scripbox/ex-xirr

An Elixir package for calculating XIRR.

ExXirr

A library to calculate xirr using the Newton Raphson method.

Usage

iex> d = [{1985, 1, 1}, {1990, 1, 1}, {1995, 1, 1}]
iex> v = [1000, -600, -200]
iex> ExXirr.xirr(d,v)
{:ok, -0.034592}

Installation

The package can be installed as:

1. Add ex_xirr to your list of dependencies in mix.exs:

   def deps do
     [{:ex_xirr, "~> 1.0.0"}]
   end

2. Ensure ex_xirr is started before your application:

   def application do
     [applications: [:ex_xirr]]
   end

Test

• Run the test suite using the following

  MIX_ENV=test mix test
  

Benchmarks

• We use benchee for the benchmark tests. Run it using the following

  mix run test/xirr_comparison_bench.exs
  

• The reports are available here.

Credits

This application is built on the fantastic finance-elixir package. Many thanks to tubedude for his work.

ex-xirr's Issues

spawn_link overhead is slow

Describe the bug

Thanks for the library. We are currently using LegacyFinance.xirr() which matches our expectations for the results. It is quite a bit slower than the newer version, but the slowness is unnecessary. Removing pmap results in a dramatic speed up. As a test, I created a Finance based on LegacyFinance module that replaces pmap with Enum.map. This is the result using your benchmark:

##### With input date_values #####
Name                     ips        average  deviation         median         99th %
finance               3.93 K      254.18 μs     ±2.71%      252.99 μs      272.99 μs
ex xirr               3.25 K      307.28 μs    ±12.77%      299.99 μs      421.99 μs
legacy finance        0.44 K     2284.52 μs     ±6.76%     2273.99 μs     2761.35 μs

As you can see, there is an order of magnitude speedup simply by making this change.

I did some experimentation with Stream.map as well, but it turned out that Enum.map was the fastest.

disregard

diregard

Invalid result from big calculation

Running this calculation results in an invalid value 0.006097 (correct value is closer to 0.120581416 according to google spreadsheets).

d = [{2038, 2, 4}, {2037, 11, 2}, {2037, 8, 2}, {2037, 5, 3}, {2037, 2, 4}, {2036, 11, 2}, {2036, 8, 2}, {2036, 5, 3}, {2036, 2, 5}, {2035, 11, 2}, {2035, 8, 2}, {2035, 5, 3}, {2035, 2, 4}, {2034, 11, 2}, {2034, 8, 2}, {2034, 5, 3}, {2034, 2, 4}, {2033, 11, 2}, {2033, 8, 2}, {2033, 5, 3}, {2033, 2, 4}, {2032, 11, 2}, {2032, 8, 2}, {2032, 5, 3}, {2032, 2, 5}, {2031, 11, 2}, {2031, 8, 2}, {2031, 5, 3}, {2031, 2, 4}, {2030, 11, 2}, {2030, 8, 2}, {2030, 5, 3}, {2030, 2, 4}, {2029, 11, 2}, {2029, 8, 2}, {2029, 5, 3}, {2029, 2, 4}, {2028, 11, 2}, {2028, 8, 2}, {2028, 5, 3}, {2028, 2, 5}, {2027, 11, 2}, {2027, 8, 2}, {2027, 5, 3}, {2027, 2, 4}, {2026, 11, 2}, {2026, 8, 2}, {2026, 5, 3}, {2026, 2, 4}, {2025, 11, 2}, {2025, 8, 2}, {2025, 5, 3}, {2025, 2, 4}, {2024, 11, 2}, {2024, 8, 2}, {2024, 5, 3}, {2024, 2, 5}, {2023, 11, 2}, {2023, 8, 2}, {2023, 5, 3}, {2023, 2, 4}, {2022, 11, 2}, {2022, 8, 2}, {2022, 5, 3}, {2022, 2, 4}, {2021, 11, 2}, {2021, 8, 2}, {2021, 5, 3}, {2021, 2, 4}, {2020, 11, 2}, {2020, 8, 2}, {2020, 5, 3}, {2020, 2, 5}, {2019, 11, 2}, {2019, 8, 2}, {2019, 5, 3}, {2019, 2, 4}, {2018, 11, 2}, {2018, 8, 2}, {2018, 5, 31}]
v = [-305, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, 188.0]
iex(91)> ExXirr.xirr(d, v)
{:ok, 0.006097}

Weirdly enough, if you just change the last -5 to -4.8 the calculation will return the expected value, like this.

d = [{2038, 2, 4}, {2037, 11, 2}, {2037, 8, 2}, {2037, 5, 3}, {2037, 2, 4}, {2036, 11, 2}, {2036, 8, 2}, {2036, 5, 3}, {2036, 2, 5}, {2035, 11, 2}, {2035, 8, 2}, {2035, 5, 3}, {2035, 2, 4}, {2034, 11, 2}, {2034, 8, 2}, {2034, 5, 3}, {2034, 2, 4}, {2033, 11, 2}, {2033, 8, 2}, {2033, 5, 3}, {2033, 2, 4}, {2032, 11, 2}, {2032, 8, 2}, {2032, 5, 3}, {2032, 2, 5}, {2031, 11, 2}, {2031, 8, 2}, {2031, 5, 3}, {2031, 2, 4}, {2030, 11, 2}, {2030, 8, 2}, {2030, 5, 3}, {2030, 2, 4}, {2029, 11, 2}, {2029, 8, 2}, {2029, 5, 3}, {2029, 2, 4}, {2028, 11, 2}, {2028, 8, 2}, {2028, 5, 3}, {2028, 2, 5}, {2027, 11, 2}, {2027, 8, 2}, {2027, 5, 3}, {2027, 2, 4}, {2026, 11, 2}, {2026, 8, 2}, {2026, 5, 3}, {2026, 2, 4}, {2025, 11, 2}, {2025, 8, 2}, {2025, 5, 3}, {2025, 2, 4}, {2024, 11, 2}, {2024, 8, 2}, {2024, 5, 3}, {2024, 2, 5}, {2023, 11, 2}, {2023, 8, 2}, {2023, 5, 3}, {2023, 2, 4}, {2022, 11, 2}, {2022, 8, 2}, {2022, 5, 3}, {2022, 2, 4}, {2021, 11, 2}, {2021, 8, 2}, {2021, 5, 3}, {2021, 2, 4}, {2020, 11, 2}, {2020, 8, 2}, {2020, 5, 3}, {2020, 2, 5}, {2019, 11, 2}, {2019, 8, 2}, {2019, 5, 3}, {2019, 2, 4}, {2018, 11, 2}, {2018, 8, 2}, {2018, 5, 31}]
v = [-305, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -5, -4.8, 188.0]
iex(102)> ExXirr.xirr(d, v)
{:ok, 0.120448}