optimize-allocation

Optimize Allocation (Max-Sharpe / Black-Litterman)

Goal

Compute the optimal weight per holding and the concrete add/trim changes vs the current book that maximise risk-adjusted return. Engine: PyPortfolioOpt Efficient Frontier with Ledoit-Wolf shrinkage covariance, longs-only, 35% cap per name. Analyst price targets are blended as Black-Litterman views when available.

This is the trading-bucket reweighting tool. It WILL suggest selling overweighted names - distinct from auto-rebalance, which only adds new cash to the long-term ETF core and never sells.

Decision Memory Protocol (load first, log after)

Before forming any view, load prior decisions so verdicts stay consistent across sessions:

• mcp__aifolimizer__get_cross_ticker_lessons (max_lessons=3) - portfolio-level win/loss patterns
• For any name you issue a per-ticker BUY/SELL/TRIM/HOLD/ADD on, also load mcp__aifolimizer__get_ticker_decision_history (ticker=…, max_decisions=5) and mcp__aifolimizer__get_ticker_reflection (symbol=…, n=3). If a prior decision exists and this run flips it, state explicitly WHY (new data / catalyst / price); never silently contradict a logged decision.

After output, log every actionable reweight: for each BUY/SELL/TRIM/ADD issued, call mcp__aifolimizer__log_recommendation (skill="optimize-allocation", ticker, action, conviction, rationale, target_pct, stop_pct). Skipping breaks the cross-session feedback loop and causes drift.

When to invoke

• "What's the optimal allocation / optimal weights?"
• "How much of each should I add or trim?"
• "Rebalance my holdings for best risk-adjusted return"
• After a large drift, new capital, or a thesis change across multiple names

How to run

Step 1 - Profile + regime (call FIRST):

1. mcp__aifolimizer__get_profile - account types, capital (never hardcode)
2. mcp__aifolimizer__get_personal_context - ground the Non-Reg-vs-registered tax note in the user's actual province / marginal_tax_rate_pct / account_waterfall instead of generic text. If present=false, keep the note generic and suggest running profile-setup.
3. mcp__aifolimizer__get_market_breadth - regime; if bear_high_fear, flag that max-Sharpe on trailing returns can over-tilt to recent winners