Re: [Quantlib-users] Failed to match calculation for Interest Rate

[Brian S. <bri...@gm...>]

Many thanks.

I just have one quick question. What the option 'SimpleThenCompounded'
signifies? Is it something like - if the time difference between start
and end date is less than one year then use Simple otherwise
Compounded?

Regards,

On Mon, 23 Nov 2020 at 15:42, David Duarte <nh...@gm...> wrote:
>
> Hi,
>
> First, the ZeroCurve constructor already expects continuously compounded rates  (and uses linear interpolation), so if you just define it as:
>
> spotCurve = ZeroCurve(spotDates, spotRates, dayCount, calendar)
>
> you will get the right result.
>
> Second, the way you are inputting the conventions is not correct. The alternatives for the compounding are:
>
> - Simple
> - Compounded
> - Continuous
> - SimpleThenCompounded
>
> These "objects" in python actually just return an int. To see what I mean, try:
>
> print(Simple, Compounded, Continuous, SimpleThenCompounded)
> # 0 1 2 3
>
> If you specify Compounded, you can then specify the frequency: Annual, Semiannual, etc (these frequencies also just return an int)
>
> So you where actually building a curve with semiannually compounded rates, and that's why your calculations were off
>
> You were building the curve as (Compounded = 1, Semiannual = 2):
>
> spotCurve = ZeroCurve(spotDates, spotRates, dayCount, calendar, interpolation, Compounded, Semiannual)
>
> when what you wanted was (Continuous = 2):
>
> spotCurve = ZeroCurve(spotDates, spotRates, dayCount, calendar, interpolation, Continuous)
>
> Hope this helps
> David
>
>
>
> On Sun, 22 Nov 2020 at 21:22, Brian Smith <bri...@gm...> wrote:
>>
>> Hi,
>>
>> I use QL to calculate the Rate of Interest between 2 forward dates as below -
>>
>> from QuantLib import *
>> import math
>>
>> # Settings
>> todaysDate = Date(1, 9, 2019)
>> Settings.instance().evaluationDate = todaysDate
>> dayCount = Actual365Fixed()
>> calendar = Canada()
>> interpolation = Linear()
>> compounding = Compounded
>> compoundingFrequency = Continuous
>>
>> # Definitions
>> spotDates = [todaysDate, todaysDate + Period("1y"), todaysDate +
>> Period("2y"), todaysDate + Period("3y")]
>> spotRates = [0, 0.066682, 0.067199, 0.067502]
>>
>> spotCurve = ZeroCurve(spotDates, spotRates, dayCount, calendar,
>> interpolation, compounding, compoundingFrequency)
>> spotCurveHandle = YieldTermStructureHandle(spotCurve)
>>
>> spotCurveHandle.forwardRate(spotDates[1], spotDates[2], dayCount,
>> Continuous).rate()  # Forward rate
>> # 0.06659636746440421
>>
>> However when I manually calculate the same, I see visible difference -
>> math.log(math.exp(spotRates[2] * dayCount.yearFraction(todaysDate,
>> spotDates[2])) / math.exp(spotRates[1] *
>> dayCount.yearFraction(todaysDate, spotDates[1]))) /
>> dayCount.yearFraction(spotDates[1], spotDates[2])
>>
>> #0.06771741643835603
>>
>> Can you please help me to understand what actually attributes that
>> difference between QL's number and my calculation?
>>
>>
>> _______________________________________________
>> QuantLib-users mailing list
>> Qua...@li...
>> https://lists.sourceforge.net/lists/listinfo/quantlib-users