[R-SIG-Finance] Do you have any suggestions for spreading R in Taiwan where finance is dominated by Matlab?

[Brian G. Peterson]

On 12/08/2010 08:16 PM, JOSH CHIEN wrote:
> Dear all R user,
>
> I'm a risk specialist, quant team, for investment risk department in life
> insurance company in Taiwan.
>
> I need your guys to help me how to spread using R in finance.
>
> Any feedback and opinion is fine to me.
>
> I just want to inspire some idea about using R in finance,especially in modeling
> Credit Risk&  Market Risk.

1> transparency

As an open system, you can look at, evaluate, and modify the code.
All of it.  There are many well documented failings of Excel, SAS, SPSS, 
and Matlab statistical methods and financial toolkits.  With those 
systems, you are at the mercy of the commercial vendor to fix issues the 
effect your work.  I know from personal experience that even after 
spending tens or hundreds of thousands of dollars (or equivalent) on 
commercial licenses, the answer from your vendor is likely to be "thanks 
for the bug report, we'll look into it and add it to a future release". 
  R is of course not perfect.  Many of the packages developed for R were 
developed for teaching, not production, for example.  *All* code has 
bugs.  However, you will have the code, and could fix any problems you 
locate, with or without help from the package creator/maintainer.

2> community

The mailing lists, forums, etc for R are in many cases more active than 
those for commercial packages.  There are some serious heavy-hitters on 
this list alone.  If you are careful in asking your question (see Eric 
Raymond's excellent advice on asking smart questions) and are working on 
an interesting problem, chances are very good that you'll get a helpful, 
insightful answer.  You'll still need to do some work, of course, but 
all things of value require effort.  Conferences like the RMetrics 
conference and R/Finance in Chicago bring together people exclusively at 
the intersection of finance and R to network and present interesting 
research.

3> collaboration

Related to 2> above, but different.  If you can judge the things that 
you work on that are non-proprietary, of general utility, and contribute 
those things to the community via packages or significant code examples 
on this list or a blog, others will put out real effort to work on them 
with you.  You, in return, enhance your own professional reputation, get 
contributions from gifted individuals, free bug testing, and feedback.

4> research

Many PhD programs, and an increasing number of Masters programs in 
economics or finance are moving to the use of R.  R seems to be the 
dominant language in statistics, and is gaining ground from Matlab in 
Finance.  This increases the chance that what you need (for example in 
Credit and Market Risk) is already written in R, leaving you to spend 
more of your time on your specific problem, and less time replicating 
some analysis published in a journal or book.

5> scalability

I have run analyses on 50 or more physical nodes in parallel.  For the 
cost of commercial software, I could instead get more physical nodes 
(lots more nodes!), and therefore be more productive.


There are more reasons, but these are the most important ones to me.  No 
system is perfect, and no system does everything you want just the way 
you want, but despite (or perhaps because of) its steep learning curve, 
R is capable of getting closer than any other system I'm aware of (and 
I've used all the ones I mention above in large scale analyses).


Regards,

    - Brian

-- 
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock