/biz/ - Business & Finance

what do I have to learn to make my own bot?

>>12016606
you have to learn how to research and learn things on your own instead of asking strangers on an anime forum

>>12016625
NANI?!

>>12016606
Books and shit.

I can’t get my packages until tomorrow because of this old fuck

>>12016625
How do I learn how to research and learn things on my own?

https://algotrading101.com/

>>12016750
unfortunately you have to be born without the retard gene. it might be too late.

Also if you just wanna ride someone's algo:
https://www.darwinex.com/explore

>>12016606
>learn python
>learn about statistics and probability
>learn how to make efficient algorithms
>learn how to acquire data in real time

there goes your bot, perhaps you might lose a ton before you make a decent hit on your algorithm

>>12016851
>you might lose a ton before you make a decent hit
this. it's just the tuition if you wanna learn trading.

>>12016535
What logic is your bot doing that's so hard to just trade yourself? I found the most important work was pulling data and sort. Just dump to Excel and pick trades out its 5 minutes to make the actual trades. Or is your bot making millions of superfast trades or some shit? Lol at fees 8.99 a trade like me

why make a bot when you can just run regressions in R and see if the results make any economic sense yourself. Probably make less costly mistakes.

>>12016893
nobody uses a bot unless it's for HFT
else they're retarded with a lot of money to waste

>>12016893
you could be running a portfolio of weighted strategies
you could be trading on very short term order book dynamics
you could be doing arbitrage
etc

you'll still be looking at charts in R or whatever, but some things can't be done manually

>>12016939
not only HFT
say you're listening for a kind of event, where a set of prices deviate from some statistical relationship, and you want to act immediately upon that event
are you going to stay glued at the screen all day?

>>12016750
stop shitting on the streets

>>12016606
https://www.quantopian.com/
They even let you backtest.

>>12017014
well, there goes my holiday free time. thanks for posting this link

>>12016893
>regressions in R
what's R?

>>12016606
Keep in mind that you're also competing against Wall Street superbots who have more data and faster speeds than you

>>12017150
>doesn't know what R is

ITT: retarded NEETs think that they compete against literally the smartest people in the world with their shitty notepad++ bots

>>12017150
reverse

>>12017150
https://www.r-project.org/
>R is a free software environment for statistical computing and graphics.

>>12017377
>t, brainlet
There are a lot of strategies that work great for a NEET happy to make 6-7 figures a year that won't work for multi-billion dollar whale hedge funds because of their size

>>12017377
>notepad++
that's my favorite language as well anon

>>12016606
Not even a GED lmao you can become a programming equivalent to a 4 year bachelor degree from googling lmao

>>12017451
thank you. Scary to see the level of smarts I am competing, but I still think it's good side hobby, but the picture is - we are all in the same retail investor boat against funds.

>>12017451
thank you. Scary to see the level of smarts I am competing against, but I still think it's good side hobby. Still the bgig picture is - we are all in the same retail investor boat against funds.

>>12017486
provided you have a 120+ IQ. I've tried to help people learn programming and some people will just never be able to go beyond the basics, their brains just aren't capable of the abstraction. Most get stuck on nested loops and recursion

Like 2/3rds of American adults aren't capable of basic algebra, programming is far beyond them

Real talk: if you want to learn solid programs read an intro to programming + algorithms book, I humbly suggest this: https://introcs.cs.princeton.edu/java/home/
It assumes you know HS math.
After that you could either try your luck with learning another langue + API (sentdex has a nice series on python for this)
Or you could go a bit more theoretical, get a decent background of: calculus, linear algebra and discrete mathematics. And you should be good enough to tackle heavyweight books like CLRS.
After that you be alright to apply your demigod algorithmically knowledge to finance, biostats, or wherever you see fit.

>>12016535
Whatever youll make it back nerd.

>>12016535
Read Karl Poppers works and try to falsify your algorithm. This way, you will get better.

>>12017486
and a dolla fifty in late chaages at the public library

>>12017834
interesting anon, I will look

>>12016535

kek

>>12018419
Mind to share some of your insights?

I also developed algorithmic trading, and fortunately I had read Karl Popper at that time. It helped me a lot, to find weaknesses in my approaches - and I'm thankful of this. Hope it also helps you!

>>12019803
So you were successful with a profitable bot? Mine is based on moving average crossovers which is very profitable except when the market is very sideways and choppy. Trying to figure out how to make it not trade in these cases. Currently looking at using volatility indicators such as ATR and StdDev, and probably some price action logic. But really I just want to get a neural network hooked up so it can figure this out for me.

>>12017519
Silicon valley coders 75% are foreign :D asian + european

>>12017328
>literally 95iq humanities freshman students know what R is

>>12020279
In my opinion, I was successfull, but it's actually hard to tell. I developed several systems, but the most important thing was really to systematically test. It is really important to not just test historic data. My works on neural networks showed me, how to test things properly. You need to split your dataset to "learning" and "testing" data. Learning data is to optimize your parameters, your testing data is blind-testing this optimizations on data, which are completely new for your algo+params. Don't do the mistake to only rely on historic data. They can show earnings like > 100% per month, but in real, these are just overoptimized results and don't work. It's hard to falsify your own hard work, but at least it helps not to loose money.

After months of works, I figured out I'm far better at trading manually than by my bots. I'm less anxious, figuring out, why my neural-network bot has decided to buy or sell. Thats why I'm currently not using any bots for trading.

I'm no scientist or some mathematician. I just develop software for fun. Hope it helps!

>>12016535
take the pill
>your trading bot sucks if it lost to a dead guy

>>12016939
these are the words of someone with a 90 IQ trying to sound smart. it's clear you have no idea what you're talking about

>>12016535
>create shitty bot
>it doesn't work
>blame innocent old man who just recently passed away

Sad!

>>12021526
Good info thanks fren. I'm making a bot since I can't/don't want to be day trading all day. I test my bot on multiple samples from the past two years and for the past 6 months it has shown ~5000% gains with a max drawdown of 20% or so. My only fear is it is too optimized for this time period, but I'm building it for the current volatile market.

>>12016535
It's a shit bot

>>12017150
R is a computer language like Javascript, C, C++, Python etc. Programmers use computer languages to make a computer do what they want. Like "be an iphone" or "be a videogame".

>>12023092
>innocent old man

>>12016606
listen kid, you can keep the prices of GPU's up all you want, but crypto is an industry completely dominated by super computers (spoiler, you don't own one and never will because you can't buy them and never will be able to afford to craft one/pay people to do it)

even at that, BTC was never intended to be anything more than a ponze scheme backed by money laundering and greed- and now crypto is just a giant shitstorm of retardation that you are 1 trillion times better off avoiding in order to learn about REAL markets.