|
13 | 13 | "Keep the shared comparison group compatible with Plotly below 7 on Python below 3.15 and regenerate all isolated lock metadata." |
14 | 14 | ], |
15 | 15 | "Documentation": [ |
16 | | - "Correct executor-mode configuration, fill-eligibility descriptions, headroom and whole-share target rounding. Link the rebalancing guide to the enum, configuration and executor API reference." |
| 16 | + "Correct executor-mode configuration, fill-eligibility descriptions, headroom and whole-share target rounding. Link the rebalancing guide to the enum, configuration and executor API reference.", |
| 17 | + "Make the Zipline migration guide explicit about input and execution assumptions, with a checked synthetic target-weight example and a link to the tested Diagnostic portfolio-analysis handoff.", |
| 18 | + "Publish the local contribution guide and issue/PR templates linked to the shared policy, preserving private security reporting and the verified usage-documentation destination." |
17 | 19 | ], |
18 | 20 | "Build and compatibility": [ |
19 | 21 | "Update the build backend, pinned workflow actions and vulnerable dependency locks while preserving the frozen comparison framework identities.", |
20 | 22 | "Accept final CPython 3.15 in the canary and describe its steps accurately. Stable support remains CPython 3.12 through 3.14; no blanket optional-package support for 3.15 is added." |
21 | 23 | ], |
22 | 24 | "Compatibility": [ |
23 | | - "Results change in the corrected overnight-session, short-target and futures-cash cases. Target caps apply before whole-unit rounding and do not guarantee a continuous limit on realized exposure.", |
24 | | - "Consumers of raw metric dictionaries must handle non-finite CAGR for extreme short windows. Versioned result exports retain their documented tagged-JSON encoding for non-finite metrics." |
| 25 | + "Results change in the corrected overnight-session, short-target and futures-cash cases. Target caps apply before whole unit rounding and do not guarantee a continuous limit on realized exposure.", |
| 26 | + "Consumers of raw metric dictionaries must handle non-finite CAGR for extreme short windows. Versioned result exports retain their documented tagged JSON encoding for non-finite metrics." |
25 | 27 | ], |
26 | 28 | "Validation": [ |
27 | 29 | "Regenerate correctness evidence from frozen inputs and preserve original timing measurements, using the existing source-certification mechanism for documentation-only source changes." |
|
93 | 95 | "date": "2026-09-12", |
94 | 96 | "sections": { |
95 | 97 | "Changed": [ |
96 | | - "Remove the inert LinearImpact.permanent_fraction parameter and add source certification for changes that preserve measured behavior." |
| 98 | + "Remove the inert LinearImpact.permanent_fraction parameter and add provenance-preserving source certification for performance evidence." |
97 | 99 | ], |
98 | 100 | "Release notes": [ |
99 | 101 | "[Published release notes](https://github.com/ml4t/backtest/releases/tag/v0.1.7)." |
|
0 commit comments