Is statistics complete bullshit...

Is statistics complete bullshit? Any time I see it in action it just doesn't seem like anything more than marginally more accurate guess work when conclusions are drawn from the results.

you can draw conclusions on some basic events and certain outcomes will happen a large percentage of the time. Is it always right? Nope.

The goal is to predict the median outcome so that you can balance risk and reward.

It just seems like these "probable events" aren't as probable as these models predict most of the time. Especially with forecasting...

They account for a certain degree of error. No, it isn't perfect, but it beats having idiots like you sticking your head out the window - guessing what the weather next week will be

Not OP but how can anything more complicated than regression be applied irl? Where does this stuff even come up?

>Any time I see it in action
>when conclusions are drawn from the results.

This is where things generally go south. Usually when you see statistics "in action" it's some rube drawing completely unsupported inferences from statistical data because it serves whatever point they're trying to make. There is absolutely nothing wrong with statistics if you don't try to force the data to say something which it isn't saying, which is generally the case when you hear someone talking about "statistics show X, Y and Z" in a newspaper or some shit. Statistics, when done correctly, is never bullshit but the conclusions which people choose to draw sometimes are.

That seems like what happens more times than not from what I see. Even if the data is accurate it's still prone to human error or unforseen circumstances that apparently weren't built into the model

Statistically speaking, the more broad the conclusion, the less that conclusion is likely true.

It's pretty much all bullshit. The whole language they use to describe stuff is all wiggle words. Ask for the answer to anything and it's all hand waiving and mystery. Complete bullshit.

Statistics is the core of every quant fund and HFT firm. These types of fund have been far more profitable than traditional hedge funds for a while now. Statistics isn't just grabbing data and finding a trend line.

>Where does this stuff even come up?
economic models and analysis of policies that have loads of variables

That appears to be how most of it is applied. Look at the forecasting from these huge firms. From what I read the accuracy is around 20% or worse. You can't tell me that certain parts of statistics isn't complete bullshit.

>marginally more accurate guess work
you nailed it. it's absolute shit, but also the best we've got

>you nailed it. it's absolute shit,

You're doing a lot of people a disservice by saying it's absolute shit. Stats has saved a lot of lives. Shit. They use stats to predict tornadoes mang.

> just because I don't understand it it's bullshit
I got paid 124k to do a 5 month study to determine if people would stay in a hotel that was located at a certain location, and was able to determine how many people would stay there within a 4.8% margin, used ANOVA and ended up being 1.4% off.

damn, thats impressive, what data did you have?

This. Statistics well done is a great tool, but most of the times we see statistics are just bullshit manipulated to show what I want to show, rather than real data, as seen everytime on politics.
On science this also happens, as many researchers will force data matematically or theoretically to fit their models. But as a tool is it useful, widely used on many fields.

Yeah, data doesn't lie. People do.

Interesting, could you elaborate more? Was it a new location or existing? What existing data did you have available?

the actual science behind it isn't bullshit, but stats is known as "how to lie with numbers" for a reason

You're just retarded for not recognizing that a "marginally more accurate guess" is extremely profitable in the long run.

every model is wrong, some are useful

How to get into a statistics business like consulting or freelancing similar to do companies usually hire outside to do statistical stuff or do they usually only use in house employees?

Statistical forecasts are usually just variables that some overpaid grad student tweaked until they matched past data. Meanwhile, the data has been manipulated so much it's meaningless. The data falls apart when future events occur and the modeler makes it even more nonsensical as the variables are tweaked again to fit the new data. It's an unfortunate cycle and through encounters I've had with statisticians and data analysts I swear only 5%, probably less, know what they're doing.

Actually it doesn't seeing as random stock picks beat the majority of hedge funds that use... wait for it... statistical analyses. Data forecasts are just cover for executives to do what they want. Seriously, any decision can be justified by a model that's been manipulated enough. Are there those who use it correctly? Yes, of course. Do most use it correctly? Not even close.

Idiot business majors ITT learn to math

Statistics is a branch of mathematics, it's not "complete bullshit"

Literal retards

You fucking hypocrite.
>5%, probably
>statistics is bullshit
You use a statistical measure based on NOTHING. No data, no hypothesis even. Fuck off nigger