Add a strategy optimizer for the cash-deploy backtest

[valuecodes]

What

• The cash-deploy strategy's parameters (cash reserve, refill rate, and the drawdown→deploy tranches) were hand-picked. This adds a Strategy Optimizer that searches them automatically for configurations that beat the index (100%-stocks buy-and-hold).
• New Streamlit-free engine module optimizer.py: an Optuna (TPE Bayesian) search with a threshold-grid encoding that makes every sampled DeployRule valid by construction, a selectable objective (excess CAGR vs index by default, plus a drawdown-capped variant, Sharpe-vs-cash, and raw CAGR), and the optimizer choosing the number of buckets (1–5).
• Walk-forward validation is the honesty layer: each fold optimizes on a train window and scores the frozen winner on a later, unseen test window, so the page leads with an out-of-sample number instead of a curve-fit in-sample one.
• New app/pages/2_Optimize.py page: button-gated search with progress, the recommended config + bucket table, per-fold train-vs-test results, comparison against 100% stocks and the four presets, equity + search-history charts, and prominent curve-fitting / price-return-only warnings.
• Small refactor: lifted _sharpe_vs_cash from the cash-deploy page into metrics.sharpe_vs_cash; added Optuna as a dependency; re-exported the optimizer and cash-deploy engine from the package root.

How to test

uv run poe check
uv run poe test
uv run poe dev

In the app, open the Optimizer page, pick a start year and a small trials/folds budget, press Run search, and confirm the recommended config, walk-forward table, comparison table, and charts render, and that the out-of-sample column is shown.

Security review

No security-impacting changes.

[Copilot]

Pull request overview

Adds an Optuna-driven optimizer and walk-forward validation workflow for the cash-deploy backtest, plus a new Streamlit “Optimizer” page, so strategy parameters can be searched automatically and reported with out-of-sample results.

Changes:

• Introduces a new engine module optimizer.py implementing Optuna search, multiple objective functions, and walk-forward validation.
• Adds a new Streamlit page to run the optimizer, display recommended parameters, fold results, and charts.
• Refactors Sharpe-vs-cash into metrics.sharpe_vs_cash, adds Optuna dependency, and expands package-root exports.

Reviewed changes

Copilot reviewed 8 out of 9 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
uv.lock	Locks Optuna and its transitive dependencies (SQLAlchemy/Alembic/etc.).
pyproject.toml	Adds optuna>=3.6 to project dependencies.
src/portfolio_research_lab/optimizer.py	New optimizer engine (search space encoding, objectives, optimize + walk-forward).
src/portfolio_research_lab/metrics.py	Adds reusable sharpe_vs_cash() helper.
src/portfolio_research_lab/init.py	Re-exports optimizer + cash-deploy symbols from package root.
app/pages/2_Optimize.py	New Streamlit page to run optimization and render results/visuals.
app/pages/1_Cash_Deploy.py	Replaces local Sharpe-vs-cash helper with metrics.sharpe_vs_cash().
tests/test_optimize.py	New test suite covering encoding invariants, objective behavior, optimize(), walk_forward().
tests/test_metrics.py	Adds unit tests for metrics.sharpe_vs_cash().

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.