Beyond the Number of Stocks: Why Portfolio Diversification Depends on Correlation, Not Just Quantity

Summary: Portfolio diversification depends on the relationship between investments rather than simply the number of stocks or mutual funds held. Harry Markowitz’s Modern Portfolio Theory explains that portfolio risk is determined by individual asset volatility, investment weights and correlations between returns. An investor holding numerous stocks from the same sector may remain exposed to common economic risks, while a smaller portfolio containing assets with different sensitivities may achieve better diversification. In India, factors such as interest rates, commodity prices, currency movements, global demand and market liquidity can influence several sectors simultaneously. Correlations can also change during periods of financial stress, reducing the protection investors expect from diversification. Historical correlation matrices, sectoral index comparisons, rolling correlations and scenario-based stress tests can help identify hidden concentrations. Mathematical portfolio optimisation offers useful insights but depends heavily on estimated returns, volatility and correlations, which may not remain stable. Investors should therefore examine underlying fund holdings, sector exposure, portfolio weights, liquidity requirements and investment objectives before making allocation decisions. Regular but disciplined rebalancing can help maintain an intended risk profile, although transaction costs and taxes must be considered. Ultimately, effective diversification requires understanding how investments interact, not merely increasing the number of securities owned.

Introduction: The Comfort of Owning More Stocks

An investor may own fifteen stocks and still have a fragile portfolio. If most holdings depend on the same economic forces, they can fall together when those forces turn unfavourable. Another investor may own fewer securities but spread exposure across businesses whose returns respond differently to interest rates, commodity prices, consumer demand and global growth. The second portfolio may be better diversified, even though it contains fewer names.

This distinction is increasingly important as more Indian households enter capital markets through direct investing, mutual funds and digital platforms. SEBI’s Investor Survey 2025 (SEBI, 2026) highlights the substantial growth of the market over the past decade, alongside differences in financial knowledge, risk perceptions and familiarity with investment products. Greater access creates opportunities, but it also raises an important question: do investors understand the risks within their portfolios, or do they simply count the securities they hold?

The central insight of Harry Markowitz’s Modern Portfolio Theory (MPT) is that portfolio risk is not simply the sum of the risks of individual investments. It also depends on how those investments move in relation to one another. Diversification is therefore not a shopping list of stocks. It is a problem of combining exposures.

Markowitz’s Insight: The Portfolio Is More Than Its Parts

In his landmark paper “Portfolio Selection,” Markowitz (1952) formalised the idea that investors should evaluate investments in the context of the portfolio they create together. Instead of assessing each security only by its expected return and individual risk, the investor considers the joint behaviour of returns. The objective is to identify combinations that offer the highest expected return for a given level of risk, or the lowest risk for a chosen expected return.

This approach introduced a crucial shift in investment thinking. A security that is volatile on its own is not automatically unsuitable for a portfolio. If its returns do not closely track those of the portfolio’s other holdings, it may help reduce total risk. Conversely, a seemingly stable stock can add less diversification than expected if it moves almost in lockstep with the assets already owned.

In MPT, risk is commonly represented by variance or standard deviation of returns. The portfolio’s variance depends on each asset’s weight, its own variance, and the covariance between every pair of assets. Covariance measures whether two assets tend to move together; correlation standardises this relationship on a scale from -1 to +1. A correlation close to +1 indicates that returns tend to move in the same direction. A correlation near zero indicates a weak linear relationship, while a negative correlation indicates a tendency to move in opposite directions.

Correlation is not a forecast and does not explain why two assets move together. It is a historical statistical measure that helps describe the relationship in a selected period. Its usefulness depends on the quality of the data, the period examined and whether the relationship remains relevant.

Why the Number of Holdings Can Mislead

Imagine a portfolio holding ten bank stocks. It may appear diversified because the investor owns ten separate companies. Yet all ten may be exposed to common forces: credit growth, funding costs, asset quality, regulatory changes, interest-rate expectations and investor sentiment toward financial services. A shock to one of these shared drivers can affect many holdings at once.

Now consider a portfolio spread across banking, consumer staples, healthcare, information technology and energy. These sectors are not independent, so the second portfolio is not automatically safer. However, their earnings and share prices may respond differently to economic conditions. Consumer staples may be relatively defensive during some slowdowns, while energy companies are influenced by commodity prices. IT firms may be affected by overseas demand and currency movements, whereas banks respond to credit conditions and interest-rate expectations. These differences can create greater opportunities to reduce portfolio risk.

The lesson is not that every sector must be represented or that a fixed number of stocks guarantees safety. It is that the underlying sources of risk matter more than the count. Ten stocks driven by one factor can amount to one concentrated bet expressed through ten company names.

The same problem can occur across mutual funds. An investor who holds several funds may assume that the portfolio is diversified, but those funds may own many of the same large-cap companies or follow similar investment styles. Fund count is not the same as exposure diversification. Looking through to the underlying holdings is often necessary to understand what the investor actually owns.

The Mathematics Behind the Intuition

For a two-asset portfolio, the variance of returns is determined by the weights of the two assets, their individual variances and the covariance between them. In simplified notation:

Portfolio variance = w₁²σ₁² + w₂²σ₂² + 2w₁w₂ρ₁₂σ₁σ₂

Here, w represents the proportion invested in each asset, σ represents the standard deviation of its returns, and ρ represents the correlation between the two assets. The final term is the key to the diversification effect: it changes as the relationship between returns changes.

Consider two assets with the same volatility and equal portfolio weights. If their correlation is +1, they move together perfectly in the model and there is no volatility reduction from combining them. If their correlation is zero, the combined portfolio’s volatility is lower than the volatility of either asset under these simplified assumptions. If correlation is negative, the risk-reduction effect can be stronger. Perfect negative correlation could theoretically eliminate volatility in a carefully weighted two-asset combination, but such relationships are rare, unstable and not a realistic assumption for ordinary equity portfolios.

This example explains why diversification is not achieved merely by adding another security. The new holding must change the combined portfolio’s risk profile. Its own volatility matters, but so does its relationship with the existing portfolio. A high-volatility asset may still contribute to diversification if its returns behave differently; a low-volatility asset may provide limited protection if it consistently falls when the rest of the portfolio falls.

The formula also highlights why weights matter. A portfolio with twenty stocks can remain concentrated if a few holdings dominate its value. Equally, a portfolio with many holdings can be exposed to one sector or one economic theme. The correct question is not only “How many securities are held?” but “What proportion of the portfolio depends on each risk driver?”

The Indian Market: Beyond Sector Labels

India’s equity market provides a practical setting for this analysis. It includes sectors with different economic sensitivities, such as banking, insurance, pharmaceuticals, information technology, consumer goods, manufacturing and energy. But sector labels should not be treated as proof of diversification. Companies in different sectors may still respond similarly to global risk aversion, domestic liquidity, currency movements or a broad valuation correction.

The current investment environment also makes it important to distinguish between a company’s business fundamentals and the forces that move its share price over a particular period. Interest-rate expectations can influence financial companies and valuation-sensitive growth stocks. Changes in oil prices can affect energy producers as well as airlines, logistics firms and other fuel-intensive businesses. A weaker global outlook can influence exporters and commodity-linked companies through different channels. These relationships can shift over time.

For investors, the practical task is to test rather than assume diversification. NSE provides historical index and market data, including information for broad and sectoral indices. An analyst could compare daily or weekly total returns for indices such as the Nifty 50, Nifty Bank, Nifty IT, Nifty FMCG, Nifty Pharma and Nifty Energy. Using total-return series, where available, helps account for reinvested dividends rather than relying only on price movements.

A useful analysis would calculate a correlation matrix for a defined period, then compare it with a stressed period. The results could reveal whether sector relationships strengthened during market turbulence or remained relatively distinct. However, a correlation matrix is a snapshot, not a permanent map of risk. The selected dates, frequency of observations and treatment of missing data should be disclosed. No conclusion about current correlations should be made without actually calculating them from a specified dataset.

Diversification Under Stress: When Relationships Change

One of the most important limitations of a simple diversification strategy is that correlations are not constant. During relatively calm periods, sector returns may be influenced by different company-specific or industry-specific developments. During a broad market sell-off, common factors may become more influential. These include sudden shifts in global risk appetite, liquidity shocks and sharp revisions in growth expectations. As a result, assets that previously appeared to offer distinct exposures may decline together.

This does not mean diversification stops working. It means its protection can weaken precisely when investors most want it. A portfolio diversified across companies but concentrated in equities remains exposed to broad equity-market risk. Markowitz’s framework helps reduce avoidable portfolio risk; it does not promise immunity from market-wide losses.

Investors should also distinguish correlation from causation. Two indices may be highly correlated because they respond to the same underlying factor, rather than because one causes the other to move. Similarly, a five-year correlation estimate may hide a very different relationship during a three-month market shock. Rolling correlations, calculated over successive moving windows, help track these changing relationships. However, they remain backward-looking and can be volatile. Stress testing can complement historical correlations. An investor can ask how the portfolio might behave if interest rates rise unexpectedly, oil prices surge, the rupee weakens, domestic demand slows or foreign investors reduce risk exposure. These are scenarios, not predictions. Their purpose is to identify concentrations that a simple count of holdings would miss.

Portfolio Optimisation and Its Limitations

A natural response is to use mathematical optimisation to find the portfolio with the lowest estimated variance. Markowitz optimisation can be useful, but it is not a machine that discovers a permanently best portfolio. Its output depends on estimates of expected returns, volatilities and correlations. Expected returns are particularly difficult to estimate, and small changes in inputs can produce large changes in recommended weights.

A minimum-variance portfolio may assign very large weights to assets whose recent data make them appear unusually defensive. If those relationships change, the portfolio may disappoint. Optimisation can also produce concentrated or impractical allocations unless constraints are imposed. Transaction costs, taxes, liquidity, minimum investment sizes and an investor’s need for cash must be considered outside the basic model.

An equally weighted portfolio is not automatically optimal, but it can serve as a transparent benchmark. A careful comparison might examine three strategies: an equally weighted set of sector indices, a minimum-variance portfolio estimated using historical data, and a broad-market benchmark. The evaluation should use a clearly stated period and, ideally, an out-of-sample test: estimate the weights using past data and assess their performance on a later period not used to build the portfolio.

Relevant measures include annualised volatility, realised return, maximum drawdown and the Sharpe ratio, which relates excess return to volatility. Results should also account for rebalancing frequency and transaction costs. Without these details, a claim that one method “works better” may simply reflect the period selected or assumptions made.

A Practical Framework for Investors

The lesson from portfolio theory can be translated into a few disciplined questions.

First, identify the portfolio’s real exposures. Classify holdings by sector, company size, geography, business model and major economic sensitivities. For mutual funds and exchange-traded funds, inspect underlying holdings to identify overlap.

Second, examine relationships, not just individual risk. Historical correlations can help, but they should be reviewed over more than one period. Consider how assets behaved during both ordinary markets and periods of stress.

Third, assess concentration by weight. A portfolio with many small positions and one dominant holding may remain dependent on that dominant company. Sector and factor concentration can persist even where individual-stock concentration appears low.

Fourth, match diversification to the investor’s time horizon and capacity for loss. A portfolio designed for a long-term goal may be able to tolerate short-term volatility differently from one intended to fund a near-term expense. No mathematical allocation can substitute for an appropriate emergency reserve, realistic goals and an understanding of liquidity needs.

Fifth, rebalance deliberately rather than reactively. Market movements can cause portfolio weights to drift away from the intended allocation. Rebalancing may restore the planned risk profile, but it can also trigger costs and taxes. A defined rule is generally more disciplined than frequent changes based on headlines.

Finally, avoid confusing complexity with sophistication. A portfolio containing numerous stocks, funds and themes may be difficult to monitor without offering meaningful additional diversification. Simplicity, transparency and a clear understanding of risk sources can be more valuable than the appearance of complexity.

Conclusion: Count Risk Exposures, Not Just Securities

Markowitz’s enduring contribution was to show that investment decisions should be made at the portfolio level. The risk of a collection of assets depends not only on how risky each asset is, but also on how the assets interact. This insight remains relevant in an Indian market where access to investing has broadened and investors can assemble portfolios quickly, sometimes without examining the exposures hidden beneath the list of holdings.

Owning more stocks may reduce company-specific risk, particularly when the holdings are genuinely distinct. But the benefit diminishes when investments share the same drivers, overlap heavily or become more correlated during market stress. Diversification is therefore a relationship among assets, not a numerical target.

The practical message is simple: a well-constructed portfolio is not necessarily the one with the most holdings. Rather, it is one that combines risks thoughtfully and reflects the investor’s objectives and ability to withstand losses. The number of securities indicates how many positions a portfolio contains, while correlation helps determine whether those positions genuinely diversify risk.

References

Markowitz, H. (1952). Portfolio selection. The Journal of Finance, 7(1), 77–91. https://doi.org/10.1111/j.1540-6261.1952.tb01525.x

Securities and Exchange Board of India. (2026, January 20). Investor Survey 2025: Main report. https://www.sebi.gov.in/reports-and-statistics/research/jan-2026/investor-survey-2025-_99170.html

National Stock Exchange of India. (n.d.). Historical reports: Capital market daily/monthly archives. https://www.nseindia.com/static/resources/historical-reports-capital-market-daily-monthly-archives

National Stock Exchange of India. (n.d.). Total returns index. https://www.nseindia.com/static/products-services/indices-total-returns-index