/biz/ - Business & Finance

I'm going to go out on a limb here, but since you're posting this question on a Papua/ New Guinean comic book board, you're probably going to get your ass handed to you if you try this.

>>836260
You could make money writing crypto currency trade bots (arbitrage etc) and selling them

>>836287
getting my ass handed to me, do you mean id lose all my money because im not smart enough or does that mean something worse.

>>836324
thats interesting, but im wondering if there will be a comeback of crypto currency, it died off a bit recently.

>>836260
I am an algo trader. You can make money, you're not obligated to compete with big firms.

Most people don't know jack shit and still talk about this topic, so my suggestion is to go and make your own conclusions.

You'll either meet losers who will discourage you, winners who will discourage you to avoid competition and people who know nothing and tell you it's pointless because "muh high frequency" or some other random shit.

It's a very interesting, challenging, competitive, lucrative and clearly GOAT field to be in. Depends on your motivation to succeed.

pic related, one of muh algos.

>>836260
Algorithmic trading is profitable specifically because it allows someone to trade on very small time frames on a wide range of securities that until recently weren't really taken advantage of.

So long as your algorithm actually works on whatever time frame you think is appropriate, you should be fine, although you'll want to take in to consideration various aspects of the process, including how fast your computer running the algorithm can communicate with wherever you are entering orders and how stable your connection speed is.

>>836330
thank you thats very helpful, im very interested in this field and its good to know its not entirely saturate/rigged

>>836332
is speed really an issue, i assume you will be having at least a second delay, but when youre not competing with any big firm it shouldnd matter too much, and stability shouldnd be an issue either unless youre having some 3rd world internet connection.

>>836330
Pretty noice, interesting

May I ask what's your education sir?

>>836342
msc in finance, bsc in economics, worked 2 years as a risk quant and 1 year as an algo trader.
>>836340
speed is not an issue if you trade normal frequency. You can have perfectly good algos in 1min or 5 min timeframes where speed is not a great factor. Especially if your holding period is more than 10 minutes.

>>836332
I agree with everything you said.

>>836345
Thanks for the answer

The thing is that I'm currently in a similar situation (without working experience apart from and internship, I finish now in uni).

I was thinking about studying Django then C++ by my own (my brother is a software engineer with great experience so he'll help me out what it), do you think that this is a good idea?

>>836345
Medpack?

>>836260
Does one need to know economics? I'm just an engineer with acceptable math intuition but almost no formal economic schooling.

>>836260
Can you guys recommend some good resources about algo trading to a software engineer with very little economic knowledge?

>>836395
>>836425
op here, im not the one with the knowledge there are a lot of tutorials and books on the topic that can easily be found on the internet.

>>836330

I don't understand, your profit margins are literally fractions of a percentage point. Don't you pay commission?

>>836466
This is a good point. But maybe the image was just a simulation or something. Pretending to buy and sell on old data to see if the algorithm seems promising.

>>836466
the contract you're seeing has a multiplier value of 100. So each tick is 10$. My trades make on average 100$-500$. Comission for a trade is 4$ so it's a low fraction of the trade.
>>836383
yes, it's me
>>836372
my experience as a quant is that you need either C# or C++ or R or Matlab and VBA to do normal tasks.

>>836713
100$-500$ a day?

>>836713
Understood, I take note of this, thanks again

Let's see where I can get in the upcoming future

>>836330
Hey man, I'm very curious

My degree at uni is financial engineering -- first in the country. It's a mix of stats, some basic CS/SENG, basic maths and a fair bit of finance.

To get into something like that -- would this be a good path?

cheers.

>>836713
Lel, thought so.

>>836771
no, per trade. But not everyday has a trade and not all are winners, so my average gain per trade for these algos is 30$, which is good.
>>837072
it's a great path. If I had to redo my education knowing that I'd want to be an algo trader, I would've taken a bachelor in maths with some kind of courses in programming and then financial engineering as a master.
>>837176
zip on biz. Lel

>>837185
what kind of things should you look at for algorithmic trading? Machine Learning - Backpropagation - Examining and looking for correspondence in stocks etc Analysis
>What else should I research
>Have programming knowledge and experience, but no finance or stock related knowledge.
I understand I probably wont be able to create an efficient enough algo.
But im sure it could make for some interesting reading.

>>837185
I've been a lurker since its introduction, used to be active when /sfig/ was a thing but it's been a while so primarily lurking now.

>>837185
I've been a lurker since its introduction, used to be active when /sfig/ was a thing but it's been a while so primarily lurking now.

>>836330

We should talk.

Setup a disposable email and post it.

>>837220
When I started working as an algo, I had no idea how to do one. My boss came to me and told me basically his idea of a strat and I had to code it and see how it performed historically.

It was a rather basic idea, let's take for example a double bottom chartist pattern. You then try to recreate statistically the environment in which such a pattern occurs. For instance if you trade in 1 min timeframe, you need a horizontal channel with the 2 bottoms sufficiently spaced and blablabla, you get the drift. I don't do machine learning and backpropagation, I just watch the time series given to me and try to find something interesting in it and attempt to recreate that environment. There are a lot of different ways to algotrade and I can only offer you my advice on what I know.

Regarding learning, my thoughts are that one idea can be successful or a failure depending on the skill of the algo trader. Therefore, if you see in a book that "we did X, we got jack shit", you never know if they failed at coding or they did it correctly. It's very easy to fuck up in this field. If you want to exercise, just try to build something very basic to understand it.

For example build a buying algo which enters a position when the price is below the bollinger band. I think you learn by experience, that's what I did. Go for easy at first: 1 asset, long-only or short-only.

>>837297
hey algobro ! you can reach me on dmadan3@gmail.com

>>837220
Also, I use MultiCharts to build my algos. You can get a free trial if you want to have a look at it.

See pic related for an example of algo. you set some conditions and if they're fulfilled, then you short.

>>836425
>>836395
I don't think you need to know economics because trading based on macro economics is not done at algorithmic level but long-term investment. Like by economists who understand what impact random speech of the federal reserve will have.

You need no economics understanding as far as I'm concerned. You need to be curious and inventive with a desire to learn to succeed in algo programming. Also, it's best if you manage to find some kind of mentor to show you the ropes but not mandatory.

>>836260

This is pretty cool anon. No doubt about it. I've never really gotten into quant/algo. I once wrote a script on C++ that analyzed the prices of stock over the last few days and then determined which were near support or resistance, which were trending sideways, etc.

However I had to use data from Quandl that was unreliable and spotty in some places.

Where do you get your data feed? Do you pay for it? Is it realtime or delayed?

>currently running an algo I've customized

- should I switch to a day trading algo?
- what metrics are ideal to gauge performance/pitch to someone?

>>837392
sharpe ratio , avg trade per win, number of trades, timeframe. Max drawdown, time to recovery.

>>837400

Thanks. I feel sharpe is skewed by leverage/timeframe. Is it still applicable?

hey nerd what do you think of sites like quantopian?

is there enough info there to learn algo trading?

>>837351
>>837357
Thanks for the insight, Sure will make for some interesting research. One thing I found when conducting my own research is that it's an extremely secretive field, very few people want to share knowledge.
>If anyone has more small tidbits of knowledge, dont hold back!

>>837401
there's a different sharp which takes into account skewness, but we're talking about annual sharpe so timeframe should not have an impact.
>>837413
I agree totally, nobody will give away their golden goose. Why would they ? It's fucking hard to build one and they gain nothing by sharing their info.

To the algo users here, do you use your whole account on this to maximise returns and minimise the impact of commissions?

FX or stocks?

>>838624
Should I just stick to playing poker?

>>838611

Can you rephrase the question?

>>838659
Say you have $10000 in your account
Does your algorithm use the entire $10000 every time it trades?

>>838786
no. position sizing is important. you cannot risk all of your equity on 1 trade.

>>838791
How much percent of your account do you risk?

>>837351
Any classes you reccommend I take at my local community college to get started?

Do I need good understanding of advanced arithmetic such as calculus or something of the sort?

>>838786
Hell no. Even when a human is trading, you never use your whole capital...

>>838815
What percentage do you use
Or does it vary based on statistical advantage say that for whatever reason a particular stock would rise in the opening hour
Just want a rough estimate to satisfy my curiousity

>>838831
>>838798

It's not that simple.

Depends on your strategy, what you're trading, what your margin limits are, what your risk expectations are, ideal diversity, etc.

A lot of variables go into it.

>>838803

Computer Programming. Java or C/C++ only. Anything else is a waste of time.

There's a buttload of math, obviously. Reading formulas and translating them into code is an essential skill but you don't need to be able to do calculus in your head or anything like that.

If you want to learn it, just start playing with it.

You don't go to college to get good at World of Warcraft, you just sit down and start playing. You learn the ropes and eventually you'll be pretty good at it.

This stuff has always interested me.
I'm a junior CS student, good at C decent at C++.
I've always wondered where do you find a library or api for the stock market?
Does etrade have their own or something?
If I could find a decent library or api I'd definitely give this a try.

>>839175
https://www.interactivebrokers.com/en/?f=%2Fen%2Fsoftware%2Fibapi.php&ns=T
https://us.etrade.com/active-trading/api

You can also download daily data from yahoo at no cost if you want to try it out.
http://ichart.yahoo.com/table.csv?s=AAPL&a=0&b=1&c=2000&d=0&e=1&f=2015&g=d&ignore=.csv
This link will download the open, high, low, close and volume of apple from the first trading day after 1 January 2000 to the last trading before 1 January 2015. Months go from 0 to 11 in the link.

>>839234
Ah, exactly what I was looking for.
I found some places like tradeking and interactive brokers. But I don't want to give my social and sign up for something just to play around with some numbers.

I guess I'll try to find the "bollinger bands" as previously mentioned.

Thanks again for the help.
This seems like some decent documentation too: jarloo.com/yahoo_finance/

>>837420
to elaborate, sharing could mean that another party could build one relatively cheaply to exploit it specifically. the more its shared, the more reason for a predatory algo to be built against to identify and exploit it. its not just selfish jewbagging.

>matlab masterrace reporting.

So someone here give me advice.

I have no background in either programming or finance (Engineer here). I did pic related with python, but am good with VB and C++, and can dable with MATLAB. I know statistics and probability. If I continue self-learning - time-series, Brownian motion, and continue with programming, will self-learning be enough to break into the finance industry?

>>839175
Python at least has a Yahoo Finance module

>>839242
http://investexcel.net/free-intraday-stock-data-excel/
https://developer.tradier.com/
http://www.quantshare.com/sa-426-6-ways-to-download-free-intraday-and-tick-data-for-the-us-stock-market
I found some places to get intraday data for free. I haven't used them yet so i don't know how well they work.

>>839356
I've always been interested in Finance but more specifically the quant stuff.
I'll probably never land a job at a bank because my mediocre university and gpa but you can definitely go for it on your own time.
And if you get something good, put it on your resume and if it's worth something I'm sure you could get picked up.

>>839356
>dat pic

I fucking love the Sentdex guy, he has the best tutorials.
He never goes to deep, that is your own job but he gives a great overview.

10/10 guy, learned so fucking much from him.

>>839356
>Python at least has a Yahoo Finance module

And if you are really interested in Python Algo trading, check out Quantopian.

You can build your algo based on both technical and thousands of fundamental metrics, all easly available through their API and backtest everything for years.

It doesnt get any better than that if youre interested in experimenting.

i use trading algorithms in order to build a reservoir in order to
A:do statistical analysis
and B: better predict who/what population is going to trade when.
i never have it set to any function of buying and selling, its more a source of information than anything, a library to predict what certain ai's with specific goals, tendencies, and methodologies will do. what they would be looking at when and when they would most likely buy
i have to say if you're going for trading algorithms it pays to do a few things before your hands hit the keys
know that the bugs will be horrendous when working with most ai
look into the cognitive science of trading and assessment
look into various types of artificial intelligence
also C masterrace phoning in

What's an entry level bankroll to start with?

>>839807
25k Minimum.

Thank the SEC for that.

>>839173
Great, thanks for your help.

I might re take a math class since I've become a bit rusty.

Can you give a more specific explanation of what type of math?

Maximums and Minimum type of stuff, quadratics, etc, etc...?

Would a statistics class be useful?

I was thinking about taking one.

Thanks again

>>837420

So just ran these metrics and really like the results. Any money mgmt pointers?

Does yahoo finance lie about graphs? The csv that returns gives an average of 118 dollars a share in 1997 but on finance.yahoo it says it was trading around the lower 20s.

Am I missing something?

>>840054
INTC is the company

>>839955
Everything i know about algo trading is from reading articles on the internet. So take care to also do your own research.

A big part is position sizing, max percentage of your account you want to risk per position, knowing were to place your stop-loss, making sure there is enough liquidity, etc. As long as you know your way around percentages this should be fairly easy.
The other part is getting your buy/sell signals. Unless you want to make your own indicators, you can just look most of them up. No real math needed here.
http://www.investopedia.com/active-trading/technical-indicators/
If you want to use chart patterns instead, you'll need an easy way to compare a chart to a pattern. Working with percentages and a maximum deviation might work. I haven't looked much into this.

I don't think taking any classes will help you much. You'll most likely just end up learning a bunch of stuff you won't be using (not that it's a bad thing). If you're going to college anyway computer science is probably the best way to go about it. Personally i just do everything through self study and look things up when necessary.

Not he person you're replying to BTW.

>>840433
Great thanks for the help.

I don't mean to sound lazy, I just wanted a good good grasp of what it involved.

I don't believe I will be doing it anytime soon though, I first need to get the gist of using fundamental analysis before moving on.

I will definitely be taking a java or C++ class in the near future though,

I might actually just end up finishing up my AA in accounting and then pursue a degree in Information Systems.

Thanks again

>>836330
>>840433

How complicated are your trade signals? Do you use ML like ANNs, SVMs, GAs, etc? Why isn't buying on resistance-breakout, selling higher a suitable algorithm?