Matlab

ZIP file of presentation of Matlab C++ Open Source Automated Trading System

ZIP file of presentation of Matlab C++ Open Source Automated Trading System This ZIP file with the ATS's details and live demo that I had talked to you about. Furthermore, in the same file, you will find a brief presentation with the tools and strategies, that the team uses, as well as what else we can offer in Discretionary trading activity as well. The doc file, is some of our strategies performance, after paper trading AND live trading via the TWS from IB.. We are using MatLab and C++ (mex files) four our development process... We can support BOTH low and high latencies trading processes... Get in touch me with me at for contact: http://quantlabs.net/labs/about/contact-us

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Video on Matlab query of parallel computing with RSI and Moving Average using High frequency trading HFT tick data

Video on Matlab query of parallel computing with RSI and Moving Average using High frequency trading HFT tick data Question from a visitor: Hi there, I watched you Youtube video, and you referred to the Mathworks webinar (algorithmic trading webinar). You said it takes couple of seconds using parallel computing, however on a quad core machine for HFT data, like you were talking about it would take several days for optimization using RSI or even mov avg Best answer from: http://quantlabs.net/quant-member-benefits/hot-hft-high-frequency-trading-tips-and-tricks/ My answer is: YouTube Preview Image  

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Anybody knows or has experience to conect Matlab with quickfix in c++ enviorment? Runs ok to recived data in real time?

View CommentsWritten on May 1st, 2012 by caustic
Categories: Matlab
HI, I have a question. Anybody knows or has experience to conect Matlab with quickfix in c++ enviorment? Runs ok to recived data in real time? Many thanks.   -- Hi! I have written a couple of similar applications (not for quickfix though). There is quite a high impedence mismatch between the operational models used by QuickFix (callbacks) and MATLAB (batch/vector processing), so you will probably need to implement a queue to sit in between QuickFix and MATLAB. A simple C++ quickfix client can be used to handle the callbacks and write messages to the queue. (Use Protocol Buffers maybe?) Then, depending on your use-case, you can either: 1. (PULL) Write a simple MATLAB mex extension to read large-ish batches of messages from the queue on demand, so your colleagues can very easily write MATlAB scripts that poll from the queue. 2. (PUSH) Write a small C++ program to read large-ish batches of messages from the queue, then pass them to a MATLAB script through an invocation of the MATLAB engine. In the past I have always gone for option (1) as it gives my colleagues a more familiar model when they write their MATLAB scripts to make use of the data. You cannot get around the fact that you need to trade-off latency for throughput. You also need to accept that you can neither get very low latency nor very high iteration rates from MATLAB (or any other interpreted scripting language for that matter); but by batching items together you can get reasonably high data rates.  

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Getting the daily return from a Pairs Trading Strategy using Matlab

Getting the daily return from a Pairs Trading Strategy using Matlab Guys, I am developing a pairs trading strategy in Matlab for my thesis. My problem is the following. I modelled the system in order to buy or sell just 1 quantity of a pair of stocks. Considering that I can short sell, possible negative initial investment are possible. How can I calculate the return in this case?   -- Prudent version: cash received from shorts is a liability and serves as collateral against the borrowed shares. The return calculation should divide any profit by these short proceeds. Necessarily the max return would be capped at 100%. Less prudent version: divide the return by only the margin required to support the short sale.   -- I think you can do it in the same way as done for futures. You short sell at 100, price rises to 102, you made a loss of 2%, similarly, if it falls to 98, you made a profit of 2%. In futures, since you don't actually own anything, based on your prediction, you either pay the 2% loss to exchange, or you receive the 2% profit from them.  

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

What is the best trading API for Matlab? MBtrading? IB?

View CommentsWritten on April 30th, 2012 by caustic
Categories: Matlab
What is the best trading API for Matlab? MBtrading? IB?   -- Why do you need a trading API for matlab?   --   -- Matlab typically has faster development turnarounds, and in some cases we have noticed the matrix math is so heavily optimized that it runs faster than C/C++. Also the GPU support for Matlab makes GPU development more rapid as well. Then like it was said there are many mathematical tools available: robust control, metaheuristics, custom plotting, etc...   -- I am sure you can use matlab's API to implement it's functions in your software. That should be more comfortable for trading I think.   -- It depends from speed required and some other stuff! For example i prefer to' develop something in matlab and then exporting it as a library (jar o what else). --   I have used Matlab's code gen feature to generate C/C++ code but noticed my pinv function ran slower than it did in Matlab. When I contact Mathworks they said it was slower because Matlab uses the heavily optimized LAPACK libraries which are not export to C/C++. So Ivan have you noticed any speed decreases in your jar files? I know there is an interactive broker API for Matlab, but I was hoping to find a better FOREX broker to auto trade. The big draw for my inside Matlab is the easy of GPU coding.   = if you are definitely looking for speed you need to have ultra fast modules for other things, like risk management, transport, monitoring and so on. And this would be quite difficult to do with matlab at ultra fast speeds. About what speeds are we talking?
 
Anyway, you can always write your own API to any broker or infrastructure provider   -- Just a quick warning. MATLABs martrix modules may be fast, but nothing else is. I haven't sorted out if you are using this for High Frequency trading or more traditional analysis and trade placement. If you are intending to use this for even moderate HFT I would stay away from the final implementation being in MATLAB, even exporting MATLAB modules and incorporating them in C++ or C# projects is a bad idea. A MATLAB interface could be convenient in an Analyse & Trade scenario, but MATLAB falls apart very quickly in any form of data feed and HFT scenario. It just doesn't handle streaming data very well at all. The memory management in MATLAB is poor and I have also seen lots of times it just stops reacting to incoming data for long periods. MATLAB is an interpreted layer over Java and sometimes C/C++. It was designed for rapidly implementing mathematical ideas and not deterministic on the fly data processing. MATLAB is outstanding for rapidly prototyping analysis ideas, but that is truly all it is good at. I would highly recommend that you use either C++ or C# for the live trading system work. This is from my last 10 years trying to do exactly what you are talking about. Just my 2 cents and I hope it helps. --   Poor guy: he wanted a suggestion about brokers and we are talking about everything else. I used matlab as Java library and i was satisfied about speed and memory. Maybe it depends from what kind of matlab tools are used: i used filters, clustering working with oo phylosophy. =--   maybe it is because not many who uses matlab for direct trading?)   -- I agree, if you're looking for ultre speed, Matlab is not what you need.   -- Recently I am trading with Matlab via API. My reasons to choose matlab is, first, easy development; second, easy math lib. There are many drawbacks, such as speed, single thread, --   so how does one easily treat tick by tick data in matlab. the fints only handles minute data. let's say you have micro second data or there any tools to help   -- Back to the broker question. IB will have better pricing, but they also force you to manually log in every night. That is tough for an automated trading system. Since they will disconnect you, if you are not there to log in your system is no longer connected. This may be okay for a US stock market program, but if you are trading any 24 hour market like currencies, it is not advisable.   == Interactive broker also provides IB Gateway, which keeps you connected all the time. This may resolve your concern. --   I am using Matlab in conjunction with Quant2IB as an interface to IB. This works well for trading on one minute intervals.  

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Anybody knows or has experience to connect Matlab with quickfix in c++ enviroment? Runs ok to recived data in real time?

View CommentsWritten on April 30th, 2012 by caustic
Categories: Matlab
Anybody knows or has experience to connect Matlab with quickfix in c++ enviroment? Runs ok to recived data in real time? == I have written a couple of similar applications (not for quickfix though). There is quite a high impedence mismatch between the operational models used by QuickFix (callbacks) and MATLAB (batch/vector processing), so you will probably need to implement a queue to sit in between QuickFix and MATLAB. A simple C++ quickfix client can be used to handle the callbacks and write messages to the queue. (Use Protocol Buffers maybe?) Then, depending on your use-case, you can either: 1. (PULL) Write a simple MATLAB mex extension to read large-ish batches of messages from the queue on demand, so your colleagues can very easily write MATlAB scripts that poll from the queue. 2. (PUSH) Write a small C++ program to read large-ish batches of messages from the queue, then pass them to a MATLAB script through an invocation of the MATLAB engine. In the past I have always gone for option (1) as it gives my colleagues a more familiar model when they write their MATLAB scripts to make use of the data. You cannot get around the fact that you need to trade-off latency for throughput. You also need to accept that you can neither get very low latency nor very high iteration rates from MATLAB (or any other interpreted scripting language for that matter); but by batching items together you can get reasonably high data rates.  

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Youtube algo strategy course with programming and new HFT Matlab cluster set up

Youtube algo strategy course with programming and new HFT Matlab cluster set up

http://quantlabs.net/quant-member-benefits/slash-your-quant-learning-curve/ YouTube Preview Image

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Youtube video demo of Quant membership with courses on algo, strategy development, HFT, QuantLib, C++, Matlab, C#, Java

Youtube video demo of Quant membership with courses on algo, strategy development, HFT, QuantLib, C++, Matlab, C#, Java http://quantlabs.net/membership.htm Quant membership benefitis here: http://quantlabs.net/quant-member-benefits/ YouTube Preview Image

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Youtube video demo of Quant membership with courses on algo, strategy development, HFT, QuantLib, C++, Matlab, C#, Java

Youtube video demo of Quant membership with courses on algo, strategy development, HFT, QuantLib, C++, Matlab, C#, Java http://quantlabs.net/membership.htm Quant membership benefitis here: http://quantlabs.net/quant-member-benefits/ YouTube Preview Image      

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Benefits of learning about quant, HFT, Quantlib, strategy development, algo, job hunting, Forex, All on Youtube of course

Benefits of learning about quant, HFT, Quantlib, strategy development, algo, job hunting, Forex, All on Youtube of course Join here: http://quantlabs.net/membership.htm Learn the quant benefits here: http://quantlabs.net/quant-member-benefits/ YouTube Preview Image  

Get our FREE Open Source Historical Database by answering the 2 WORLD'S FASTEST TRADER/QUANT QUESTIONS

Post to Twitter

Follow

Get every new post delivered to your Inbox

Join other followers: