"Jerry W. Lewis" wrote in message
...
More likely it is a convergence problem due to the negative return. If
you
provide a negative initial guess (optional 3rd argument) then XIRR returns
-9.834%.

The XIRR does report negative returns elsewhere,
It's an algorithm issue, for example.

If I change the value for 26 Dec 2006 from -940 to -1881, XIRR produces a
negative return.,

If I change the value for 26 Dec 2006 from -940 to -1880, XIRR produces a 0
return.,