Holy Fuck

Holy Fuck

I just made a machine learning model for predicting future ethereum prices. I ran it through a simulation, and it multiplied my money 79 times.

Is this what printing money feels like?

Attached: Screenshot_20200816_050942.png (687x402, 61.25K)

Other urls found in this thread:

medium.com/@samuel.monnier/cross-validation-tools-for-time-series-ffa1a5a09bf9
twitter.com/NSFWRedditVideo

Attached: 1555708283909.png (657x450, 42.92K)

No it's not.

Share the code then

>start with 10k
>ends with 757
sounds accurate

>what is overfitting existing data?
you have to build the model with half data and run it with the second half at least

I did that

>you have to build the model with half data and run it with the second half at least
it's even more complex if you are analysing time series, you need to split it in train/test/validation and remove the embargo parts. there's a book that explains it really well

or you can downlod the already made code from here medium.com/@samuel.monnier/cross-validation-tools-for-time-series-ffa1a5a09bf9

I actually converted data from 20 day spans into graphs labeled with future movements, then used resnet34 to notice movements in the graph.

you still need to remove the embargo parts or you're crossvalidation is garbage

do you guys wanna build a crypto autistic hedge fund?

ethereum is finished

Your machine hasn't learnt anything if it thinks the price is going to continue going up.

The thing is, there's always some shitcoin going up 20% or some shit. If I can make a model that can predict those big movers, I could amplify returns significantly.

>can't follow naming conventions
>unnecessary parentheses
>using .format() instead of f-strings
>if(imagePrediction > 0.1 and isStaked == False):
> isStaked = True
>if(isStaked == True):
>if(imagePrediction < -0.01):
> isStaked = False
OP you must be 18+ to post here.

Ah yes. Because what's more important than speed, agility, and making useful shit?

Oh right. Naming conventions.

I don't work with anyone else, so these are all retarded points.

i was literally just thinking about such a hedge fund today

Look at your if statements. Look at the logic of your code. You started programming less than a month ago.

Ya got me. I'm crying with my 7900% predicted return.

we can stop insulting each other for a moment and start building something
OP spent time trying to learn stuff and model something, so other people did, and regardless if it's leet code or not he has the drive to learn, try, make mistake and build
let's organize into an autism powered team:
drop a mail: [email protected]
we can make a slack for example
I have some experience with backtrader, built some models as well, we can try something, do peer-review, and then collect funds from autists here, make a website, get a hedge fund license and build a company, why not? it takes very little to build one in malta or Seychelles or some other non cucked country

sent

Just sent. craigslist

>you have new mail
>thanks for joining blacked.com

henlo henlo peer-reviewer, do you mind joining?
I have some ideas

Is this larp?

why? some of us are spending the weekends playing with this stuff, so why we don’t work together?
several things should be discussed, what kind of organization form, what rules, what roles, but nothing impossible

ITT fucking lowest plebs keep dreaming
>hurderpoo ima build de A-EI for de moni printer
You don’t even grasp what you have to read to understand what you have to do to make that shit work.
>inb4 but le backtraderrr, le alpha

Attached: D1B2F1C0-304B-4EF9-A100-C606738DA00C.jpg (700x557, 40.62K)

If not it is reddit tier autism, the worst kind of it

Will it be algo based? I am actually working on something like that myself

Meant for