Re: [Quantlib-dev] Negative price with Forward Rates

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

thanks, yes I agree, this is more a design question than of real
practical interest. The log-linear interpolation in zeroyield is very
briefly discussed in the Hagan / West paper on yield curve
construction, playing in the same league as linear in zeroyield (which
is yet very popular I think) or linear in discount. Log-linear in
forwards is not even mentioned there, so probably we should not make
that work in QuantLib either :-)

On 1 February 2014 16:47, Luigi Ballabio <lui...@gm...> wrote:
> Yes, possibly. In that case, we'd have to make the change in
> InterpolatedForwardCurve<T>::zeroYieldImpl. But since we don't have
> different exception types, we'd have to catch the exception and check
> the error message it contains to decide whether to switch to the
> base-class method (because if it's a different error, we'd want to let
> it bubble up to client code). It wouldn't look very pretty :)
> Another alternative would be to implement the numerical integration in
> the LogLinear interpolation. I'm not sure that I like a situation in
> which some curves are reasonably fast and others are much slower
> depending on the interpolation method. Sure, the current situation is
> that some interpolations don't work, so it might be an improvement
> anyway...
>
> (Oh, and a final note: while it makes sense to use log-linear with
> discounts, because they're given by exp(-rt), I'm not sure that it
> makes sense to use it with zero and forward rates.)
>
> Luigi
>
>
>
> On Tue, Jan 28, 2014 at 1:21 PM, Peter Caspers <pca...@gm...> wrote:
>> Luigi, just out of curiosity, in ForwardRateStructure there seems to
>> be a general purpose implementation to get the zero yield by
>> integration over the forwards. Wouldn't the situation here be a use
>> case for exactly this (i.e. catch the exception and invoke the base
>> class method instead) ?
>> best, Peter
>>
>> On 28 January 2014 12:37, Luigi Ballabio <lui...@gm...> wrote:
>>> Yes. the curve is trying to integrate the forwards to get the
>>> discounts, but the LogLinear interpolator doesn't provide a formula
>>> for that. You'll have to choose another interpolation.
>>>
>>> Luigi
>>>
>>> On Mon, Jan 27, 2014 at 5:50 AM, varun yadav <var...@ya...> wrote:
>>>> Hi Luigi,
>>>>
>>>> I passed evaluation date as the first date in liborDates. Also added
>>>> corresponding rate in liborRates. But now it gives exception -
>>>> "LogInterpolation primitive not implemented"
>>>>
>>>> Below are new values -
>>>>
>>>> liborDates -
>>>>
>>>> January 24th, 2014
>>>> July 24th, 2014
>>>> January 24th, 2015
>>>> January 24th, 2016
>>>> January 24th, 2017
>>>> January 24th, 2018
>>>> January 24th, 2019
>>>> January 24th, 2020
>>>> January 24th, 2021
>>>> January 24th, 2022
>>>> January 24th, 2023
>>>> January 24th, 2024
>>>> January 24th, 2029
>>>>
>>>> liborRates -
>>>>
>>>> 8.4
>>>> 8.4955
>>>> 8.36907
>>>> 8.38416
>>>> 8.47267
>>>> 8.62786
>>>> 8.69832
>>>> 8.71107
>>>> 8.85289
>>>> 8.99607
>>>> 8.80085
>>>> 8.7099
>>>> 9.00594
>>>>
>>>> Is exception coming due to InterpolatedForwardCurve<LogLinear> ...?
>>>>
>>>> Thanks in advance!
>>>>
>>>>
>>>> On Thursday, January 23, 2014 5:02 PM, Luigi Ballabio
>>>> <lui...@gm...> wrote:
>>>> You should pass the evaluation date as the first date in liborDates.
>>>> You'll need a corresponding extra element in liborRates; just
>>>> replicate the first you have for constant rates from today to the
>>>> first actual Libor date.
>>>>
>>>> Luigi
>>>>
>>>> On Mon, Jan 20, 2014 at 8:12 AM, v17 <var...@ya...> wrote:
>>>>> Hi Everyone,
>>>>>
>>>>> I have following piece of code which gives Exception "negative time (-0.5)
>>>>> given"
>>>>>
>>>>> vector <Real> &forwardRates;
>>>>> const int size = liborRates.size();
>>>>> vector<Date> liborDates(liborTenors.size());
>>>>> for (int i = 0; i < size; i++)
>>>>> {
>>>>>        curDate = ycDef.cal.advance(ycDef.evalDate, liborTenors[i]);
>>>>>        liborDates[i] = curDate;
>>>>> }
>>>>> yieldCurve = shared_ptr<YieldTermStructure>(new
>>>>> InterpolatedForwardCurve<LogLinear>(liborDates, liborRates, ycDef.dc,
>>>>> ycDef.cal));
>>>>> yieldCurve->enableExtrapolation();
>>>>>
>>>>> size = liborTenors.size();
>>>>> forwardRates.clear();
>>>>> forwardRates.resize(size);
>>>>> try
>>>>> {
>>>>> forwardRates[i] = yieldCurve->forwardRate(ycDef.evalDate, ycDef.evalDate +
>>>>> liborTenors[i] , ycDef.dc, Compounded, Annual);
>>>>> }
>>>>> catch(QuantLib::Error& e)
>>>>> {
>>>>> cout << e.what() << endl;
>>>>> throw;
>>>>> }
>>>>>
>>>>> liborDates has below values -
>>>>>
>>>>> July 20th, 2014
>>>>> January 20th, 2015
>>>>> January 20th, 2016
>>>>> January 20th, 2017
>>>>> January 20th, 2018
>>>>> January 20th, 2019
>>>>> January 20th, 2020
>>>>> January 20th, 2021
>>>>> January 20th, 2022
>>>>> January 20th, 2023
>>>>> January 20th, 2024
>>>>> January 20th, 2029
>>>>>
>>>>> while liborRates is
>>>>> 8.53854
>>>>> 8.40083
>>>>> 8.39311
>>>>> 8.44955
>>>>> 8.56089
>>>>> 8.65359
>>>>> 8.68869
>>>>> 8.80096
>>>>> 8.9133
>>>>> 8.74109
>>>>> 8.66277
>>>>> 8.98536
>>>>>
>>>>> eval date is January 20th, 2014
>>>>>
>>>>> What is causing negative time exception?
>>>>>
>>>>> Thanks,
>>>>> Varun
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> View this message in context:
>>>>> http://quantlib.10058.n7.nabble.com/Negative-price-with-Forward-Rates-tp14868.html
>>>>> Sent from the quantlib-dev mailing list archive at Nabble.com.
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> CenturyLink Cloud: The Leader in Enterprise Cloud Services.
>>>>> Learn Why More Businesses Are Choosing CenturyLink Cloud For
>>>>> Critical Workloads, Development Environments & Everything In Between.
>>>>> Get a Quote or Start a Free Trial Today.
>>>>>
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> QuantLib-dev mailing list
>>>>> Qua...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/quantlib-dev
>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> <https://implementingquantlib.blogspot.com>
>>>> <https://twitter.com/lballabio
>>>>>
>>>>
>>>>
>>>
>>>
>>>
>>> --
>>> <https://implementingquantlib.blogspot.com>
>>> <https://twitter.com/lballabio>
>>>
>>> ------------------------------------------------------------------------------
>>> WatchGuard Dimension instantly turns raw network data into actionable
>>> security intelligence. It gives you real-time visual feedback on key
>>> security issues and trends.  Skip the complicated setup - simply import
>>> a virtual appliance and go from zero to informed in seconds.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=123612991&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> QuantLib-dev mailing list
>>> Qua...@li...
>>> https://lists.sourceforge.net/lists/listinfo/quantlib-dev
>
>
>
> --
> <https://implementingquantlib.blogspot.com>
> <https://twitter.com/lballabio>