History construction

[Andre Louw-2]

Hi all,

I have historical rates where the weekends/non-business days are missing.

While looking at History I notice that feeding the History(dates,value)
constructor with lists with missing dates/values causes it to fill these
blanks with Null<double> values. Any objection to me propagating the last
known value into these slots (in my example the Friday's rate will then be
used as the last known rate)? I could always pass a flag to specify whether
this must be done or not?

André Louw
Decillion Limited - "Your Risk Is Our Domain"

Email: [hidden email]
Office: [phone]
Mobile: [phone]
Fax: [phone]

 
-------------------------------------------------------------------------
This e-mail is intended only for the use of the individual or entity named
above and may contain information that is confidential and privileged,
proprietary to the company and protected by law. If you are not the intended
recipient, you are hereby notified that any dissemination, distribution or
copying of this e-mail is strictly prohibited. Opinions, conclusions and
other information in this message that do not relate to the official
business of our company shall be understood as neither given nor endorsed by
it.

[Luigi Ballabio-2]

Hi Andre,

At 02:35 PM 5/8/03 +0200, Andre Louw wrote:
>I have historical rates where the weekends/non-business days are missing.

A common occurrence :)

>While looking at History I notice that feeding the History(dates,value)
>constructor with lists with missing dates/values causes it to fill these
>blanks with Null<double> values. Any objection to me propagating the last
>known value into these slots (in my example the Friday's rate will then be
>used as the last known rate)? I could always pass a flag to specify whether
>this must be done or not?

Hmm. Well, those fixings *are* missing after all, so I'd like them to be
null as default behavior. You could add a flag (with default = false) that
causes last values to be propagated.

Bye,
         Luigi

[Alex McGuire]

Andre Louw wrote:

>Hi all,
>
>I have historical rates where the weekends/non-business days are missing.
>
>While looking at History I notice that feeding the History(dates,value)
>constructor with lists with missing dates/values causes it to fill these
>blanks with Null<double> values. Any objection to me propagating the last
>known value into these slots (in my example the Friday's rate will then be
>used as the last known rate)? I could always pass a flag to specify whether
>this must be done or not?
>
>  
>

I wouldn't recommend that.

There are several ways the missing rates could be filled (last good
value, next good value, interpolation ...), so implementing this in the
constructor  would need more than just a flag, complicating things somewhat.

Also, what does it mean to assign rates to weekends and bank-holidays,
when no such rates exist? Am I right in assuming you want to do this so
as to analyse the rates using some tool that expects daily rates? If so
might it not be a better idea to improve the analysis to take account of
weekends etc.


Supposing you really do need rates for every day, I think it would be
better to add a method to History to either
    a) fill it's own gaps
    b) return a copy with the gaps filled in
in either case taking some parameter that determines the rule used to
fill the gaps. I'd choose b) as I'm an immutability nut, but whatever.

cheers,

Alex McGuire

[Luigi Ballabio-2]

At 03:07 PM 5/8/03 +0100, Alex McGuire wrote:

>Andre Louw wrote:
>>While looking at History I notice that feeding the History(dates,value)
>>constructor with lists with missing dates/values causes it to fill these
>>blanks with Null<double> values. Any objection to me propagating the last
>>known value into these slots (in my example the Friday's rate will then be
>>used as the last known rate)? I could always pass a flag to specify whether
>>this must be done or not?
>I wouldn't recommend that.
>
>There are several ways the missing rates could be filled (last good value,
>next good value, interpolation ...), so implementing this in the
>constructor  would need more than just a flag, complicating things somewhat.

Good point.

>Supposing you really do need rates for every day, I think it would be
>better to add a method to History to either
>    a) fill it's own gaps
>    b) return a copy with the gaps filled in
>in either case taking some parameter that determines the rule used to fill
>the gaps. I'd choose b) as I'm an immutability nut, but whatever.

I'd second that, and being a polymorphism nut, I'd dump the method and
write a few history processors---each of them taking a History and
returning a copy with gaps filled in according to its one rule.

Bye,
         Luigi