Spain
Mexico
Brazil
Poland
Taiwan
Philippines
South Korea
The essentials
- Different asset names can conceal the same issuer, sector or currency exposure.
- Correlation describes a return relationship that can change during stress.
- A weighted return calculation does not prove that a portfolio is well diversified.
Count exposures, not just labels
A portfolio's number of line items says little about its independent exposures. Several funds and direct shares can all depend on the same companies or industry. Asset allocation describes weights across asset classes; diversification also examines how exposure is distributed within those classes and whether the holdings share important risks.
Look-through arithmetic makes overlap visible. If a fund is 40% of a portfolio and holds 10% in company X, its indirect X exposure is 4% of the portfolio. Adding a direct 6% X holding gives 10% total exposure. This is a hypothetical calculation; actual analysis also needs comparable holdings dates.
Correlation and covariance intuition
Correlation summarizes the linear relationship between two return series on a scale from −1 to +1. Values near +1 indicate closely aligned movements; values near −1 indicate opposing movements. Near zero means little linear relationship in that sample, not that every possible relationship or shared risk has disappeared.
Covariance considers whether returns depart from their own averages together and by how much. Portfolio variability depends on these joint movements as well as each asset's volatility and weight. Two asset pairs can have the same correlation but very different volatility, so correlation alone cannot rank their portfolio risk or establish causation.
Work through a 50/50 example
Suppose a hypothetical portfolio begins with equal amounts in assets A and B. Over the same period, A returns −20% and B returns +5%. Their contributions are −10 and +2.5 percentage points, respectively. The portfolio return is 0.5 × (−20%) + 0.5 × 5% = −7.5%.
This assumes no fees, taxes, external cash flows or rebalancing during the period. An initial 100 becomes 40 in A plus 52.5 in B, totaling 92.5. The loss is smaller than holding A alone, but it remains a loss. One period cannot estimate correlation or establish that this mix is suitable.
| Hypothetical asset | Initial weight | Period return | Portfolio contribution |
|---|---|---|---|
| A | 50% | −20% | −10 percentage points |
| B | 50% | +5% | +2.5 percentage points |
| Portfolio | 100% | −7.5% | −7.5 percentage points |
Relationships can change under stress
Historical correlation depends on the observation window and return frequency. A weak relationship in monthly data may not describe simultaneous daily losses during disruption. Shared shocks can make apparently varied assets fall together, although this does not mean all correlations inevitably become one or every asset responds identically.
Federal Reserve stress-model documentation explicitly discusses limitations in estimated correlations and their relevance to current stress conditions. A useful reading check is whether an analysis separates calmer and stressed samples and states its window, currency basis and return definition. A stress scenario exposes assumptions; it does not predict the next market move.
Audit a diversification claim
Start with holdings and weights from the same date, then inspect overlapping issuers, sectors and regions inside funds. Next check whether return comparisons share a currency and period. Strong past performance, or many different product names, cannot by itself demonstrate that exposure to a common risk has been reduced.
Weights drift as prices change, and equal money weights need not mean equal contributions to risk. Rebalancing changes those weights and can incur transaction costs or taxes. The calculations here teach how to read exposure and returns; they do not prescribe an allocation or recommend buying, selling or rebalancing particular holdings.
Official sources
An explanation of financial mechanics based on official sources. Hypothetical calculations are not actual trading results or forecasts.
