Spread Modeling - ML Approach

[AaronDeb]

Purpose

Implementation of the second element of the ML Roadmap - Spread modelling. Implementation of the concepts for the paper "Modelling and trading the gasoline crack spread: A non-linear story" by Christian L. Dunis, Jason Laws and Ben Evans.

Approach

Applying ML to model pair spreads.

Tests for New Behavior

The following unit tests were added:

• test_feature_expander.py
• test_filters.py
• test_futures_roller.py
• test_neural_networks.py
• test_regressor_committee.py
• test_spread_modeling_helper.py
• test_tar.py

Checklist

Make sure you did the following (if applicable):

• Added tests for any new features or behaviors.
• Ran ./pylint to make sure code style is consistent.
• Built and reviewed the docs.
• Added a note to the changelog.

Learning [to be updated]

Describe the research stage

Links to blog posts, patterns, libraries or addons used to solve this problem

[PanPip]

Checked the PR. Filters and models (TAR, RNN, MLP, PiSigma) look good. Will have to check the roller classes.

Waiting for the Notebook to check how the created structures work and if we should change anything in terms of input/output.

[PanPip]

Great draft PR! Roller classes look good.

Waiting for the final code adjustments and docs to do a more detailed review.

[PanPip]

Reviewed the code part of this PR. There are places to fix, as some docstrings are lacking descriptions. Overall, the code itself is approved.

Please check every place where the functionality was later changed, but the variables are still present in the docstrings. (If the functions work as you expect them to)

Fixed some small details here and there.

[PanPip]

These docs are great, a lot of work here. Awesome Futures Rollover section 👍

Please finish the code examples and add references for the images where needed. This is important! Otherwise, the docs are approved.

[PanPip]

Unit tests look good 👌

I see some sections are yet to be finished. Will check them again once the coverage is passing for this PR.

[PanPip]

Thank you for finishing the docs and the unit tests. This PR is almost complete now. 🙂

We should add links to the notebooks in the docs, and I think the unit tests need improvement. Will discuss in more detail over the call.