A portfolio health check examines the structural integrity of your holdings — identifying risks that only manifest during drawdowns. Like a structural survey of a building, it reveals vulnerabilities that are invisible under normal conditions but critical under stress.
KlirInvest's Health Score (model portfolio_health_score@2.1.0) combines six weighted convex penalties into a single composite from 0–100, using the formula "Health Score = 100 − Σ(weight × penalty)". Here is what each of the six signals measures, why it matters, and what the data shows.
Signal 1: Volatility (weight 0.20)
What it is: Annualised portfolio standard deviation of returns.
Why it matters: Volatility is the most basic measure of how much a portfolio's value can swing in either direction. Higher volatility increases both the size and the emotional cost of drawdowns.
What the data shows: Higher volatility is strongly associated with deeper interim drawdowns, even when long-run expected returns are similar.
Signal 2: Tail Risk / CVaR (weight 0.25)
What it is: Expected Shortfall at 95% — the average loss in the worst 5% of modeled outcomes. The standing metric uses the disclosed Gaussian baseline; Student-t assumptions apply only when Tail Risk Mode is explicitly enabled.
Why it matters: Standard deviation tells you about typical moves; CVaR tells you what happens in the bad ones. Real return distributions have fat tails, so portfolios that look similar on volatility can differ sharply on tail risk.
What the data shows: Portfolios with elevated CVaR experience disproportionately larger losses during crises than their volatility alone would suggest.
Signal 3: Concentration (weight 0.15)
What it is: The largest single-position weight and the sum of the top-5 weights. These two weight-based measures penalise portfolios where a small number of holdings dominate.
Why it matters: JP Morgan Asset Management's Agony & Ecstasy analysis of Russell 3000 constituents (1980–2020) reports that approximately 40% of stocks experienced a permanent decline of 70% or more from their peak with no recovery. Concentrated portfolios have less capacity to absorb a single such position.
What the data shows: Portfolios with elevated single-name or top-5 weight tend to experience materially deeper drawdowns during broad corrections than well-diversified portfolios.
Signal 4: Correlation Stress (weight 0.15)
What it is: The instability of pairwise correlations — how much correlations rise under stress relative to calm-period correlations.
Why it matters: Correlation spikes during crises. The average pairwise correlation in the S&P 500 rose from 0.27 to 0.79 during the 2008 financial crisis. Portfolios that look diversified in calm markets can behave as a single bet during drawdowns.
What the data shows: Portfolios with higher correlation instability tend to exhibit substantially deeper maximum drawdowns than portfolios whose holdings remain more independent under stress.
Signal 5: Drawdown (weight 0.15)
What it is: A combination of drawdown depth (how deep the current trough is) and drawdown proximity (how close the portfolio is to its prior maximum drawdown). Convexity exponent is higher here (k=2.5) because deep drawdowns matter disproportionately.
Why it matters: If a drawdown causes panic selling, the portfolio's risk profile exceeds the investor's psychological capacity — regardless of what their stated risk tolerance suggests.
What the data shows: Dalbar data consistently shows that investors who experience drawdowns exceeding their psychological tolerance sell at or near the bottom, locking in losses and missing the recovery.
Signal 6: Data Confidence (weight 0.10)
What it is: A data-quality proxy covering history depth, outliers, and coverage of price/return inputs for your holdings. It is NOT a market-liquidity / bid-ask / ADV measure — the DB column name is retained for backward compatibility only.
Why it matters: Every other component above is computed from input data. Sparse history, gappy coverage, or outlier-heavy series reduce the reliability of volatility, tail-risk, correlation, and drawdown estimates. Treating data confidence as a structural signal acknowledges that a score is only as trustworthy as the data behind it.
What the data shows: Portfolios with thin or noisy input data typically show larger gaps between modelled and realised drawdowns during crises, because the other five components are themselves estimated less precisely.
The Health Score: All Six Signals Combined
KlirInvest combines the six signals into a single Health Score from 0–100:
- 80–100: Structurally robust — well-balanced across all six dimensions
- 60–79: Generally sound with identifiable improvement areas
- 40–59: Structural vulnerabilities present in one or more dimensions
- Below 40: Significant structural issues — portfolio is vulnerable to volatility, tail, concentration, correlation, or drawdown stress
The Health Score updates as market conditions change. A score of 75 during calm markets might decline to 60 during a volatility spike if correlation instability intensifies — revealing stress-dependent vulnerabilities that static analysis misses.
Related Diversification Concepts the Health Score Does Not Directly Score
The signals above are the actual inputs to the Health Score. Several other diversification concepts are valuable when interpreting your portfolio but are not direct inputs to this specific score:
- Sector balance — distribution across market sectors. Useful context, but the score's concentration penalty above is weight-based, not sector-based.
- Geographic distribution — domestic vs international exposure and home-bias considerations.
- Asset-class balance — allocation across equities, fixed income, alternatives, and cash relative to horizon and risk tolerance.
- Risk-adjusted return ratios — Sharpe, Sortino, and related efficiency ratios.
Treat these as complementary lenses, not as Health Score inputs.
When to Run a Health Check
- Quarterly: Full review of all six signals
- After major market moves: Check whether weights, correlations, or drawdown proximity have shifted
- After life changes: Marriage, retirement, inheritance, or income changes alter appropriate risk profiles
- Before major portfolio changes: Evaluate the structural impact of proposed adjustments before executing
See how your portfolio scores — run your free analysis at [KlirInvest](/free-tools?tab=analysis).