[R-sig-Finance] portfolio package update and RFC on package names

David Kane dave at kanecap.com
Tue Jun 13 15:20:34 CEST 2006


Hi,

1) Version 0.2-1 of the portfolio package is available on CRAN. An
introductory article is available in the latest issue of R News. We
are thankful to many people for thoughtful comments, especially Dan
Rie,  Dirk Eddelbuettel, and Patrick Burns. As we noted in our initial
announcement:

https://stat.ethz.ch/pipermail/r-packages/2006/000171.html

we hope to recruit other developpers to this project. At this point,
our main concern is with the design of the class. Informed commentary
and suggestions (and contributions!) would be welcome.

2) We will be making signficant additions to the portfolio package in
the coming months, including extensive trading facilities and
improved graphics, including a Map of the Market plot.

3) We are working on two other related packages and would like to
request comments on their names. The first package, which we have
tentatively named "backtest", will provide --- you guessed it! --- a
basic backtest capability for financial applications. This will *not*
use the portfolio package. It will instead provide simple results like
decile spreads, Sharpe ratios, turnover and other measures of
historical performance for a quantitative trading signal.

RFC: Is "backtest" a good name for such a package?

4) The other package will *require* the portfolio package and will
provide a *much* more serious environment for backtesting. Some people
in finance distinguish between "backtests" --- meaning simple
approaches to seeing if a given signal would work without worrying too
much about trading rules, investibility and the like --- and
"simulations" --- meaning testing environments in which you do things
as much as possible as you would in the "real world.". 

An example of the difference between "backtests" and "simulations"
concerns funds under management. A typical "backtest" --- looking at
decile spreads for a price-to-book signal on a monthly basis for the
last decade --- does not explicitly worry about the amount of money
under management. It is implicitly assumed that you could trade into
the desired positions whether you are managing ten million or ten
billion dollars. A "simulation" would take position size seriously. It
might take days/weeks more to invest the ten billion than the ten
million, time during which you do not get the benefit of holding those
stocks. Our simulation package will take these and related issues
seriously.

RFC: "simulation" does not seem like a good name for this package
since the terms overlaps too much with other R communities. What would
be a good name?

Thanks,

Dave Kane



More information about the R-SIG-Finance mailing list