[COMPUMAX-EC]

Checklist before merging this PR:

• Read and understood the contribution guidelines including AI/agent usage.
• Linked at least one approved issue below (an issue with label ok).
• Summarized the updates of this PR under Summary.
• Added an entry under Unreleased in the Changelog.

Fixes #3310.

Summary

Adds a signed: bool = False parameter to darts.metrics.ope(). With signed=True, the absolute value is not taken, so the metric returns the signed overall percentage error (percentage bias). The default behavior does not change.

• The sign follows the existing ope formula, (Σy − Σŷ) / Σy: positive means under-forecasting and negative means over-forecasting. If you'd rather have positive mean over-forecasting, as the issue suggests, it's a one-line change.
• Zero-division handling, quantile and stochastic predictions, multivariate series and sequences of series all work as they do for the unsigned metric.
• A new test, test_ope_signed, covers both signs, the default, multivariate and multi-series inputs, stochastic forecasts with q, and zero division.

Other Information

• pytest darts/tests/metrics/test_metrics.py: 487 passed locally (Windows, Python 3.12). Ruff 0.7.2 check and format --check pass.
• AI use: I used Claude Code to draft the implementation, tests and docstring, and to run the checks. I reviewed and verified every change and can explain the design.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.15%. Comparing base (580358e) to head (127cbed).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #3319   +/-   ##
=======================================
  Coverage   97.15%   97.15%           
=======================================
  Files         175      175           
  Lines       19468    19469    +1     
=======================================
+ Hits        18914    18915    +1     
  Misses        554      554           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[dennisbader]

Thanks a lot for the PR @COMPUMAX-EC 🚀

Everything looks good to me 💯
I simplified the tests to reuse the reference OPE.