riskquantify-devel Mailing List for Risk Quantify Financial Engine

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Risk Quantify has a new look! Check it out at http://riskquantify.net/.

Cheers, Brett
--
Brett Hutley - Quantitative Developer
mailto:br...@hu...
http://hutley.net/brett

I've release version 0.7.6 of Risk Quantify. This version has the  
following improvements:

* Added a rq_tree_rb_remove() function so that we can remove nodes
from the red-black tree. Also it doesn't seem as though an in-order
insertion of keys into the tree would actually produce a balanced
tree. I've fixed this up.

* Added a new test_tree_rb test for testing red-black trees.

* Added generic iterator support to most of the container classes.

* Added some matrix functions.

* Added a poisson random number function.

* Added a multi-factor monte carlo pricing function.

As usual the latest version can be downloaded here:

http://sourceforge.net/project/showfiles.php?group_id=10419

Please let me know if you find any problems with this release.

Cheers, Brett

--
Brett Hutley - Quantitative Developer
mailto:br...@hu...
http://hutley.net/brett

I added support to Risk Quantify to remove nodes from red-black =20
trees. I also found that an in-order insertion into the red-black =20
tree didn=E2=80=99t actually rebalance the tree, so I=E2=80=99ve fixed =
this bug as =20
well. This may mean that existing code will have better performance - =20=

especially if the data being loaded into the tree is already ordered, =20=

and there are lots of searches. I=E2=80=99ve added a rq_tree_rb_size() =20=

function to get a count of the number of nodes in the tree.

I=E2=80=99ve created a file to test the new red-black tree functions, =
called =20
test_tree_rb.c which is in the test/code directory. If you invoke it =20
with an argument, you can get a prompt allowing you to test the =20
functionality interactively. The test file is designed to fill the =20
tree with integers, so if you are using it interactively; type in =20
something like =E2=80=9Ca 1=E2=80=B3 to add a node with the key =E2=80=9C1=
=E2=80=B3, =E2=80=9Ca 2=E2=80=B3 =20
will add a node with the key =E2=80=9C2=E2=80=B3, etc. =E2=80=9Cp=E2=80=9D=
 will print the =20
current state of the tree. =E2=80=9Cr 2=E2=80=B3 will remove the node =
with key 2 =20
from the tree.

Please let me know if you find any problems with the implementation.

--
Brett Hutley - Quantitative Developer
mailto:br...@hu...
http://hutley.net/brett

Happy new year, everyone.
I've added generic iterator support to most of the container classes  
(specifically rq_tree_rb, rq_linked_list, rq_double_linked_list and  
rq_array). I still have to do rq_vector and rq_array_partitioned.  
rq_array_double will probably be removed in favour of rq_vector. The  
old iterator interfaces are still there, but everyone should now use  
the new generic interface functionality, as it means that generic  
algorithms can be created that take rq_iterator objects as parameters  
and it doesn't matter what type of container you have underneath.

The call to get a generic iterator is rq_xxx_get_iterator() - so, for  
example, to get a red-black tree iterator you would do something like:

rq_iterator_t iter = rq_tree_rb_get_iterator(tree);

Cheers, Brett

--
Brett Hutley
mailto:br...@hu...
http://hutley.net/brett

Hot on the heels of version 0.7.4 - version 0.7.5 has been released! 
Here are details of what has changed in this release (from the NEWS 
file):

Version 0.7.5 (2005-04-05)

* Yield curves when interpolating or extrapolating rates using
   zero-coupon rates, now have a default day count convention, which is
   usually set to either ACT/365 (default) or ACT/360. The simple yield
   curve bootstrapper and composite yield curve bootstrapper set this
   value based on the days_per_year value of the underlying currency on
   the yield curve.

   Additive and Multiplicative factors are now applied to rates
   calculated using the appropriate day count fraction.

* There is also a new helper function rq_termstruct_get_type() to
   determine what kind of term structure one is dealing with.

* There are some helper functions in rq_rate_conversions.c to go from
   a discount factor to an annual compounding zero-coupon rate and back
   again.

Cheers, B
--
Brett Hutley {MAppFin,CISSP,SANS GCIH, SANS GCIA}
mailto:br...@hu...
http://hutley.net/brett

I just release version 0.7.4 of RQ on SourceForge. The source can be 
downloaded from here:

http://sourceforge.net/project/showfiles.php?group_id=10419

The main change in this release is that the payment amount on the fixed 
flows in the rq_floating_flow structure are now calculated by dividing 
the fixed rate by the number of periods per year, and then multiplying 
by the notional amount. This replaces the old method which was to use 
the day count convention and the period itself to work out the 
effective rate. This (I believe) emulates what occurs in the market. 
The code will fall back to using the day count convention and the 
period to calculate the payment amount, if it can't work out the number 
of periods per year from the start and end dates.

Let me know if you think I am wrong with this.

Cheers, Brett

--
Brett Hutley {MAppFin,CISSP,SANS GCIH, SANS GCIA}
mailto:br...@hu...
http://hutley.net/brett

Version 0.7.2 of Risk Quantify has been released. There has been some 
work done on the yield curves in this release, and there is an MSVC 
project file for building a command-line tool - "rq_yc.exe" - for 
bootstrapping yield curves.

There is a sample configuration file rq_yc.cfg in the 
"src/apps/cmdline" directory.

Of course, you can build this program in the Unix/Linux/Mac OS/X 
environments as well.

A sample command line for running the program is:

$ ./rq_yc -c rq_yc.cfg -C "YIELD_CURVE:AUD BBSW" -D 2005-01-01 -L
2005-01-02      0.999863032461  0.050000000000  0.050000000000
2005-02-01      0.995687174185  0.051000000000  0.051000000000
2005-04-01      0.987340402510  0.052000000000  0.051287671233
2005-07-01      0.974390872227  0.053000000000  0.052217402257

Here it is outputting the points on the bootstrapped curve "AUD BBSW" - 
the date, discount factor, zero rate, and quarterly swap rate are 
output.

I am still testing this program, but would definitely appreciate any 
comments/feedback/bug-fixes.

The tarball can be downloaded from sourceforge:
http://sourceforge.net/project/showfiles.php?group_id=10419

Cheers, B
--
Brett Hutley {MAppFin,CISSP,SANS GCIH, SANS GCIA}
mailto:br...@hu...
http://hutley.net/brett

OK, I've released the new version of RQ on SourceForge. Here is a 
summary of the latest changes from the NEWS file:

Version 0.7.1 (2005-03-16)

* Lots of work done on term structures, floating flows, day count
   conventions, roll conventions, and calendars!

   There was a bug fix done for calculating the number of months
   between dates in rq_date_month_diff.

   I've added a new type - rq_floating_flow - for storing the
   information needed to value asset-flow-based trades such as swaps or
   bonds.

   I've added functions for calculating day count fractions according
   to the different conventions (in rq_day_count).

   I've added functions for calculating the different roll dates
   according to the different conventions.

   I've also changed the roll date calculation functions so they can
   take a list of calendars instead of only one.

   I've extended rq_yield_curve so it can support being a composite
   yield curve on two other yield curves.

   I've *started* to extend rq_calendar to support composite calendars.

   I've added the weekend dates to the rq_calendar object, in order to
   support calendars that DON'T have Saturday and Sunday as weekends.

   I've fixed up one bug in the rq_tree_rb_clone() function.

   I've made the first part of any term structure object
   (rq_yield_curve, rq_forward_curve, etc) be a standard rq_termstruct
   type. This is so there can be standard routines to determine what
   KIND of term structure a generic termstructure object is.

As usual, any help is much appreciated. Please let me know of any 
issues you find with the release.

Cheers, Brett

--
Brett Hutley {MAppFin,CISSP,SANS GCIH, SANS GCIA}
mailto:br...@hu...
http://hutley.net/brett

OK, I'm trying to put together some objectives for future RQ 
development.

Here is some stuff I've got down so far:

1. metatype building of RQ objects: I want to change RQ so that the C 
code for the objects in the system are constructed from XML schema 
files. This will mean that every object will support a standard set of 
functionality including loading/saving from/to XML and loading/saving 
from/to ODBC (UnixODBC on Unix, Linux and Mac OS/X).

2. More pricing functions: I want to add more pricing functions, 
especially beefing out the numerical support.

3. Enhanced yield curve support: More algorithms for modeling term 
structures in various ways. Eg Hull and White. Calibration of models.

4. Redo FpML support via metatype framework.

5. Start building front end to demonstrate and exercise the Risk 
Quantify library.

6. Get Excel Add-ins and Mathematica support building out-of-the-box. 
Include the Excel Add-in in a binary release.

7. Better developer documentation!!!!

Some general notes:

I want to use expat (http://expat.sourceforge.net/) as the XML parser 
in RQ. The project will be modified so that this will become a 
dependency (much as I hate having dependencies in the project).

I might need to have a development code stream as well as a "stable" 
code stream so that existing users aren't impacted by large changes in 
the code base. I'm not quite sure how to manage this, so would 
appreciate any input.

Cheers, Brett

--
Brett Hutley {MAppFin,CISSP,SANS GCIH}
mailto:br...@hu...
http://hutley.net/brett

I want to put a list of objectives together for RQ ver 0.8. If anyone 
has anything they want to add to the list, please send me an email.

Cheers, Brett
--
Brett Hutley {MAppFin,CISSP,SANS GCIH}
mailto:br...@hu...
http://hutley.net/brett

I've updated the include path to the files in the rq and rq/fpml
directory.
Originally my idea was that with a unix installation, the header files
would be copied into /usr/include/rq and /usr/include/rq/fpml, etc and
the library would be installed into /usr/lib. People could then put
#include <rq/rq.h> in their source files and drag in the risk quantify
includes.
Unfortunately this doesn't work so well if you don't have this layout.
Eg under Windows, it means to include stuff, I have to set the include
path to be the directory *underneath* the rq directory, or in my case
the src/riskquantify/src directory, which is pretty weird.
I've also updated the MSVC project. I think the unix Makefiles are now
out of date. I'll fix that up once I get back on unix.

Cheers, Brett

2002	Jan	Feb	Mar	Apr	May	Jun	Jul (1)	Aug	Sep	Oct	Nov	Dec
2005	Jan (2)	Feb	Mar (2)	Apr (2)	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec (1)
2006	Jan (3)	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec

riskquantify-devel Mailing List for Risk Quantify Financial Engine

riskquantify-devel — Risk Quantify Developers Mailing List