[R-SIG-Finance] Curve fitting the South African yield curve

[Thomas Browne]

Thanks very much for this. And what would you suggest I use to obtain  
the zero curve from the coupon curve (for the inputs to Nelson Siegel)?

South African bonds are quoated semi-annual Act/365.



On 29 May 2011, at 03:18, Joshua Ulrich wrote:

> Hi Tom,
>
> On Sat, May 28, 2011 at 4:43 AM, Thomas Browne  
> <thomas.browne at mac.com> wrote:
>> Hi,
>>
>> I am looking to do historic cheap/dear analysis on the South African
>> bond and IRS yield curves. What I'd like to do is fit a curve each  
>> day
>> and see what the historic deviation from that fitted curve each bond
>> has exhibited. The curve I guess should not be a cubic spline type
>> since that would go through all the bonds). I want a fairly simple
>> model because to start with and I've heard of Nelson-Siegel. How  
>> would
>> I go about doing this using R? Is there a library which could help?
>>
> A search for "Nelson Siegel" on www.rseek.org returned 3 packages on
> the first page.
> 1) YieldCurve
> 2) fBonds
> 3) termstruc
>
> I've used YieldCurve.  I've never used fBonds.  I looked at termstruc
> but I needed to modify some of the functions and it was easier to
> understand the YieldCurve code (it's very simple and straight
> forward).
>
>> As I'm just starting out I don't really want to have to bootstrap the
>> zero curve first (too much work! Unless there's help in R for that
>> too!). I'd just like to fit a fairly simple curve which makes sense  
>> to
>> the YTM versus Mduration (or PV01) scatterplot.
>>
>> Thanks for any help.
>>
>> Tom
>>
>
> Hope that helps,
> --
> Joshua Ulrich  |  FOSS Trading: www.fosstrading.com