[Obsoletionist]

I've heard people say "the stock market is a tool to transfer money from the stupd to the smart.", that being the case would it be practical to use a test control group of stock trading simulation, to make AI learn to trade stocks, then let it loose in the wild? 🤝

The test groups, and tracking their metrics should give me practicality of "how well it fares"

[morph]

An AI can in practically all cases can be better at a task than a human. So, if you have the time and the knowledge, go for it.

[6577657534735]

If you have to ask you can't do it. Don't even try, you will get cleaned.

[lockhead]

Check out https://www.quantopian.com/
Here's a YT tutorial: https://www.youtube.com/watch?v=GlV_QO5B...Z&index=13
It's Python, not C though.

Also it doesn't use AI but if you don't have at least a basic understanding of trading, finance and AI, ML Neural Networks I highly doubt you'll be able to pull of anything profitable without dedicating years of your time

[Obsoletionist]

If you have to ask you can't do it.  Don't even try, you will get cleaned.

I'll actually take this advice, and go for it. Thank you for being blunt on what its going to take.

[6577657534735]

I'll actually take this advice, and go for it. Thank you for being blunt on what its going to take.

Keep in mind that professionals at proprietary trading firms design models specifically meant to exploit every naive model they can find listed in every textbook and blog post, to make money off people like you. Also keep in mind that to make money you need low latency access to the exchange you're trading on which costs millions, as well as access to stock data feeds, which costs in the tens of thousands per month for some exchanges. Don't forget that backtested trading strategies which make money often fail to make money when deployed in real life.

You need deep low level systems programming knowledge to build a low latency trading platform, as well as deep statistical and machine learning knowledge to make it behave correctly. Prop trading firms hire people with PhDs in statistics and machine learning for a reason - because trading strategies aren't one-size-fits all, they are novel and need to be constantly updated. It's not about just "making a trading bot" or some other little HF level project. I'm not telling you this to discourage you, I'm telling you this to inform you of the fact that you have literally no chance of ever making money. I say this as a person who has professional experience in this field. This is a massive team effort between systems programmers, statisticians, data scientists, and traders. To top it off you need lots of capital, far more than you likely have

You can't just trade with a couple thousand dollars, you need to drop at least $100k to even attempt to implement a basic trading strategy. That's a lot of money to bet on your systems programming knowledge and adherence to sound control theory. Did you account for flash crashes? Did you account for program states that were unexpected which result in money-loss feedback loops? Did you account for every conceivable bug in your trading algorithm with hardened fuzz-testing suite and manual test case set? Did you account for things like network latency, out of order UDP packets (trading firms rarely use TCP), and factors such as this?

Once again I'm not trying to discourage you, but rather inform you of the fact that there is no hope for one person to do this, let alone a person who is starting from no knowledge of the topic. If you think you're capable of doing this, you're better off trying to get a job at a proprietary trading firm where you can execute your strategies on their capital and have them give you a few million to play with on your algorithms.

It's an honorable goal though, I encourage you to study these topics to the point where you can get hired, but I'm just informing you that you cannot do this by yourself, it's not a solo activity for newcomers, it has a very high barrier of entry and it takes a large village to do this effectively in a way that will actually generate income instead of it just being a toy project. If you think you can write a little AI implementing naive formulas like "buy high sell low" or train a simple neural network on historical data and have it actually make you money, you are very sadly mistaken

[int]

bruh just do 1000 nested if statements, basically the same as an AI in 2019 ROFL

[0xDEAD10CC]

bruh just do 1000 nested if statements, basically the same as an AI in 2019 ROFL

You don't even know what you're talking about.

To learn machine learning algorithms and have your first project be something that has potential to make you go bankrupt though, is not a smart idea.

[int]

You don't even know what you're talking about.

To learn machine learning algorithms and have your first project be something that has potential to make you go bankrupt though, is not a smart idea.

it was a fucking joke you mongoloid.
joke explanation because you dont seem to understand:
a majority of AI startups dont have a functioning ai, just a bunch of if statements.

[Legi0NHover]

it was a fucking joke you mongoloid.
joke explanation because you dont seem to understand:
a majority of AI startups dont have a functioning ai, just a bunch of if statements.

just shut up

joking or not you dont know jackshit

[int]

just shut up

joking or not you dont know jackshit

thank you mr xbox random capitals name, your input is valued!

[0xDEAD10CC]

it was a fucking joke you mongoloid.
joke explanation because you dont seem to understand:
a majority of AI startups dont have a functioning ai, just a bunch of if statements.

It doesn't change the real functionality of an AI just because startups can't implement it properly you baboon.. Learn to implement a neural network that can even understand to recognize digits before you comment next with something else that doesn't provide any benefit.

[int]

It doesn't change the real functionality of an AI just because startups can't implement it properly you baboon.. Learn to implement a neural network that can even understand to recognize digits before you comment next with something else that doesn't provide any benefit.

getting defensive after u realize ur an idiot for getting mad at my joke?

[0xDEAD10CC]

getting defensive after u realize ur an idiot for getting mad at my joke?

https://hackforums.net/showthread.php?tid=5857495

^ You don't even realize how stupid something like this is to benchmark with the information you provided in the thread. You must think all CPU's, and all compiler optimizations work the same... 🙃 It also goes to show that you don't understand CPU caching. How is someone like you going to ever provide something of value to this thread?

You barely know the basics, yet you attempt to make a joke here. You're the idiot. ❓

[int]

https://hackforums.net/showthread.php?tid=5857495

^ You don't even realize how stupid something like this is to benchmark with the information you provided in the thread. You must think all CPU's, and all compiler optimizations work the same...  🙃  It also goes to show that you don't understand CPU caching. How is someone like you going to ever provide soemthing of value to this thread?

You barely know the basics, yet you attempt to make a joke here. You're the idiot. ❓

imagine getting mad and using a thread from 6 MONTHS ago to try to flame me, rofl