[ts-gen] Re: new user

[R P Herrold]

On Tue, 17 Apr 2007, Andrew McClure wrote:

> Excellent, I never knew of this web site.  I'd read 
> Education of a Speculator, and Practical Speculation many 
> years ago, and being an avid checkers and table tennis 
> player, loved them from the start. Will be watching this 
> forum with interest!

> No, I came across shim some months ago from doing a google 
> on MySQL and IB - I figured someone must be further down the 
> road than I was, and sure enough ...  Will take the covers 
> off it today, and hopefully get on my way.

> Have you any thoughts about using MySQL to plug in to a 
> charting pack - i.e. perhaps write an interface into an 
> existing backtesting/charting application? I always 
> considered that MySQL (or any RDBMS?) just wouldn't be able 
> to cut it in real time.

Hooks into R, and gnuplot, from a MySQL source are 
straighforward -- scripting glue is easy enough -- I like php, 
and Bill, Ruby, each of which are trivial to hook into the 
backend datastore.  Obviously the perl-DBI is out there as 
well (genius trader);  We are studying retrospectively, rather 
forecasting predictively, as our lead focus of inquiry, and as 
noted, particularly on the Other Voices part of the website, 
we have added outlinks to several promising and usually FOSS 
approaches.  One thing I note is that most authors' packages 
are not well set up for sub-day visualization, backtesting, or 
charting.

As to the speed of MySQL, it is surprisingly fast -- I have 
hit a burst rate in a call center application I wrote a few 
years ago, in excess of 1000 TPS, in a 90/10 R/W mix in the 
past; subsetting and pre-selection of datasets, and cacheing 
which the MySQL engines do help as well.  Tonnes and tonnes of 
ram also help, and SMP, so that the entire dataset can migrate 
into ram on the MySQL server.

> Are you using TA Lib perhaps for technical indicators?  Are 
> you familiar the work of Jeff Katz and C-Trader? 
> Personally, I quite like the TS2000i environment, but 
> getting data into it is a perennial chore, and of course 
> it's not supported anymore.

I made a decision to move to an all *nix with a very strong 
FOSS bias back a decade ago -- there are a couple of Windows 
boxes around for the IB WebEx content, and for MICR check 
printing, but that's about it; If the API to use a tool is not 
freely available, we are not likely to code to it.  The Other 
Voices part of the trading-shim site have ta-lib; genius 
trader is also known to me; The Katz book is known to exist by 
me, but I have not been pursuaded by the Amazon reviews to buy 
it:

 	Katz, Jeffrey Owen,Ph.D., McCormick, Donna L., _The
 	Encyclopedia of Trading Strategies_, 2000. (ISBN 0-07-058099-5)

Similarly the Stridsman book is out there ...

 	Trading Systems That Work: Building and Evaluating Effective
 	Trading Systems (Hardcover) by Thomas Stridsman, McGraw-Hill;
 	1 edition (November 20, 2000), 007135980X

but I have not exhausted other resources I want to complete 
examination of enough to go seeking new ideas.

There is no dearth of indicators -- indeed far more exist than 
one can ever sensibly test.  The Aronson book, Evidence-Based 
Technical Analysis: Applying the Scientific Method and 
Statistical Inference to Trading Signals, Wiley (November 3, 
2006)  0470008741 struck me as more useful for the interests I 
have, as I mentioned on the Daily Speculations piece;  Prof 
Aronson was kind enough to send along a review copy, and I 
think highly of it.

> You may be interested to learn that Saxo bank are developing 
> an API for their Saxo trader product - definitely worth 
> checking out for range of instruments if nothing else. 
> Anyway, no more procrastinating - looking forward to that 
> manual.

I have a markup draft home with me tonight, and Bill and I 
discussed ways to make the rater extensive detail in the 
website more accessible, with a Sitemap, and precis for the 
various pages.  I was familiar with the product:

   http://saxotrader.saxobank.com/

and had discarded consideration of it in that it lacked an API 
I could develop sensibly to, when I did my parket survey a 
couple years ago.  A FIX approach is the more likely next path 
if a major extension is needed, along with alternative 
datastream and historical data retrieval interfaces.

-- Russ Herrold