"
By the way, to all those people that wanted the math posted THERE ISN'T ENOUGH DATA IN THE VIDEO TO DO REAL STATS ON IT, BUT THE SAMPLE SIZE IS PLENTY LARGE ENOUGH, ALSO I DID SOME MATH:
You only need 30 samples if you want a basic picture of what drop rates are like. There is talk that this 'minimum' should be raised for serious scientific work, but you can get a fairly small confidence interval with just 30 datapoints.
There are two ways to run this experiment: by doing it with no other variables besides Biscos, where you clear 30+ white, unmodified maps, without any extra forms of item find or whatever,
or by acting to the best of your ability to raise item drop rates in general, like the streamer in the video did. Use all the sextants, alch every map, wear a bunch of item rarity, etc.
I'm gonna show you the basic, highschool level of stats that could be applied here.
We call this Confidence Interval for Two Independent Samples.
First, what numbers do we have?
He did 60 maps each time, and got 108 total from the control set and 160 from the biscos set.
The population, or n, of our data sets are 60 and 60, for the maps run with and without biscos. Call them n1 and n2.
The Means of out dataset are 108/60 and 160/60, or 1.8 and 2.666..., call THEM u1 and u2.
Problem though, we also need a standard deviation for both of these sets. Standard deviation is a sort of measurement of randomness. You would normally get this by comparing each datapoint to the mean of the dataset, sort of measuring how far each point is from the average.
Since we don't have that, I'm going to make up two datasets, one being super random (where every map got dropped in the same run) and the other super average (where rngesus is a merciful god and you got 1-3 each run). Both of these are extremes, and don't represent real drop rates.
Maths for Standard deviations spoilered
Now that we have our standard deviations, we need to pool them together, since we're comparing two sets of data. We call this the pooled estimate of the common standard deviation, and it means we assume that the two datasets have similar variance.
More maths
Alright, NOW we can do a confidence interval.
This amounts to taking the difference in the means, and comparing it to the pooled standard deviation modified by a T value and the populations.
The T value we just get from a table since this is high school maths. In this case we have 60 data points and want a 5% confidence, so we come up with 2.000, yay for easy math.
Maths the mathining
We don't actually have a standard deviation though, so none of this is actually more than an example.
The 'line' where we move from not confident to confident comes at a standard deviation of around 2.5, which is relatively low. If you have for or five runs with more than 5 maps that run, we probably won't be able to prove anything either way.
Or maybe, just maybe he should have count all maps, not just strand. Or maybe just maybe he should have run all the maps white with no sextants and no rolls on maps. Also if he get same ammount of maps he would probably not post it at all, because publication bs, when you don't want to post something underwhelming. Maybe he got less maps with Bisco's like 3 times and posted only when he got sattisfied results? Who knows.I don't think there is need for such walla of text.
Last edited by SunL4D2 on Jan 12, 2018, 1:48:32 AM
|
Posted bySunL4D2on Jan 12, 2018, 1:46:40 AM
|
"
everyone posting his results...
you DO realize he only counted SHAPED STRANDS, right? he NEVER mentioned how many actual MAPS he dropped.
the entire argument based on his results is flawed from the beginning...
|
Posted bySarevokczon Jan 12, 2018, 1:48:42 AM
|
"
"
everyone posting his results...
you DO realize he only counted SHAPED STRANDS, right? he NEVER mentioned how many actual MAPS he dropped.
the entire argument based on his results is flawed from the beginning...
Dude, you are the only person who noticed it too BibleThump
|
Posted bySunL4D2on Jan 12, 2018, 2:55:50 AM
|
"
"
everyone posting his results...
you DO realize he only counted SHAPED STRANDS, right? he NEVER mentioned how many actual MAPS he dropped.
the entire argument based on his results is flawed from the beginning...
You do realize that he forced every T9 and T10 drop to be a shaped Strand, right? Also, unless for some magical reason Bisco's affects either the tier or the layout of the maps dropped, it doesn't matter which map he counts. He could literally be counting any T3 map, and only that T3 map, and as long as he does so long enough to get a good statistical sample, he can test the hypothesis "Bisco's doesn't affect map drops" on any confidence level he wants to test it on.
Remove Horticrafting station storage limit. Last edited by Char1983 on Jan 12, 2018, 6:01:15 AM
|
Posted byChar1983on Jan 12, 2018, 6:00:56 AM
|
"
"
"
everyone posting his results...
you DO realize he only counted SHAPED STRANDS, right? he NEVER mentioned how many actual MAPS he dropped.
the entire argument based on his results is flawed from the beginning...
You do realize that he forced every T9 and T10 drop to be a shaped Strand, right? Also, unless for some magical reason Bisco's affects either the tier or the layout of the maps dropped, it doesn't matter which map he counts. He could literally be counting any T3 map, and only that T3 map, and as long as he does so long enough to get a good statistical sample, he can test the hypothesis "Bisco's doesn't affect map drops" on any confidence level he wants to test it on.
Ahahah, so what if he uncompletecd t9s and t10s? There are lower maps, too. He could have gotten 160 strands and 100 lower tier maps, while in the other test he could have gotten 100 strands and 160 lower tier maps, we simply DONT KNOW. randomness of the maps is so volatile 60 runs isnt nearly enough to make a statement, especially when he didnt even include ALL the drops, or, like he specifically said, made arbitrary decisions on specific things (like not running backwards abysses)
IF he really wanted to do a proper test, he would have...
And its funny, how his results are brough as some kind of proof, when the other guy, who did 40/40 atolls and had 20% more map drops without bisco just to show, that map drops are very random was discredited as unreliable and low sample size etc.
|
Posted bySarevokczon Jan 12, 2018, 6:13:20 AM
|
"
Ahahah, so what if he uncompletecd t9s and t10s? There are lower maps, too. He could have gotten 160 strands and 100 lower tier maps, while in the other test he could have gotten 100 strands and 160 lower tier maps, we simply DONT KNOW. randomness of the maps is so volatile 60 runs isnt nearly enough to make a statement, especially when he didnt even include ALL the drops, or, like he specifically said, made arbitrary decisions on specific things (like not running backwards abysses)
IF he really wanted to do a proper test, he would have...
And its funny, how his results are brough as some kind of proof, when the other guy, who did 40/40 atolls and had 20% more map drops without bisco just to show, that map drops are very random was discredited as unreliable and low sample size etc.
Someone run AToll with Bisco's to test as well? hmmm.
|
Posted bySunL4D2on Jan 12, 2018, 10:46:14 AM
|
"
"
everyone posting his results...
you DO realize he only counted SHAPED STRANDS, right? he NEVER mentioned how many actual MAPS he dropped.
the entire argument based on his results is flawed from the beginning...
Technically, his 'experiment' only suggests that Strand maps might be more likely to drop while using Biscos.
But unless you have a good reason to think a specific variety of maps is more or less volatile than other varieties of maps, there isn't a big difference between using only Strand maps and using all kinds of dropped maps. The data would be more robust, but not more accurate. We have no reason to think that including all maps dropped would do anything but raise the average number of maps found per run.
Again though, without knowing the specifics of the data we can't make any conclusions. His example raises flags that something could be going on, and somebody who cares enough should do a more well-documented set of runs.
"
And its funny, how his results are brough as some kind of proof, when the other guy, who did 40/40 atolls and had 20% more map drops without bisco just to show, that map drops are very random was discredited as unreliable and low sample size etc.
Mind linking me to that guy? 20% more map drops and really random sounds like you wouldn't be able to make a yes/no with any confidence, and thus legitimately need more data to make a valid decision.
But like, it'd be nice to have some actual data.
Last edited by Duodecimus on Jan 12, 2018, 3:14:55 PM
|
Posted byDuodecimuson Jan 12, 2018, 2:32:19 PM
|
"
Bisco's collar is like 60c in standard (around the correct price compared to all other MF Items) and 18ex+ in ladder.
Just wanted to quote this again to show how retarded the sheep are in this game. Is bisco's good? Yes. Is it 18ex good? for 99% of the people that own one... no fucking way.
If I dropped one today I would sell it instantly. Streamers have hyped this shit to the end of days...
The benefit is that as a result all other really good amulets are dirt cheap right now... so I hope they keep up the hysteria!
|
Posted byPrizyon Jan 12, 2018, 3:51:51 PM
|
|
Some of the ventors are 60-100ex. Like 140+ all res with top life and MF. People just need something to spend on. No way will you make back what a setup like that costs compared to no MF
Git R Dun! Last edited by Aim_Deep on Jan 12, 2018, 4:25:54 PM
|
Posted byAim_Deepon Jan 12, 2018, 4:24:09 PM
|
|
I love bisco. I can't play this game without bisco.
|
Posted bykpham1991on Jan 12, 2018, 10:15:04 PM
|
