From Qaunt Premium Member, the truth of open source vs money sucking proprietary software like Tradestation or Multicharts
This is a series of queries for:
Big announcement as I move back to Matlab for my quant model and trading strategies instead of R for my custom HFT
Queries are below followed by –>
Once you have the C++ translations for analysis modules, can you use Scilab and/or Octave? Can Scilab generate translated modules? I believe particularly Scilab has a lot of recent effort put into it so it can integrate Matlab code and toolkits.
—> The end system is all done in C++ where the Matlab M script code is converted thanks to the Matlab Coder toolbox. I will drop that C++ into my HFT platform.
The rest of this email, is just FYI, some background to inform you about my particular personal interests and experience.
I am not at all sure if there is any good way to integrate with R though, even though I am aware that many traders’ analysts are rallying around R, whether the traders wear both analysis and trader hats at the same time, or are working as teams dividing the specialities among partners. I am still in Stone Age struggling with TradeStation for historical data and Multicharts for “dot net integration” to be able to use the legacy Easylanguage file writing of compressed data sitting around in various traders’ old backup TradeStation caches becoming no longer readable with new TS versions or Windows operating systems. But new traders can save tens of thousands of dollars sharing those old caches between others who just happened to save them. Many of these historical tick data can be found in forums on the internet already.
–> Tradestation is propietary which I am not a fan of. I like to work with things that are not so black box. R is fine but it does not quite live up to tick by tick analysis in real time. You can easily generate your own tick files thanks to IQFeed.net. http://www.youtube.com/watch?v=Ry7igGVN1uQ&feature=plcp
So why spends of thousands of dollars?
Isn’t it suspicious that the proprietary trading platform developers and sellers do not willingly encourage or even allow even simple ASCIII file READING of the same files they are easily capable of writing???
–> See that Youtube video above to have your mind blown.
The database work you are doing is fascinating, but the trader I am working with recently in New York by Teamview still has faith in the stored up caches of TradeStation tick data sitting around which I think will soon be almost impossible to gather, maintain, and read if it is not translated and recorded in a less proprietary format. I still am trying to find more time to study your data base proposals soon.
–> There is no proprietary technology to the above video. It is live tick by tick or a any frequency historical data you can generate on your free will thanks to IQFeed
But I still have to spend time babysitting TradeStation, Multicharts, and Interactive Brokers accounts while I try to learn the more flexible escape routes from proprietary platforms without losing use of proprietary software I accumulated over many years.
–> Wow! Unfortunately you are getting left behind my friend. Programming will be a critical in the ‘next generation of trading’. Most of what I do is based on FREE open source technology but the only thing holding you back is learning the languages.
In terms of what kinds of data analysis I am interested in, it is mostly signal processing, as my background is electrical engineering, and I have for many years studied filtering, neural networks, audio technology which involves a lot of filtering.
–> Most strategies I am working on generate those signals but this video will change your views on how to trade in the next generation of software
Pretty cool huh?
I particularly like the java coded constraint-based filter design work in differential evolution by Rainer Storn, ( I own source code license) the Singular Spectrum Analysis work by N. Golyandina et al. of Gistat Group, the Biocomp software by Carl Cook, and the neural network tools by the late Steve Ward for all of which I have long ago paid up licenses. Most of these are trading-oriented filtering and mapping tools useful for time series decomposition , compression, and reconstruction. Some I have seen no substitute yet in open source, others are just becoming available in R, Octave and Scilab. Of course, Matlab has been way aheadfor a long time, but I was hoping to escape the large matlab fees which I now am too “small” to afford…..
–> Sure I hear you on the costs of Matlab but fortunately for my Premium Membership will make these C++ generated strategies available for my members. People in the future will be paying thousands for these. Once I get past the development of my HFT platform, it will open this world very quickly. Remember I can use R but the previous video will show the bottlenecks with R which meant I had to resort back to Matlab for the Coder toolbox.
These tools will be quite useful for pre-processing historical data into a form useful for saving a lot of time now wasted in re-processing history redundantly in the proprietary platforms. Better ( IMO, personally) to have available a “library” available for instant look-ups to determine what state a market is most probably in so intelligent choice of current trading strategy does not have to always be repeated from scratch analytics. State change time and price-action intervals with useful repetitive persistence attributes are visible using these tools ( in retrospect) down to the minute, or even single tick in history for quick easy reference. This will be valuable time, equipment, and screen real estate saver when trying to pin down and classify causal evidence of state changes later. Many of these are quite simple in nature, thus useful for simple strategies, such as “what are typical retrace ranges of this instrument ? , and how longare they likely to persist, and why, because what intrinsic and extrinsic events are probably the state change correlated causes?” Being able to quickly relate quantifiable events outside of price data observation alone, especially scheduled events, to quantifiable persistent price activity behavior and its periodic change events is one valuable and essential approach in the toolbox. Isn’t that why traders call people quantifying, classifying, and identifying market state changes related to “events”……. “quants”?
–> Many people can use whatever type of analysis but the pro toolboxes from Matlab enable me to bang out stuff pretty quickly or at least test it within Matlab
MR X (member trying to find more time to follow your work, and escape from proprietary software platforms)
Thanks for your invitation on Linked In! Am looking forward to correspondence more in future, and your opinions about my own points of view.
–> No worries as things will get more exciting after the HFT platform proves itself. Thanks for being a Member, it is guys like you that enable me to continue what I do.
So JOIN NOW so you don’t get left behind.
Memebership benefits listed
Or join my free newsletter for more info
I now post my TRADING ALERTS
into my personal FACEBOOK ACCOUNT
. Don't worry as I don't post stupid cat videos or what I eat!