Explain Portfolio

Call export_price_series for each holding (parallel) with the inferred period. Compute:

• Per-holding return over the period (close-to-close)
• Weighted contribution to portfolio return: holding return × weight
• Total portfolio return (sum of weighted contributions)
• Rank holdings by contribution (biggest detractors first)

Compare computed return against the client's stated figure. If they diverge significantly (>1%), note the discrepancy.

Step 2 — Attribution layer 1: Market and regime (parallel)

After Batch: call macro_analyst with component="tactical" for each home market (cap at 2). This establishes the macro backdrop: is this a market-wide drawdown, sector rotation, or idiosyncratic?

If quick_portfolio_scores fails or returns <50% coverage, don't retry — Step 3's get_score_analysis per holding is the more reliable factor source and will provide all needed score data.

Step 3 — Attribution layer 2: Factor and thematic

Call get_score_analysis for each holding (parallel) with weeks as int 13 (lookback window, non-default — see conventions §0.1). This is the primary factor data source — more reliable than quick_portfolio_scores for score trend analysis.

Analyze:

• Which factor scores changed most across the portfolio? (e.g., momentum collapsed across the board → factor rotation)
• Are score changes correlated across holdings? (yes → systematic/thematic; no → stock-specific)
• Portfolio-level factor tilt: is the portfolio concentrated in a factor that's out of favor per the regime?

Cross-reference with get_telemetry divergences — do any thematic baskets in the telemetry data match the portfolio's exposure?

Step 4 — Attribution layer 3: Stock-specific

For the top 3 detractors by weighted contribution (parallel):

Note on get_peer_snapshot: The tool may return a different company as the "target" (see Convention #2). Extract the queried stock's scores from wherever it appears in the peer list — do not trust the target_company field blindly. Cross-check against get_company_info if the name looks wrong.

Determine for each: is this stock down because of (a) broad market, (b) sector/factor rotation, or (c) company-specific news?

Step 5 — Score-vs-price divergence (the key insight)

For each holding, compare:

• Price change over the period (from Step 1 — daily data, up to today)
• Score change over the period (from Step 3 — weekly data, may lag by up to 7 days)

Important: If a major event occurred after the last score data point (e.g., earnings miss, indictment, regulatory action), note that scores may not yet reflect this development. The divergence classification still applies but should be flagged as provisional for affected holdings.

Classify each holding:

Portfolio-level verdict: if majority of weighted holdings show "Transient" → overall drawdown is likely noise. If majority show "Fundamental" → drawdown reflects real deterioration.

Step 6 — Custom advice

Based on the verdict:

If transient (scores stable, regime-driven):

• "Your portfolio's fundamentals haven't changed. Quality scores are still [X], defensive scores are [Y]. The drawdown is driven by [regime: risk-off / factor rotation / sector selloff]. Historically these reversals take [timeframe context from macro tactical]. Stay the course unless your risk tolerance has changed."

If fundamental (scores declining):

• Identify which holdings have deteriorating scores. For each, suggest: deeper analysis (/parallax-deep-dive), trim, or replacement.
• If the factor tilt is the problem (e.g., heavy momentum in a mean-reversion regime), suggest rebalancing toward favored factors per the macro tactical outlook.

If mixed:

• Separate the transient holdings (hold) from the fundamental ones (investigate). Prioritize by weighted contribution to the loss.

Output Format

• What Happened (computed portfolio return over the period, compared to client's stated figure; 1-sentence summary of the loss magnitude)
• Performance Attribution (table: each holding with return, weighted contribution, and primary driver tag — Market / Factor / Stock-Specific)
• Market & Regime Context (regime tag, mechanism, 2-3 sentences on what's driving markets; is the broad market down too?)
• Factor Exposure (which factor tilts helped/hurt; connection to the current regime)
• Top Detractors (for each of top 3: what happened, why, and whether scores agree with the price move)
• The Key Question: Noise or Signal? (score-vs-price divergence summary — portfolio-level verdict with per-holding classification table)
• What To Do (conditional advice based on the verdict; specific next steps framed as suggestions, not directives)

Keep tone calm and explanatory. The client is worried — the output should reduce anxiety with clarity, not amplify it with jargon.

Always end with: "This is informational analysis based on Parallax factor scores, not investment advice. All outputs should be reviewed by qualified professionals before any investment decisions."