Transfer Learning using ahead::ridge2f on synthetic stocks returns

This article was first published on T. Moudiki's Webpage - Python , and kindly contributed to python-bloggers. (You can report issue about the content on this page here)
Want to share your content on python-bloggers? click here.

In https://github.com/thierrymoudiki/2025-09-05-transfer-learning-ridge2f, I pretrain ahead::ridge2f (also available Python) on 1000 synthetic stock returns using Bayesian Optimization, and test its performance on real market data.

In order to reproduce results from https://github.com/thierrymoudiki/2025-09-05-transfer-learning-ridge2f, either:

Run 2025-09-07-transfer-learning-stock-returns.Rmd

or

Execute the .R files in the order in which they appear.

Results on 4 major European indices:

[1] "\n=== MEDIAN PERFORMANCE ACROSS ALL SERIES ==="
   Method    Winkler Coverage Interval_Width
1  fgarch 0.05925044     93.5     0.04735842
2  ridge2 0.06024835     94.5     0.04753165
3 rugarch 0.05919827     93.5     0.04753477

Results on 10 CAC40 stocks:

  Method    Winkler Coverage Interval_Width
1  fgarch 0.09799624 96.42857     0.06746349
2  ridge2 0.09573495 95.71429     0.07500853
3 rugarch 0.09797592 97.14286     0.06758644

More details about this model (actually used in an industrial setting):

To leave a comment for the author, please follow the link and comment on their blog: T. Moudiki's Webpage - Python .

Want to share your content on python-bloggers? click here.