Diversification
/Diversification Analysis
Diversification Analysis
Learn how to read portfolio clusters, diversification metrics, and low-correlation candidates.
Diversification Analysis shows whether a portfolio is spread across genuinely different sources of risk or whether several holdings have historically behaved like the same position.
The tool looks at diversification from four angles:
Portfolio clusters
Groups holdings that behaved similarly during the selected analysis period.
Number of Effective Assets
Shows how evenly the portfolio weight is distributed across its holdings.
Diversification Ratio
Shows how much combining the holdings reduced overall portfolio volatility.
Diversification candidates
Surfaces eligible ETFs and stocks that historically moved differently from current holdings.
These views answer different questions. A portfolio can have evenly distributed weights but still be concentrated in assets that move together. Read the results together rather than treating any single value as a complete diversification score.
Portfolio Clusters
A cluster is a group of holdings with similar historical return behavior. If several technology funds appear in one cluster, for example, the portfolio may have less behavioral diversification than the number of tickers suggests.
The cluster weight shows how much of the portfolio belongs to that group. A large cluster means a large part of the allocation has tended to respond similarly to market conditions.
PortfoliosLab also flags pairs with correlation of 0.95 or higher as highly correlated. These positions have moved very closely together during the analysis period, although their future behavior can change.
The Asset Correlations guide explains how to interpret the correlation values behind the clusters.
Number of Effective Assets
The Number of Effective Assets measures weight concentration. It answers: “How many equally weighted positions would create roughly the same level of concentration?”
For example, a portfolio may contain 10 assets but have an effective number of only 3 if most of the money is held in a few positions. If all 10 assets have equal weights, the effective number is 10.
A higher value means the capital is distributed more evenly. It does not mean the holdings behave differently from one another. That is why this metric should be read alongside the Diversification Ratio and portfolio clusters.
Diversification Ratio
The Diversification Ratio compares the portfolio's volatility with the weighted average volatility of its individual holdings.
A value near 1.00 means the combination provided little volatility reduction. A higher value means the holdings offset more of one another's movements during the measured period.
PortfoliosLab provides context using the distribution of diversification ratios across thousands of portfolios:
- Below 1.20 is in the bottom quarter of observed portfolios.
- 1.20–1.80 is the range where most observed portfolios sit.
- Above 1.80 is in the top 5% of observed portfolios.
These thresholds describe how the result compares with other portfolios in the dataset. They are not targets, and a high historical value does not guarantee lower future risk.
Diversification Candidates
The candidate tables surface ETFs and stocks that have historically moved differently from the portfolio's current holdings. A lower Avg Correlation means the candidate had a weaker average relationship with the holdings for which correlation data was available.
Candidates are screened before they appear. The results therefore form a research shortlist rather than a complete market ranking.
A low-correlation candidate does not automatically improve the portfolio. Its effect depends on its allocation weight, volatility, exposure, costs, liquidity, and whether the historical relationship persists.
Diversification candidates are historical matches, not forecasts or recommendations. Review the asset and its role before changing an allocation.
How Candidates Are Selected
PortfoliosLab evaluates ETFs and stocks separately. It first builds an eligible group of up to 3,000 symbols of each type, ordered by site usage and then by Risk / Return Rank.
All candidates must:
- Trade in a market represented by the portfolio when market information is available.
- Have a PortfoliosLab Risk / Return Rank of at least 50.
- Have at least ten years of quoted price history for the current all-time search.
- Be outside the current portfolio.
- Not be delisted, marked as low quality, or classified as inverse.
ETFs must also have more than $100 million in assets under management and be classified as 1x, which excludes leveraged products. Stocks must have more than $1 billion in market capitalization.
The analysis then checks each candidate against the portfolio holdings. A candidate passes the correlation screen only when its highest available correlation with any holding is 0.50 or lower. The displayed Avg Correlation is the average of its available holding correlations, with lower averages shown first.
The calculation uses available stored pairs and does not require a correlation record for every holding. Treat a candidate with limited overlapping history or missing pairs with additional caution.
Calculation Details
Portfolio clusters are based on the correlation matrix for the selected portfolio analysis period. PortfoliosLab converts correlation into a measure of distance, then uses hierarchical Ward clustering to group holdings with similar behavior. The result contains no more than ten clusters.
The Number of Effective Assets is calculated as:
1 / Σw²
Here, w is each holding's portfolio weight. Squaring the weights gives larger positions more influence, so a concentrated allocation produces a lower effective number.
The Diversification Ratio compares the weighted average volatility of the individual holdings with the volatility of the combined portfolio:
weighted average individual volatility / portfolio volatility
Because the combined portfolio volatility reflects how the holdings moved together, the ratio incorporates weights, volatility, and correlation. The table shows the value over 1Y, 3Y, 5Y, 10Y, and all available history when those periods can be calculated.
Limits of the Analysis
Diversification Analysis is based on historical prices and available metadata. It does not show:
- Whether past relationships will persist.
- Whether correlations will rise during a future market decline.
- Whether a candidate matches an investor's objectives or constraints.
- Future returns, valuation changes, trading costs, or tax consequences.
- Every possible holding overlap or shared economic exposure.
The Understanding Diversification guide covers the broader concept. Use Asset Correlations when you want to inspect individual relationships in more detail.