Introduction into Stock Trading && Algorithmic trading

Come on, doesn't somebody on /biz know about algo trading?? Please help a guy out

>>958089
>where can I publicly read about proprietary trading methods guise??
When you graduate, instead of using your meme degree to get a meme job, work at a bank.

>>958154
Your reply is extremely irrevelent, but yeah working at a bank isn't >> being an entrepreneur....so do you know anything about algo trading to help me complete my meme degree?

File: 1444715511574.jpg (1004KB, 3741x2490px) Image search: [Google]

1004KB, 3741x2490px

>>958089

>>The Evaluation and Optimization of Trading Strategies by Robert Pardo
should answer most of your questions.

It takes years to get good at algo trading and most of that time is spent developing a war chest of strategies.

It took me just over four years to go from the first lines of code to making a living from it.

>>958542
Thank you I will look into it

>>958542

Thanks anon. Also check out: https://www.quantopian.com

Looks pretty cool

>>958589

Quantopian is a joke, but I appreciate the thought.

>>958805
Why a joke?

File: 11247362_10206596639425458_2051017525_n.gif (222KB, 367x206px) Image search: [Google]

222KB, 367x206px

>>958813
Laughably limited data resolution,
unrealistic order execution emulation,
sloppy scripting language for strategy,
zero support for optmization.

God forbid you actually cook up a good strategy, you won't be able to protect it there.

If you wanna be an algo trader, you'll need to build your own tools to do it properly.

Retail traders don't stand a chance with 'black box' trading. You simply cannot compete with the algo and quant guys at mega firms.

On the flipside, you -can- use your discretionary tactics to try to gain an edge, but the only real things that work (when removing discretionary flaws such as emotional responses) are risk management and a staunch trading plan with discipline.

The quant guys will win every time if you play their game. They just will. Your game should be developing a sound discretionary system starting with risk management and sticking to your guidelines. There is no system to predict where a stock will be tomorrow with great certainty, or any certainty at all. All the algo guys trade milli second to milli second.

File: silver.jpg (95KB, 753x500px) Image search: [Google]

95KB, 753x500px

ITT: We talk about the only thing that works in trading stocks as a retail investor: risk management and a strict trading plan.

Years of trading has lead me to the conclusion that there's only a few things we can do as retailers to grow our money let alone not lose all of it and that involves risk management and strict trading parameters and not deviating from them in terms of cutting loser trades immediately and allowing winners to run 'on a leash.' Maybe your plan is to cut a losing trade when it loses 2% while on the flipside your plan is to capture 6-8% for each winning trade. Now, circumstances will vary from trade to trade such as the environment overall down to smaller niche circumstances but our overall risk:reward ratio should always be 1:3+ if we want to win in the long run.

Portfolio management goes hand in hand with the risk management aspect, never risking more than % amount of your portfolio in just one trade. However, the downside to this is we (mostly all of us) are starting out with small portfolios. 200 dollars in one trade to make 20 dollars is not going to get us anywhere soon, nor should it (things that come fast go leave faster) but we can't put all of this effort and time into trades to earn a measly 20 dollars (assuming we start with a 10k) account.

So, what are some of your trading protocols to remain in the game, yet earn an above average return? I know there's a few guys with high-high risk plans that are working currently, but just as everything else, unless you are incredibly intuitive and can shift with the change of the markets the outliers will eventually burn your plan to the ground.

Let's bounce some ideas around.

>>959868
I had a weird idea to analyze the S&P 500 and analyze it for factors that were independent of the markets - basically looking for animal spirits whose movement could be predicted in the future (the ol' 'recessions happen in Fall' kind of things). I got a statistically significant correlation algorithm, though it only accounted for a tiny amount of the overall effect. I figured it might give me an edge if I used it to predict when I should rebalance.

Trouble is, it consistantly told me that I should invest now rather than later...

>>960362
What's the general idea? Also, what time period? You could've been curve fitting unknowingly and this out of sample may fuck you over. If unsure, test in a paper account for a while before committing. A lot of strategies work really well until they don't and then they are woeful. Its not even a case of getting crowded out most of the time but a case of regime change in the market.

>>960362
>dimwit doesn't recognize that 'time in market' is a statistically significant return factor
Why do you think you'll alway lose to indexing? Because its not what you buy; its how long you own it.

>>960590
>Because its not what you buy; its how long you own it.
what. the. fuck. That's the most ridiculous thing I've ever heard. Its all about what you buy. Selecting indexes is an example of being selective with your choices.

Holy shit, please leave this board. Way too meme for me man. I mainly do macro indexes and focus on portfolio theory but efficient market/ index fanatics like you are fucking retarded. If you find systematic mispricings in the market you can increase your return - just look at Buffet who's alpha is purely the low beta phenomenon and cheap leverage from the insurance company.

>>960595
Kek. Check the science noob. Asset allocation is one of the least statistically significant performance factors.

Or keep praying to your ouija board fantasies. You'll still be poor, but you can roleplay with the best of 'em.

>invoking Buffett in his defence
>how I know you're full of shit

>>960603
>invoking Buffett
next time someone will ritual summon graham :-D

>>960588
Basically, use astrology to predict the animal spirits which in turn predict the stock market. The time period I fit it to was from 1954 and the longest cycle I analyzed was 12 years. So I could certainly have been curve fitting with those, but I also tested it against other indices and they all showed pretty similar patterns.

It's a really small factor, so small it would take a while to reproduce in a paper account. But it's also one I bet typical models wouldn't show. I think though that I would have better luck prettying it up and selling it as a 'scientifically tested horoscope generator' than using it as the basis of all my trading.

>>Trouble is, it consistantly told me that I should invest now rather than later...
>Dimwit doesnt recognize that 'time in market' is a statistically significant return factor
Dafurq? I was joking about that very effect. In doing my analysis the time in market overrode every other factor, so in the end my weird and unconventional analysis just confirmed the same truism everyone learns in investing 101.

Are you the same guy that told me to just buy American last time I posted about my portfolio? Because if so that's kind of spooky.

>>960929
second half was meant for this guy >>960590

Hey guys, I'm a junior software engineer.

Are you guys actually coding algorithms to predict stock movement?

>>961008
In my case, I'm coding algorithms to learn about genetic algorithms and machine learning. If the algorithm happens to be better at picking stocks than a random number generator, that's an unexpected but very welcome bonus.

File: 1420437804174.jpg (28KB, 600x600px) Image search: [Google]

28KB, 600x600px

>>961008

In my case, Yes, but "predict" is a strong word. It's more about anticipation.

You can't predict your chess opponent's moves before he makes them, but you can position your pieces to exploit the moves he is most likely to make.

If you're very good, this "positioning" cannot be easily turned against you.

>>959868

you can trade derivatives if you have a smaller account. new traders can make more through the leverage they have. however, they can, and often will, also lose more.

>>958089
If you're an undergraduate, I would just soak in everything taught. Know how databases work, know what the various OS's are, know how data structures and algorithms work.

Pay attention in discrete math and statistics and probability. You may have classes or electives in machine learning or AI. You may use things from continuous math classes as well.

Just learn broadly a little bit about everything they teach, while also focusing on discrete math, statistics, probability, machine learning, artificial intelligence etc.

Other things can factor in as well. Just knowing how to clean up data quickly is helpful. Hedge funds have a division of labor, and being able to build tools for the quants can be lucrative as well.