Thread replies: 8
Thread images: 1
Thread images: 1
File: 1473921687432.jpg (30KB, 514x536px) Image search:
[Google]
30KB, 514x536px
OK so i'm guessing this has been done 1000 times but im just thinking about why this would or would not work. I am an electrical engineer with decent coding skill (Java/C/C#) and I have an interest in the stock market.
Steps:
1) Create a Neural Net and train it on the shit load of stock history data until you get a net positive outcome for arbitrary stocks. (So the net will make on average a return of 10% in a month say)
2) Get around 1000 USD and let the Neural Net have fun and start trading.
3) Profit?
>>
>Implying all the big trading houses don't already have million dollar software running 24/7 trying to do the same.
Go for it. Just understand that you've got some really well financed competition.
>>
It's already been done and its already in widespread use. They have mixed results. They do really well with futures market stuff but suck at picking stocks.
Do a google search or something dude.
>>
I don't understand why this is a problem (since I dont really have a huge amount of experience in finace). So surely the trading houses are really well financed and can operate much faster and make transactions much faster than I can. But why is this a problem? My system may pick up different trends than theirs making it so that we are not competing.
>>
>>1669788
"billion".
>>
>>1669796
do you seriously think you can outdata, outmath, outcompute the big boys?
reality: they have teams of people smarter, better connected and with numerous technological edges over you. you are talking about probably 50+ years of data. and each time period will start to behave differently due to macroeconomics and technology. you can build something that will overfit and have crazy backtested returns and then go to complete shit over the next year. what timeframe? 1min, 1hr, 1day, 1wk, 1mnth. all those timeframes behave differently.
let's assume you have a 120+IQ. ok. then build a program that can sniff orderflow from the firms and piggyback off that. i don't even think you you're that smart. because 1000 with commission of 10 in and out is already a 1% loss. then your order needs to cross the spread. that'll be another 1% probably. that amount of starting capital is good for gambling. nothing more. the fact you are even suggesting to start with 1000 tells me you haven't read anywhere near enough about the markets.
10%/month over 10+ years would make you fucking legendary. books, courses, schools and trading systems will be named after you. you will literally be the man that breaks and makes markets.
you and i are 30 years too late for this game.
>>
>>1669777
If you can consistently hit 10% returns on your investments, you will be the wealthiest person alive in 50 years time.
>>
Thread posts: 8
Thread images: 1
Thread images: 1