/biz/ - Business & Finance

lol

>>502978
>INSTEAD EXPLAIN HOW THE EQUIPMENT WORKS

You buy some sort of data feed or direct line to the exchanges. These feed data to a computer you own with programs you wrote on it, and, your algoirhtms monitor the stream of data and send buy or sell decisions as nessecary.

>HOW DOES THE LAW WORK

Same as for regular investors. You're still subject to SEC, CFTC, etc. regulations.

>HOW DOES THE BUSINESS WORK

You buy and sell financial instruments with the intent to make more money than your expenses.

>>502993
where do i get these data feeds, what are these data centers called where they are in close proximity and such?

>>502995
also what kind of message does my computer have to send out to where to make a buy or sell?

>>502988
my thoughts exactly

>I want to start a specific type of very competitive business with high barriers to entry but can't Google the basic business model
Great success pending

>>503008
google or the public dont know shit about this

>>502995
>where do i get these data feeds

Interactive Brokers offers direct exchange access for $10k per month, I believe. A few other online brokers also offer direct access. It's expensive. Expect to spend 6 figures a year at least.

>>502996

Just use the API that they offer.

>>503013

Lol you're clearly retarded. I see I've been posting in a troll thread.

>>502978

I've been doing it for years. Get ready to spend a few building your own trading engine, then another few cooking a library of robust trading strategies.

Interactive Brokers is your only option for a broker with an API.

Start with
>> "All About High-Frequency Trading"
and
>> "Quantitative Trading: How to Build Your Own Algorithmic Trading Business"

Read both of those tonight, they'll answer all of your questions.

>>503455
can i do it from the west coast on a home computer or will i be kilt by latency

>>503494
you don't want to do this on a home computer. rent one in NY

>>503494
closer to the exchange the better

how do i compete against mit whiz kids who do this 10 hrs/day

>>503527

By doing it 12 hrs/day.

You must live and breathe Algo Trading. It isn't for everyone. It's a very lonely road. Think about if this is really for you. If you decide it isn't, that's nothing to be embarrassed about.

But if you really want this, go read those books and post back here when you're done.

>>503539
>Zero Lag Exp Moving Avg

Thank you, I'm looking into this now :)

I built a series of micowave towers betwen NY and Chicago to exploit price movements between the two exchanges

I made mad dough.

If you look at any stock level II orders. yeah brah thats me with the 100 lot orders on each side

>>503546
Why are you not still makin mad dough with those?

>>503553
cause some fucker drilled though a mountain and placed some fiber so he scoops his orders right before me. Its all good though, I've got a room in the nasdaq

>>503561
why don't you just use all that money you made to manipulate the market at certain times

>>503455
thank you my /b/ro

maybe i can buy you dinner at the gold club next year

>>503517
how do i get my machines in the same server rooms as the exchange?

>>503539
>very lonely road
i dont care if it was a road of rusty nails and shoes where out of fashion.
as long as there is CASH at the end of the road i will MAKE it

there is two types of men in the world
winners and then there is people who give up when the going gets tough

>>502978
http://en.wikipedia.org/wiki/Series_7_Exam

>>503455

It's been two days. Did anyone do the homework?

>>504764
anon what should i study in uni if i want to work for a quant company

>>504764
I think I had that Ernie Chan book somewhere on another laptop, but I'm reading it now and it's pretty good. Wondering why you recommended a 5 year old book, though.

I have two questions for you:

What instruments do you trade?
Do you use Matlab and Excel for your analysis? (I'm much more familiar with R and Python, so I would want to port his examples)

Thanks.

>>504960

The books are a tad dated but they're only to cover the FAQs. I've got a library of titles to suggest if anyone starts asking more apt questions.

>> what instruments do you trade?
I trade nasdaq securities exclusively for now. I'm experimenting with currencies and foreign securities. The goal, of course, is around-the-clock trading.

>> what tools do you use?
I built my own analysis and trading platform specifically for designing and executing autonomous strategies. It seemed the proper way.

>>505041
Thread was pretty kek until you showed up, so I'll bump one last time with some more questions if you're up for it:

How many self-directed guys are there that do this successfully do you think? Do you know any? Is it really as lonely as you say?

>I built my own analysis and trading platform specifically for designing and executing autonomous strategies.
Mind if I ask what language(s) you went with? Do you use a traditional RDBMS for storage or is it something exotic like kdb+?

Do you incorporate order book data or regime analysis into your strategies, or is it just 'simple' stat-arb on raw time-series data?

Lastly, what do you tell friends, family or the guy you're sitting next to on a plane what you do for a living? I was in financial services for many years and it was easy to discuss work back then, but I wouldn't tell my dog that I was a quant/HF day trader.

>>506331

Quant trading != HFT. You're buying or selling an instrument in the expectation of a price change, not in expectation of earning a spread.

Also note that when people ask after your job, replying "I'm a day trader" == "I'm an unemployed moron".

>>504194
The problem is that there is also cash at the beginning of the road, so you'll also have to make that.

>>504194
Simply epic

>>506331

>> is it really so lonely?
Never met another. Tried to mentor a few, but they flew.

>> order book data
Moot. Pure noise. Ice bergs, stops, book stuffing, etc.

>> regime analysis
The holy grail. The key to ForEx. I've got a few promising threads of research there, that's all I can say. 'Simple' stat-arb, or 'Simple' mean-reversion seems to remain king.

>> language, database
Java. Everyone loves to knock it, but it's still the best for uni-platform concurrent processing applications. "Write once, run anywhere."

I use in-memory databases to save microseconds. Casper datasets is a pretty great library for it. Easily tailored to your needs. More to the point, all of the trading math is built from the ground-up to run in an on-line capacity; which is to say it's built not to require any iterating over past data except in a few unavoidable cases like donchian-type math.

>> what do you tell people you do for a living?
"I'm a Computer Programmer." is my go-to answer. Usually followed by "It's very boring" and a subject change. Like I said, it's a lonely road... but if she's hot I'll say "I teach computers to trade stocks."

>> this thread.
I always prefer to piggy back. Keeps out the riff-raff. See you next time, Space Cowboy.

>>506579

>nobody is good enough for me to mentor
>order book data is useless
>simple stat-arb
>>>>>java

You sound like a massive faggot, anon, and a shit trader to boot. Writing Java programs to trade moving averages on the NASDAQ doesn't make you a badass.

At least you have the decency to not go around telling people you're a quant.

>>506579
>Java
what happens when a garbage collection routine interrupts your HFT algo between a calculation result and its network request?

>>506579
Where are you going?
I too have questions!

>>506598
In all seriousness anon, what say you for prospective HFTr's who do not have privileged access to something like BATS?

Is there any way one can just paddle out there with ~30 thinkcentre workstations and expect make any sort of gain? (considering you have a platform already)

What can give a goober some sort of edge today?

there was an awesome course on coursera on automated trading. it didn't do HFT of course, but it covered what you want and how to implement it in python (but i followed along in C++ instead, just personal preference)

>>507115
cool!
https://www.coursera.org/course/compinvesting1

I'm getting into machine learning myself. Still haven't figured out exactly what my infrastructure will look like, but I'm already focussing on mining the necessary data.