peliot
commented
9 years ago Sorry it is not working for you. Here's a quick explanation and some other thoughts. XIRR is a numerical solver (as you see from the code), which is solving for the discount rate at which NPV=0. When you have a series of cash flows with a large negative return (as you do here), NPV as a function of discount rate looks pretty ugly as function, in such a way that makes it difficult for numerical solvers to converge on a solution. I wish the error messages were more informative, but when you get the error message you got, it really means that the solver could not converge on a solution (the reason for that particular error message is that the solver is testing a value for the discount rate for which NPV has no solution). Some other thoughts:
- My goal was to be consistent with OpenOffice Calc. If you try the same set of cash flows in Calc, you will find that it also throws an error, even if you give it the correct answer as a guess. So, this implementation is consistent with OO Calc in that regard (though I realize that doesnt help you!)
- I realize the explanation above was not super clear, but what it comes down to is that XIRR does not perform well for cash flows that represent substantial losses (when the answer is near or above -100%). Your example is one of these cases.
- Given your cash flows, I would suggest using an IRR function, rather than XIRR. There is an IRR function in Numpy. When you have the same period between each cash flow, IRR works. XIRR is only necessary when the intervals between cash flows are irregular. Since IRR can be solved without a numerical solver, it does not suffer from the errors that XIRR yields in cases where the answer is a large negative number.
Regards, Philip
On Fri, Jan 9, 2015 at 10:19 PM, enpassanty notifications@github.com wrote:
I tried using the function to calculate XIRR for the following cashflow:
[(datetime.date(2011, 12, 29), -9000), (datetime.date(2012, 1, 29), 305.38), (datetime.date(2012, 2, 29), 305.38), (datetime.date(2012, 3, 29), 305.38), (datetime.date(2012, 4, 29), 305.38), (datetime.date(2012, 5, 29), 305.38), (datetime.date(2012, 6, 29), 305.38), (datetime.date(2012, 7, 29), 305.38), (datetime.date(2012, 8, 29), 133.04)]
which returns the following error:
return sum([cf/(1+rate)**((t-t0).days/365.0) for (t,cf) in chron_order]) ValueError: negative number cannot be raised to a fractional power
Any idea what's going wrong?
— Reply to this email directly or view it on GitHub https://github.com/peliot/XIRR-and-XNPV/issues/3.
enpassanty
I tried using the function to calculate XIRR for the following cashflow:
[(datetime.date(2011, 12, 29), -9000), (datetime.date(2012, 1, 29), 305.38), (datetime.date(2012, 2, 29), 305.38), (datetime.date(2012, 3, 29), 305.38), (datetime.date(2012, 4, 29), 305.38), (datetime.date(2012, 5, 29), 305.38), (datetime.date(2012, 6, 29), 305.38), (datetime.date(2012, 7, 29), 305.38), (datetime.date(2012, 8, 29), 133.04)]
which returns the following error:
return sum([cf/(1+rate)**((t-t0).days/365.0) for (t,cf) in chron_order]) ValueError: negative number cannot be raised to a fractional power
Any idea what's going wrong?