algotrading

>>1457489
Trying to learn it but fuck I am so dumb

File: guacaveli.png (488KB, 600x450px) Image search: [Google]

488KB, 600x450px

Java/C++, Interactive Brokers, Casper, Proprietary.

>>1457489
Seems like a meme desu. I can make decent money trading already and modern algorithms aren't making as high returns as when algorithm trading was new.

>>1458285
algorithms will always do better than people though desu fampai

>>1458285
As an algo trader who once traded by hand, I feel precisely the opposite way. I make more from "Algo Scalping" than I ever made the old fashion way.

I think we are two sides of the same ecosystem, senpai. The Marlins and the Trout.

File: 1462921646307.jpg (77KB, 777x728px) Image search: [Google]

77KB, 777x728px

>>1457489
My algo deals only with dank memes.

>>1458271
Are you doing well? Stock or forex?

>>1458581
Yuhp. Stocks.

Quantopia.com

Then realize you aren't smart enough to compete with the rain men doing this.

File: 1471598637631.jpg (224KB, 940x672px) Image search: [Google]

224KB, 940x672px

>>1458778
Did you write it yourself? Was java for the interface to the sites? Just you working on it or do you have a team?

I am finishing up my engine, soon I can work on communicating with the websites. Not much of a web Dev, but I can and will learn.

No one here is doing algo trading unless they work at a big firm.

Also algo is literally killing the markets

>>1459084
You need to get on the Interactive Brokers API and skip the websites thing.

>>1459042
>Quantopia.com
>Rain men

>No one here is doing algo trading unless they work at a big firm.
At my business school I once gathered a few students to start a Algo trading project (website smartphone-capital.com) and we were straight business majors without much IT background.

Trading was quite good for the first two years with +30% with very conservative mean reversion strategies. However the last years didn't go so well. It's not that we had big losses, but we didn't have wins either and the market was just in a kind of twilight zone were none of our strategies would really go to work.
Actually I am thinking about stopping the project altogether since it just steals my time and my discretionary trading and analysis of crypto currencies has yielded far far greater rewards.

>>1459128
Btw. we used interactive Brokers which was connected to NinjaTrader where we just had to program the actual strategies without having to deal with the API stuff beyond selecting the right ports.

>>1459128
>website smartphone-capital.com
Meant to write smartstone-capital.com

File: 1471591630347.jpg (76KB, 865x604px) Image search: [Google]

76KB, 865x604px

>>1459115
They have an API? This will be much easier than expected, thank you kind sir.

https://www.interactivebrokers.com/en/index.php?f=1325

Holy hell, I cannot thank you enough for this. They even have c++ specific.

>>1459184
discord gg/92xaj
ask for guac

File: 1457730297153.jpg (3MB, 1082x9999px) Image search: [Google]

3MB, 1082x9999px

>>1457489
we need a discord specifically for algo trading. can some1 make dis

>>1459267
does this work?

https://discord.gg/hBrE29J


https://discord.gg/hBrE29J

https://discord.gg/hBrE29J

>>1459267
>>1459273

ahem
>>1459265

>>1459275
could u post the link plz. with the whole URL.

>>1459275
Sorry, I was coding. ntx2.

File: Big-Sur-top-view-without-cover-GPU-removal.jpg (1MB, 1920x1100px)

1MB, 1920x1100px

As much as I hate memecoins, one thing Bitcoin has going for it is that you can easily get bid/ask data for free for most of the major exchanges going back a LONG time.

There's also the history of every trade on every major exchange ever available: http://api.bitcoincharts.com/v1/csv/

And those exchanges never close.

Has anyone tested strategies against memecoin exchanges as a proof of concept? I'm currently working on some GPU-based neural networks with my Walk Forward Analysis framework for bitcoin since I just got a GTX 1070

My goal is to figure out some interested non-price "features" (NN inputs) and eventually use them on either forex or futures trading

>>1459477
interesting*

And I'm using ArrayFire if anyone else is interested in getting into GPU-based ML/NN/deep learning stuff:

http://arrayfire.com/

Bump.

Also get in the chat >>1459273

>>1458299
>"Algo Scalping"
Are your algos no directional bias and make the market as in trading between the bid/offer, if so how do you even compete?

>>1460509
i too would like to know this

HFT intern here ama

>>1460534
>HFT intern here ama
How do I get this?

>>1460558
you don't want it. trust me. i work on the s&t inst. floor at morgan stanley 1585. HFT is looked on as a fucking joke

>>1460919
Dont IB guys look down on everyone?

File: 474955430.jpg (29KB, 320x320px) Image search: [Google]

29KB, 320x320px

Couldn't make my NN recognize text.

>>1461037
Not even the MNIST dataset?

http://arrayfire.org/docs/machine_learning_2neural_network_8cpp-example.htm

Bump.

There's also a forex-related chat here which is more active: https://discord.gg/HmxfP

>>1461084
Thanks man, it's been years since last worked with c++.
Any good book about algotrading for beginers?

>>1461668
For beginners? No idea. I recommend "The Optimization and Evaluation of Trading Strategies" and "Master the Markets (3rd edition)".

The former explains in detail the "walk forward optimization" which is necessary to avoid overfitting and prove robustness. The latter has the best explanation of the significance of volume that I've come across yet.

http://bookzz.org/s/?q=Robert+Pardo+The+Evaluation+and+Optimization+of+Trading+Strategies+2nd+Edition&yearFrom=&yearTo=&language=&extension=&t=0

>>1460534
what degree?
what GPA?
anyone from non prestigious schools that works there?
Much experience working in finance beforehand at lower level stuff?
Had experience trading-was it successful?
What programming languages do you know?

>>1459098
>No one here is doing algo trading unless they work at a big firm.

big firm... kek

it is small, lean and innovative firms that are cleaning up in algo trading... big banks etc.. can't compete against the likes of jump, virtu, KCG, XTX etc..

>>1459136
and no doubt abstracting yourself from the micro-structure of the relevant exchanges and using some pseudo C# language to generate 'signals' from some TA style indicators...

this is the wrong approach - forget about ninja trader - you'll just end up in a big waste of time curve fitting exercise

>>1459477
>My goal is to figure out some interested non-price "features" (NN inputs) and eventually use them on either forex or futures trading

what makes you think that this egde you intend to develop in the bitcoin market will be applicable to FX or futures markets

though people do a lot of data analysis in systematic trading it isn't completely in isolation - if you find what might be an edge it is good to be able to attribute a cause too else what you've found may well be spurious

>>1460919
it is a joke at morgan stanley, they're not able to compete which is why they sell access to their dark pool to others

HFT guys you've got are mostly glorified developers working on execution algos

as far as electronic market-making on public exchanges is concerned you'd get slaughtered along with any other bank

>>1462670
I just think it would be interesting to see if there's any crossover. I don't have any expectations of ever actually building a successful trading algorithm. I'm in this for the intellectual curiosity, not the 1% chance of hitting it big.

>>1463165
you might find a general momentum type strategy/trend following approach could be applicable to all of those markets, the edge is probably going to be quite small though (or at least it is in futures and FX) - but that is one thing that might well exist in BTC etc.. markets too

problem is that if you really want to exploit that sort of strategy the thing to do is actively trade a few dozen markets simultaneously otherwise there is just too much variance... small edge and markets are full of noise

large CTAs take this approach with say 50 or so different futures markets + FX forwards

>>1464357
Thanks for the ideas

>>1464868
I run a few algos in the background of my manual trading. Mostly to control and dynamically adjust risk etc.


Works well. I make the trading decisions and the algo keeps me disciplined

>>1463165
look for the whales