c-1908142" wrote:
If you’re aggregating data, mine were hovering at around a 20% drop rate after 3 days (roughly 250 attempts for 50 pieces).

Nope. Just tracking my own because that's what I can specifically verify and I don't want to go through the hassle of sorting out everyone reporting in different formats or creating a standardized reporting format that everyone will ignore anyways. And if I were aggregating, I wouldn't accept approximatelies.

"YaeVizsla;c-1907965" wrote:

"ZAP;c-1907836" wrote:
Hmmm, you said the exact same thing I said with a bunch of “stuff” in between.

What I said was absolutely true and you even confirmed it in the above bolded sentence.

It’s no different than percentages in poker hands. Those percentages are derived from millions of trials, not a handful. The more trials, the more you narrow the range, we’re saying the same thing, aren’t we?

No, we did not say the same thing. For reasons you are doubling down on by going up into the millions.

Your original post:

"Hundreds is not even close to figuring out averages. Hundreds of thousands is getting warmer, but will still create some outliers"

This is an extremely disingenuous take.

You set an impossible goal. You set a nigh unattainable standard to achieve the impossible goal. Then you concern troll a vanishingly implausible potential failing even if you achieve the nigh unattainable standard.

That's not an accurate assessment of statistics. That's rhetoric to dismiss statistics as an accessible tool.

Fact of the matter is smaller data sets are valid ways to draw conclusions with appropriate caveats. Being able to say with 99% confidence that the drop rate on finishers is between 15 and 30 percent is meaningful, actionable data. It does not require millions of trials. And while a narrower range would be lovely and will come in time, you only ever need so much precision and an arbitrary degree of precision for all practical purposes is attainable LONG before the hundreds of thousand of trials you initially stated.

Ah I see now we are not talking about the same thing and we are both correct.

You are correct that for practical purposes, namely our own personal experiences and expectations, we can assign a range of expected percent based results, based on smaller datasets.

I was also correct in that assigning more specific percentage outcomes happens through millions of trials.

In poker and I assume drop rates in SWGOH, when something is expected to happen X% of the time, that “X” was derived from a high number of trials and not a low number of trials.

It’s all the, “I did 100 runs and only got 15 drops, so the drop rate must be 15%” is what I was responding to.

And you can assign whatever perceived drop rate range you want, but that doesn’t change what the true or nearing true drop rate truely is.

In regards to omegas, my personal tracking put it at less than 1%.

When I was working on Veteran Han and Veteran Chewie, I tracked my omegas gained. I stopped tracking around six hundred sims, because my omega count was two (it was getting depressing). I know that represents a small sample size, but it was enough for me personally to feel confident that it is no higher than 1% and most likely less.

In those same sims, I had gained around 200 shards for the characters. The average seemed in line with the expected drop rate I’ve seen posted online.

I went looking before I made this post, but it appears I didn’t keep the data to post here.

"ZAP;c-1908426" wrote:

"YaeVizsla;c-1907965" wrote:

"ZAP;c-1907836" wrote:
Hmmm, you said the exact same thing I said with a bunch of “stuff” in between.

What I said was absolutely true and you even confirmed it in the above bolded sentence.

It’s no different than percentages in poker hands. Those percentages are derived from millions of trials, not a handful. The more trials, the more you narrow the range, we’re saying the same thing, aren’t we?

No, we did not say the same thing. For reasons you are doubling down on by going up into the millions.

Your original post:

"Hundreds is not even close to figuring out averages. Hundreds of thousands is getting warmer, but will still create some outliers"

This is an extremely disingenuous take.

You set an impossible goal. You set a nigh unattainable standard to achieve the impossible goal. Then you concern troll a vanishingly implausible potential failing even if you achieve the nigh unattainable standard.

That's not an accurate assessment of statistics. That's rhetoric to dismiss statistics as an accessible tool.

Fact of the matter is smaller data sets are valid ways to draw conclusions with appropriate caveats. Being able to say with 99% confidence that the drop rate on finishers is between 15 and 30 percent is meaningful, actionable data. It does not require millions of trials. And while a narrower range would be lovely and will come in time, you only ever need so much precision and an arbitrary degree of precision for all practical purposes is attainable LONG before the hundreds of thousand of trials you initially stated.

Ah I see now we are not talking about the same thing and we are both correct.

You are correct that for practical purposes, namely our own personal experiences and expectations, we can assign a range of expected percent based results, based on smaller datasets.

I don't think you're fully understanding what Yae is saying or what confidence intervals are.

I'd you do fully understand then you are not explaining yourself very well.