Muhammad wrote: >> âTraders Studio allow programmers to make there own codes to run on portfolios and calculate any thing they want. None of the currently available softwares provide such power.â Iâm a programmer and a user of TS2000i and TraderStudio doesnât meet my requirements. I couldnât find any description of features or performance characteristics on their web page. They just have tutorials. I donât want to watch, âclick thisâ or âpoint your mouse hereâ and try to deduce what I would be buying. I want to know what their optimization engine is, what objective functions are available, what built-in indicators are available, what there new hybrid language looks like, how fast is the optimization engine? Etc. As a programmer it was easy to write my own platform for portfolio testing and I was confident that it was going to what I wanted because of the software development process I went through to develop it: Requirements analysis, Design, Development, Integration and Test. For example my requirement was to test with intraday and tick data. The implementation details for this are trivial if that requirement was there at the beginning. If I were developing a product I would make a major effort to understand the needs of my customers. If one looks at the total number of users backtesting only EOD verse intraday, EOD only is in the minority. A larger customer base uses intraday data. And as I said intraday testing is such a trivial implementation detail if you specify and design for it from the beginning. The difference is that a bar of EOD data has only a date field while intraday has date and time. Adding that field after the fact as to be rippled though all of design and code. One would have to change everything that touches a bar of data from the data feed interface, to the database, to the charting package to the backtesting engine, etc. Another import user requirement is speed. TS2000i code is interpreted and some traders have strategies that take hours to optimize. Imagine doing that for a portfolio of 25 stocks. Interpreted code runs at least 10 times slower than compiled code. So my platform had a performance requirement for speed. I wanted to optimize, test and data mine large portfolios and I knew that I had to have speed to do that. Initially a portfolio the size of the NDX, but my requirement had no limitations on the size of the portfolio. The obvious design choice was for compiled code not interpreted code like in TS2000i. Its interesting Iâve only come across one commercial backtesting product that compiles strategies â someone had done their homework.
Excellent reply Wolf! Thanks for the insight. What is the name of commercial product you mentioned? Smartquant perhaps?
Don't let the negative comments get to you, first none of the people who made them own the product. They have there own issues about why they don't want to see this software succeed. We have a special offer with my book, Trading in Technology which offers a 60 day money back guarantee for 499.00. Here is the link. http://tradersstudio.com/Default.as...alogID=5&psnavcmd=CatalogItemDetails&tabid=38 We have live chat for tech support, a help desk as well as E-mail. We also have our own forums. I am sure you will be happy with TradersStudio. If you have any question , you can private PM me at EliteTrader. In answering other peopleâs comments. We have our language help up on the site . You need to be registered and logged in to view them. http://tradersstudio.com/Default.aspx?tabid=58 Our reports are as complete as TradeStationâs, we have very similar reports, Summary, Trade by Trade,Annual breakdown,monthly breakdown, active orders, Equity Curve chart. Start Trade Drawdown, Underwater equity and more. We also have a walk forward optimizer macro built in. I wish people would not make comments about TradersStudio without owning the product. In terms of real time development, yes it easier if you do it correctly in design, I know we are doing it right. The big problem is supporting all of the data feeds that a big issue, since they change their API, that was an issue I needed to address In terms of speed for 90% of users it will not be an issue. We are working on version 2.0 EOD that will be a free upgrade to existing users. This is still a script language but it about 2-3 times faster than 1.3.6. The point these other people are forgetting is that we can optimize on a portfolio, you can get the best parameters across the basket. In addition we have a compiler under development as an addin for version 2.0. We have done our homework. Comments on this site on how easy it is to develop a trading platform I see up here sometime are funny. If you want to develop a reliable platform where you can trust the results of commercial quality, it will require 15-20 man years to develop a competitive platform, at 100,000 per man year with overhead that as much as two million dollars.
I fully agree. Optimization speed is key factor. What I can suggest is to give AmiBroker a try. You would be surprised how fast it is. In addition to very quick exhaustive search optimization, there is a free "intelligent" optimizer script, written by one of AB users, that will perform "particle swarm" optimization of billions of parameter combinations within minutes, not hours. http://finance.groups.yahoo.com/group/amibroker-ts/files/IO.zip You can use free trial to evaluate both. Regards, Tomasz Janeczko amibroker.com
Murray Ruggiero: It took only a few man-months (my man-months) to develop my platform. And most of that time was spent in requirement analysis, data modeling and object oriented design. Development was quick because I used Delphi for rapid prototype development and I had used as many existing components that satisfied the requirements that I could find. Some I had lying around (GA, NN, etc.) and others were commercial components (charting, database, etc) I had bought. Software reuse is a key factor in low cost software development â why reinvent the wheel? Speaking of reinventing the wheel, why write a new compiler? Thatâs a big costly effort. Is it because you arenât using a standard language and have to start from scratch? My requirement was to use a standard language. I use Delphi to compile my strategies. Its language is Pascal which is very similar to TS2000i's EasyLanguage. If I were developing a commercial product I would have looked at licensing an existing commercial compiler and putting a wrapper around it. The unit cost for such a license has to be far lower than developing a new one (not to mention the added life-cycle maintenance cost). And speaking of realtime data feed, you can use ActiveX to interface to eSignal. The development effort would be man-days or weeks at the most. I donât know that would be the âbestâ method, but eSignalâs ActiveX interface to TS2000i is very fast. So if I had a commercial product that was already designed for realtime I would have initially come out with a RT version whose only data feed was eSignal and incrementally added others later.
We are talking about apples and watermelons. A platform which takes only a few months to develop , can be very complete , well tested and flexible. A good charting module requires , 3-6 months work as a starting point . A real time data server ,with a high speed data access to save historical data for later use is a large complex project and again even for a brilliant program 6-12 months of work including very complete testing. Our TradersStudio language as at least 2.5 man years invested in it. It was written from scratch. My point was to do it right , not to just put something for you to use together, if you are a programmer. One point you are forgetting is that if you read some of the components license agreements , you can't develop commercial trading product with them, without paying royalties or in some cases not at all. You need to read the agreement very closely. Another point is that we wanted to build a commercial platform, which was our business so we wanted to own the code. If you are playing for yourself, buying off the shelf tools might be ok. Please don't confuse the too. I am sure that the developer of AMIBroker would agree developing a commercial backtesting platform is not a few months work. A toy to play with maybe but not a real product.
Re: âA toy to play with maybe but not a real product.â My platform is not a toy; its capabilities far exceed your product â by light-years. Thereâs a reason that agile model driven software development and rapid application development are popular methodologies in the commercial software development world. There are both life-cycle cost and reliability reasons for this. These are not toy methodologies of home developers but are used mainly by large commercial organizations for industrial strength mission critical applications. If you have some spare time you may want to research model driven software development and rapid application development. The big boys donât think of these approaches as toys.
From your posting I can see you know nothing about my product. Do not make negative comments about it , since you do not know what you are talking about relative to my software.