Investing · South Africa
How Many Shares Should You Hold in a Diversified Portfolio?
There is no share count that makes every portfolio adequately diversified. The weights, underlying businesses and risks shared by the holdings matter more than the number of lines on a statement. Twenty shares at roughly equal weights are different from twenty shares with most of the money in three companies. Funds can also repeat exposures you already own directly.
Choose a portfolio size by testing what one failed investment or common shock could do, whether the holdings provide genuinely different exposures, and whether you can maintain the necessary research and administration. If you cannot do that work, a suitably broad fund may provide a better starting point than assembling a collection of individual shares to reach a target number.
In this article
What a share-count rule leaves out
A count treats a 30% position and a 1% position as one holding each. It does not distinguish two companies exposed to the same commodity price from two businesses driven by different customers and conditions. It says nothing about debt inside those businesses, liquidity in their shares or the price paid.
South African research has addressed the weighting problem directly. The publisher's abstract for Bradfield and Munro's 2017 study contrasts equal-weighted portfolios with the more concentrated weighting schemes relevant to the local market. Its conclusion is that the construction method affects the number of shares required for risk reduction. That is a reason to inspect the method behind a quoted number, rather than adopt it without the study's conditions. Publisher's research abstract
Historical index names also need context. The JSE's harmonisation notice states that calculation and dissemination of SWIX indices ceased from 2 January 2026. An older paper or mandate mentioning SWIX is not a description of the current index arrangement. Check a fund's current mandate and benchmark before using old research to infer its exposures. JSE harmonisation notice
Twenty names can carry very different amounts of concentration
Consider two invented R2 million share portfolios. One holds twenty companies at 5% each. The other puts 30%, 20% and 10% in its largest three holdings, with the remaining 40% divided equally among seventeen companies. Both statements show twenty names.
| Measure | Twenty equal holdings | Twenty top-heavy holdings |
|---|---|---|
| Largest position | 5% / R100,000 | 30% / R600,000 |
| Largest three together | 15% / R300,000 | 60% / R1,200,000 |
| Portfolio loss if only the largest holding falls 50% | 2.5% / R50,000 | 15% / R300,000 |
| Equal-weight concentration equivalent | 20.0 | About 6.7 |
The last row is a mathematical description of weights only. Square each portfolio weight, add the squares and take the reciprocal. Twenty positions of 5% give 1 divided by 0.05, or 20. The top-heavy portfolio gives approximately 6.7. The full holding weights and calculation are supplied with the article.
This does not mean the second portfolio has “6.7 independent investments”, nor does it predict its volatility or loss. The measure does not include correlations, business quality, valuation or market risk. Its use is narrower: it reveals why a reassuring literal count can conceal a much smaller spread of capital.
The 50% fall is also an invented stress, not a forecast. It holds every other investment unchanged to isolate position size. In a real downturn, other holdings may fall too. The result is a starting point for asking whether the financial consequence is affordable, not a certification that the equal-weighted portfolio is safe.
Look through funds before counting additional diversification
Suppose the top-heavy direct portfolio is part of a larger arrangement: 60% of the total is in the direct-share basket, 25% in Fund A and 15% in Fund B. The direct basket's largest company accounts for 30% of that basket. The same company represents an assumed 8% of Fund A and 12% of Fund B.
| Route to the company | Allocation to the route | Company within that route | Exposure in the total portfolio |
|---|---|---|---|
| Direct-share basket | 60% | 30% | 18.0% |
| Fund A | 25% | 8% | 2.0% |
| Fund B | 15% | 12% | 1.8% |
| Combined exposure | 21.8% |
On a R2 million total portfolio, that is R436,000 exposed to the same company. A 50% fall in that company alone would remove R218,000, or 10.9% of the total, under this static illustration. Looking only at the direct holding would miss R76,000 of exposure and R38,000 of that assumed loss.
The correct operation is multiplication within each route, then addition across routes. Adding “30% + 8% + 12%” would be wrong because those percentages use different portfolio denominators. So would assuming the two funds are two independent protections against the direct holding's decline.
FINRA's investor guidance specifically recommends checking fund holdings for overlap with other funds and directly owned securities. It also distinguishes correlated investments from simply numerous holdings. Those are general investment principles, not South African regulatory limits. FINRA: concentration and fund overlap
In an actual review, use holdings from consistent dates where possible. A fund's top-ten disclosure may be insufficient to calculate all overlap. Record what is missing rather than treating undisclosed exposure as zero. Different share classes of the same company and related businesses may require grouping for the risk question, while retaining their distinct legal and trading characteristics.
Diversify the reasons an investment can disappoint
After measuring weights, examine the businesses. A bank, retailer and property company can have different labels while sharing sensitivity to household income, financing conditions and the local economy. Several exporters can share exposure to one commodity or customer market. Geography of listing, reporting currency and sources of earnings are related but different descriptions.
For a South African household, the portfolio also sits beside employment income, a business, property and rand spending obligations. This article concerns the share portfolio; the household concentration question tests that wider balance sheet. Neither assessment should assume that an offshore listing automatically removes every shared risk.
There is also a risk of missing a small group of exceptional long-term performers. Bessembinder's research programme, described by Arizona State University, documents highly concentrated wealth creation across large historical stock samples. That is different from saying that an investor can identify the eventual winners in advance. It is one reason a narrow selection can disappoint even when the wider market performs well. The evidence is historical and does not supply a South African optimum holding count. ASU research and data overview
Buying more names helps only if the additions improve the portfolio's exposures or implementation. Adding an unfamiliar, weakly researched company for the sake of a count can introduce another avoidable mistake. Owning a broad fund can reduce the need to select individual winners, but a sector-specific or concentrated fund still requires scrutiny.
Set a loss budget before a target count
A position limit can begin with a consequence the investor chooses to test. Suppose, purely for illustration, the investor wants a loss in one company to reduce the total portfolio by no more than 3%. If the tested company loss is 50%, the corresponding maximum weight is 6%: 3% divided by 50%. If the test instead allows for losing the entire holding, the maximum weight falls to 3%.
Neither figure is a recommended limit or a prediction of the company's downside. The comparison shows how much depends on the loss assumption. A reassuring limit based on a mild stress can conceal an unacceptable result under a more severe one. Shared losses across several companies can exceed the single-company budget even when every individual holding meets its limit.
Use this arithmetic to make the discussion concrete, then assess business risk, the total portfolio and the household's needs. It cannot replace that assessment. The resulting count is an implementation consequence, not the original definition of safety.
A smaller deliberate portfolio can be defensible
The strongest case for fewer holdings is that an investor has a clear mandate, sufficient research, conviction grounded in evidence and the financial capacity to tolerate being wrong. A specialist portfolio can also form a limited part of a much broader household arrangement. Equal weights are not automatically the correct allocation, and excessive diversification can dilute a genuine analytical advantage while increasing work and costs.
The burden is to show the downside and why it is acceptable. Expertise does not make a business immune to loss. A holding that has grown successfully can become too large for the investor's present needs even if its original purchase was sensible. Review the current position, not only the history that created it.
Before trimming or adding, include tax, dealing costs, liquidity and the alternative use of the money. There may be ways to improve the spread through new contributions or other accounts rather than an immediate sale. A change should improve the whole arrangement, not merely make a concentration statistic look better.
Use a written portfolio test
Ask for the largest single-company and combined exposures after looking through funds. Identify the principal shared business risks, the effect of an explicit loss scenario and whether essential household needs would remain funded. Explain the role of every substantial holding and the evidence that would change that role.
Then assess maintenance. Can you follow the businesses, corporate actions, changing debt, valuation and mandate? What information is missing? A portfolio that is manageable at a small size may become burdensome when inherited accounts, offshore platforms and several funds are added. The count should emerge from a coherent investment process, rather than replace one.
For the next decision, read how a portfolio can withstand a market crash and offshore exposure. Share Portfolio Guidance is relevant when direct-share holdings need a documented assessment of their role, valuation and combined risk. The calculations above remain usable without commissioning a review.
Sources and calculation basis
Sources were accessed on 6 September 2026. The South African study is published in volume 31 of the South African Journal of Accounting Research, 2017; only the publisher's abstract was available, so no numerical holding-count recommendation or full-paper methodological finding is adopted. The JSE notice supplies the subsequent index change. FINRA's explanation is dated 15 June 2022; ASU's page identifies its historical research periods separately.
All portfolio values, companies, fund weights and stresses here are invented. The data show every weight and the arithmetic behind the concentration and overlap comparisons. They do not estimate correlations, expected returns or a probability of loss.
General education only. A personal portfolio decision requires actual holdings, tax circumstances, objectives and capacity for loss.
Download portfolio holdings and diversification (Excel)
Inspect the fixed worked-example records in Excel, with readable headings and filters. The workbook preserves the article’s calculation data; it is not a personal planning calculator.