IRR formula: weird pattern

[via sarpi]

I found some older post, so it is known that the IRR of +10, +10 and
-15 is -18%
while if I have the -15 as first number instead, IRR is +22%

Why is that? my understanding of IRR is that the later the negative
outflow, the better the IRR should be.

-18% 10 10 -15
22% -15 10 10

 

[joeu2004]

I found some older post, so it is known that the IRR of
+10,  +10  and -15 is -18% while if I have the -15 as
first number instead, IRR is +22%

Why is that? my understanding of IRR is that the later
the negative outflow, the better the IRR should be.

The problem with unqualified generalizations is that they tend to be
over-generalizations.

If you change -15 to -25, you will find that the IRR function returns
the relative results that you expect: 15.83% in the case, -13.67% in
the second case.

I cannot explain why based on the concept of time-value of money.

But I can tell you how to determine if the IRR function result is a
(perhaps one of many) valid discount rates.

If A2:A4 contains the cash flow ({10,10,-15} or {-15,10,10}) and A1 is
=IRR(A2:A4), note that =NPV(A1,A2:A4) is close to zero.