Re: financial computations

[Al Christians <achrist@easystreet.com>]

Robert Dewar wrote:
> 
> What artithmetic do you use in these functions?
> Note that the use of floating-point for many financial
> computations, e.g. interest rate calculations for many
> instruments, is illegal.
> 

This would be way off-topic for this newsgroup, except that you
are stating it as a reason to use your firm's product, so I'll
inquire:  What do you mean?  What instruments?  Illegal where?
According to what law or ruling?   Are you typing about the SIA
rules for bond calculations?  Certainly not the truth in lending
law?  The Cobol Programmers' Full Employment Act?  

Fixed-point decimal exponentiation is what ought to be illegal, on 
the grounds that it takes to long, as anyone who has done this in
Cobol knows too well.  Interest calculations often require 
exponentiation with non-integer exponents.  The results of these 
calculations cannot be computed exactly, so any means of computation, 
fixed or floating point is an approximation.  I can see that the law 
might require some certain accuracy in the approximations, but I 
don't expect that the law would care whether the decimal point was 
fixed or floating through the intermediate calculations.   With
enough precision, either approach could match the monetary values
produced by the other.


Al