Skip to main contentChartCompass
Menu

All Insights · strategy-risk

Portfolio concentration and scenario testing

Look through ticker counts to economic exposures, then use transparent shocks to find where correlated losses and liquidity needs may cluster.

By ChartCompass · Fri, 04 Sep 2026 07:34:33 GMT · 5 min read

Owning many tickers does not guarantee diversified economic exposure. An index fund, a technology stock, and an employee share plan can all depend on the same large companies. A domestic company may earn most of its revenue abroad. Several apparently distinct positions may respond together to interest rates, oil, a currency, or market liquidity.

FINRA describes concentration risk as the potential for amplified losses when a large portion of holdings is exposed to one investment, asset class, or market segment. Investor.gov likewise explains diversification as spreading money among investments to reduce risk. A useful concentration review therefore looks through labels and asks what can lose together.

Establish the measurement boundary

State which accounts and exposures are included, the valuation timestamp, base currency, and price source. Include cash, funds, derivatives, employer equity, and material exposures outside the displayed brokerage account when they are in scope. Excluding an account can make a precise percentage answer the wrong question.

For each position, retain market value, quantity, instrument type, currency, issuer, sector classification and provider, geography, duration or rate sensitivity where relevant, liquidity indicators, and any derivative notional and direction. Avoid adding option notional directly to stock market value: options require a scenario-dependent valuation or a clearly defined sensitivity approximation.

Look through wrappers and classifications

Fund labels are not exposures. Use dated holdings when available and record the coverage and publication lag. Allocate a fund's weight across underlying issuers, sectors, countries, or factors, then add direct positions to reveal overlap. The residual “other” bucket should remain visible instead of being forced into known categories.

Classification is a model choice. A payments company may be labeled financial or technology; a multinational's listing country may not represent revenue exposure. Use more than one lens where it matters:

  • issuer and ultimate parent;
  • sector and industry;
  • revenue or asset geography;
  • currency of cash flows and instrument denomination;
  • interest-rate duration or financing sensitivity;
  • commodity input or output exposure;
  • broad equity factors such as size or growth; and
  • liquidity and funding needs.

Employer stock deserves special attention because income and portfolio value may depend on the same business. That connection is not visible in a holdings-only correlation matrix.

Distinguish weights from behavior

Position weight is observable at a timestamp. Correlation and beta are historical estimates. Report the return interval, sample window, missing-data handling, currency treatment, and whether returns include distributions. Compare more than one horizon and include stressed periods; a low average correlation can rise during a market selloff.

Do not use correlation as the only diversification test. It is symmetric and summarizes a particular sample, while losses may be nonlinear or driven by common tail events. Two assets can have modest daily correlation yet both depend on refinancing access that disappears under stress.

Build scenarios as questions

The Federal Reserve's supervisory guidance is written for banking organizations, but its core stress-testing discipline is broadly instructive: use multiple measures, document assumptions, consider severe outcomes, and recognize model limitations. A personal research tool should keep a much narrower claim. It can apply transparent hypothetical shocks to holdings; it cannot certify resilience or prescribe an appropriate portfolio.

Define scenarios by changes in risk variables, not by a dramatic story alone. Examples include an equity-index decline with volatility and spread widening; a parallel interest-rate shift with sector-specific effects; a commodity move combined with currency changes; or a liquidity shock with larger haircuts for less-traded instruments. State whether shocks are instantaneous or unfold over time.

For a simple linear approximation:

scenario contribution = position value × assumed percentage shock

Sum contributions, but display each one. For bonds, options, leveraged funds, and instruments with path dependence, a flat percentage shock can be materially misleading. Use instrument-aware repricing if available and label any proxy.

Scenarios are not forecasts and do not need assigned probabilities to be useful. Their job is to reveal dependencies and decision points. If one moderate assumption dominates the result, investigate the exposure, the data, and the assumption rather than hiding it in a portfolio total.

Connect losses with liquidity

Market loss is only part of stress. Estimate near-term cash needs, collateral or margin calls where applicable, and the ability to transact under wider spreads and lower volume. A position can be a small market-value weight yet create outsized liquidity risk if it is leveraged or difficult to sell.

Record stale prices and non-trading assets explicitly. A smooth valuation series can indicate infrequent marks rather than low risk. If a look-through holding or price is unavailable, preserve the unknown weight and test a conservative alternative instead of treating it as zero.

Turn the map into review questions

A concentration report should show top direct weights, top look-through issuers, sector and geographic distributions, shared factor sensitivities, scenario contributors, data coverage, and timestamps. Then ask: which exposure was intentional, which arose accidentally, which is duplicated outside the account, and which missing data could change the conclusion?

Those are research prompts. Whether a concentration is acceptable depends on objectives, financial circumstances, time horizon, taxes, constraints, and risk capacity that a market-data screen cannot determine.

Limitations

Holdings disclosures can be delayed, classifications subjective, and issuer exposures difficult to decompose. Historical correlations can fail abruptly. Linear shocks understate nonlinear instruments, path dependence, forced selling, and market impact. External income, property, debt, taxes, and private assets may be omitted, while scenario severity and combination remain judgment calls.

This framework is educational and does not assess personal suitability or recommend allocation changes. Open ChartCompass to inspect position overlap, declared scenario assumptions, data coverage, and source timestamps together.

Sources

Return to Insights