Testing technical indicators
On 6/1/06, Jeff Ryan <jeff.a.ryan at gmail.com> wrote:
I too have been looking for a more elegant solution than my current perl/R approach. Jab - my current (yet uncoded) thinking has been to migrate everything except the raw data manipulation to R. I'd be interested to hear why you use python for some of your systems and/or test(s). The only thing keeping me back is from a signal generation standpoint - seems to be easier to leave that outside of R.
Several factors lie behind that decision, but the most important is
that it is the easiest/fastest way for me to do things. I agree that
everything could be done in R, but it would take a huge amount of
effort for _me_ to get there. I did start on an R package for all the
important technical indicators and some classic systems, but my
progress was too slow to allow me to do anything important in the time
I had available. For example I ran into problems involving the details
of S3 and S4 classes which seemed very hard to solve and the package
its, which I relied on, was orphaned. As I already had these things
coded, debugged and working in Python, I just didn't have the
time/will to recode everything. The bottom line is I can code systems
very quickly in Python and relying on R for analytics rather than
coding the stats myself makes perfect sense to me.
jab
John Bollinger, CFA, CMT www.BollingerBands.com If you advance far enough, you arrive at the beginning.