[ANN] PortfolioOptimisers.jl: Ape together strong

Package Announcements
, , , ,
4

PortfolioOptimisers.jl v0.14.2 Release

Breaking changes

  • Lots of quality of life breaking releases since the inaugural post.
    • Better names for abstract and concrete types.
    • Shorter and unambiguous fieldnames.

Bug fixes

  • Lots of minor and quality of life bug fixes. Most don’t affect any results aside from disallowing certain types/values that would error further down the chain rather than at instantiation.
  • One pretty big bug one relating to high order prior statistics using low order factor prior statistics.

New features

  • New risk measures and formulations.
  • Enable existing measures on more optimisation types.

Docs

  • Lots of new API docs, including instructions on defining interfaces for new estimators and algorithms.
  • A few new examples.
  • Better readmes.

Project maturity

  • Stabler API, still expect breaking changes with v0.X releases.

Roadmap

I don’t have a timeline or a serious plan, but there are a few features I’d like the package to have. In order of desirability:

  1. Cross validation prediction.
  2. Pipelines, i.e. data validation → optimisation → prediction → post-processing
  3. Report generation.
    3a. Better plots.
    3b. Perhaps excel output.
  4. A non-shitpost logo (though tbh I kinda like that it’s a shitpost).

Feedback and contributions welcome

The package is quite functional, but it is missing a lot of docs and sample use. If anyone out there is using/wants to use the package I’d love to hear about how it can be improved and would love some contributions/help with implementation and design.

2 Likes
[ANN] PortfolioOptimisers.jl