How PORTIQA evaluates a portfolio.

The evaluator is deterministic: the same holdings, risk profile, engine version, and data snapshot produce the same result. The score is a structured assessment of the portfolio, not a return forecast.

Last reviewed: 3 August 2026 · By PORTIQA Editorial Team

1. Normalise the portfolio

Positive market values are converted to weights using total invested value. Cash is tracked separately. Tickers that cannot be resolved are reported as missing instead of silently receiving invented values. The chosen risk level selects a target profile for concentration and exposure planning.

2. Measure concentration

Position concentration begins with maximum position weight and the Herfindahl-Hirschman Index:

The same weighting logic is applied to sector groupings. Effective positions answers a more useful question than holding count: how many equally weighted positions would have the same concentration as this portfolio? The concentration guide gives a worked example.

3. Build component evidence

The engine calculates separate components for diversification, risk, quality, signal, and trend. It also measures classification confidence, factor-evidence confidence, market-regime alignment, and input coverage. Keeping these components separate is deliberate: a high-quality group of holdings can still be dangerously concentrated, and one blended number should not conceal that disagreement.

4. Apply the risk profile

Low, medium, and high risk profiles use different targets, including different effective-position expectations and exposure ranges. The same portfolio can therefore receive a different exposure plan under a different stated risk level. That is expected behaviour, not score instability.

5. Produce the verdict

Component evidence is combined into a 0–100 health score and a readable status. The response also contains the component scores, trajectory, holding-level analysis, exposure plan, diagnostics, insights, and prioritised suggestions. Treat the top-line score as triage; use the components and diagnostics for decisions. See how to interpret a portfolio score.

Coverage and uncertainty

Missing beta, grade, dividend, trend, macro, classification, or factor evidence reduces what the engine can support. The response exposes coverage percentages, confidence measures, low-confidence weight, missing-factor weight, and a data_gaps list. Integrations should display or act on those fields rather than presenting weak evidence as certain.

Versioning and change control

Every response includes engine_version. Store it with saved evaluations. Thresholds, component logic, or context handling may change between versions, so historical comparisons should compare like with like. Material methodology changes will update this page and its review date.

Limitations

• Scores describe portfolio structure and current evidence; they do not predict future returns.
• Sector labels and factor relationships simplify real companies and can change over time.
• Correlation and regime evidence is historical and can fail during structural breaks.
• A score cannot incorporate tax position, liabilities, time horizon, liquidity needs, or holdings not included in the request.