Concentration Risk Cheatsheet: When Your Index Fund Is a Tech Bet

Buying "the whole market" through an S&P 500 fund is the textbook way to diversify. Five hundred companies, eleven sectors, one low fee. But in July 2025 the ten largest stocks in the index made up 37.3% of its value, and one company, Nvidia, weighed 7.83% on its own, more than each of five of the index's eleven sectors, according to Reuters.

That does not make index funds a bad idea. It means the label "500 stocks" says less about your risk than it used to. A large share of the money in a broad US index fund now rides on a small group of mostly technology and technology-adjacent companies, priced at higher multiples than the rest of the market. If you own an S&P 500 fund, a total US market fund, or even a global fund, you own a tech bet whether you chose one or not.

This cheatsheet covers how cap weighting produces concentration, how concentrated the market is in mid-2025 versus history, how to measure it with the effective number of holdings, and what equal weight, global diversification and factor tilts actually change, including what they cost.

Educational content, not financial advice.

1. How cap weighting works

Three properties of cap weighting explain both its popularity and its concentration:

• It is self-rebalancing. When a stock rises, its weight rises automatically. The fund does not need to trade, which keeps turnover, trading costs and taxable distributions low.
• It mirrors the market. A cap-weighted fund holds what the average investor holds in aggregate. Not everyone can beat it, because together everyone is it, before costs.
• It is momentum by construction. Winners grow their weight; losers shrink. When a few companies compound faster than the rest for years, the index ends up dominated by them. Nothing in the rules caps a single company's weight.

The last point is not a flaw in the formula. It is the formula. Concentration is what cap weighting looks like after a long run of a few big winners.

2. How concentrated is the S&P 500 in mid-2025?

The numbers below come from dated reports published before this article. Different providers count slightly differently (for example, whether Alphabet's two share classes count as one company), so expect a point or two of disagreement between sources.

Two observations:

• Concentration is in a theme, not only in names. Most of the top 10 are technology companies or depend on the same drivers: AI spending, cloud, advertising, semiconductors. If those drivers stall, the companies are likely to fall together, so ten names behave more like a few bets than like ten independent ones.
• Concentration and valuation go together. The largest stocks are also the more expensive ones. That is not a forecast; it is a description of where the risk sits.

3. Historical comparisons

Is this unprecedented? It depends on the measure and the time window.

The honest reading:

• On the top-10 measure, mid-2025 is a record for the modern S&P 500 data (since 1975), well above the 2000 peak.
• On longer history, it is not unique. The 1930s, 1950s and 1960s saw single companies with weights larger than any stock today. Lamont's argument in the same Goldman report is that concentration follows profits: the biggest firms are where the earnings are.
• High concentration did not predict short-term returns in Goldman's analysis: no relationship over the following week, month, six months or year. Over ten years, Goldman's David Kostin found high concentration associated with lower index returns, and his model with a concentration variable forecast about 3% annualized for the S&P 500 over the next decade. That is one model's estimate, not a fact, and other strategists disagree.
• Global markets are often more concentrated. Lamont points to Switzerland and France, to Taiwan and South Korea (where one stock is over 20% of the market), and to Finland two decades ago, where one stock was over 70%.

4. Measure it: the effective number of holdings

Top-10 weight is easy to read but ignores everything below rank 10. A better single number is the effective number of holdings, the inverse of the Herfindahl-Hirschman Index (HHI):

• HHI = sum of squared weights (weights as fractions that sum to 1)
• Effective N = 1 / HHI

It answers: "how many equally weighted stocks would give the same concentration?" An equal-weight 500-stock index has an effective N of 500. A portfolio with one stock has an effective N of 1. MFS put the S&P 500 at about 45 at the end of 2024: 500 names, diversification of roughly 45 equal ones.

def effective_n(weights: list[float]) -> float:
    """Inverse Herfindahl-Hirschman Index: 1 / sum(w^2), weights normalized to 1."""
    total = sum(weights)
    return 1 / sum((w / total) ** 2 for w in weights)


def top_share(weights: list[float], n: int = 10) -> float:
    return sum(sorted(weights, reverse=True)[:n]) / sum(weights)


def cap_weights(weights: list[float], cap: float) -> list[float]:
    """Cap each weight, spread the excess pro rata over uncapped names, repeat."""
    w = [x / sum(weights) for x in weights]
    while max(w) > cap + 1e-12:
        over = [x > cap for x in w]
        excess = sum(x - cap for x, o in zip(w, over) if o)
        free = sum(x for x, o in zip(w, over) if not o)
        w = [cap if o else x * (1 + excess / free) for x, o in zip(w, over)]
    return w


# Illustrative 500-stock index: top 10 hold 37%, a long tail holds the rest
top10 = [7.5, 7.0, 6.0, 4.0, 3.5, 2.5, 2.0, 1.6, 1.5, 1.4]       # percent
tail_raw = [(i + 25) ** -1.2 for i in range(490)]
tail = [63 * t / sum(tail_raw) for t in tail_raw]
index = {
    "cap weight": top10 + tail,
    "5% cap": cap_weights(top10 + tail, 0.05),
    "equal weight": [1.0] * 500,
}

for name, w in index.items():
    print(f"{name:12s} top 10 = {top_share(w):5.1%}   effective N = {effective_n(w):5.0f}")

# cap weight   top 10 = 37.0%   effective N =    48
# 5% cap       top 10 = 32.6%   effective N =    66
# equal weight top 10 =  2.0%   effective N =   500

The weights are illustrative, not real index data, but the shape matches mid-2025: a 37% top 10 on a long tail lands close to the effective N of 45 that MFS reported. Use it on your own holdings:

• Look through funds. Combine the holdings of every fund you own, weighted by how much you hold of each fund, then compute effective N and top-10 share on the combined list. Two "different" funds often hold the same ten names.
• Count themes, not tickers. Group holdings by what drives them (AI and semiconductors, consumer platforms, banks) and compute the same metrics on the groups. Ten correlated stocks are fewer than ten bets.
• Track the trend. The level matters less than whether it is rising fast. A portfolio can drift from effective N of 80 to 45 without a single trade.

Effective N measures weight concentration only. It does not see correlation: 50 equal-weight bank stocks have an effective N of 50 and still move together in a banking crisis.

5. Option A: equal weight

The S&P 500 Equal Weight Index, launched by S&P in January 2003, holds the same constituents as the S&P 500 at equal weight and rebalances quarterly.

Equal weight changes the risk you hold rather than removing it. You swap concentration in mega-caps for more exposure to smaller, more cyclical companies. In Goldman's November 2024 report, Kostin recommended that non-taxable investors such as pension funds shift some allocations to equal-weighted benchmarks. That is advice for institutions in a specific tax position, quoted here as context, not a recommendation for anyone else.

A middle ground is a capped index (no stock above, say, 5%), as in the Python example. It keeps most of the cap-weighted structure and lowers the extremes. The effect on effective N is modest compared with equal weight.

6. Option B: global diversification

The intuitive fix is to own the world instead of one country. It helps less than people expect if you stay cap-weighted.

• In Goldman's November 2024 report, the US was about 70% of the MSCI All Country World Index, and the 10 largest US stocks were over 20% of the entire global index.
• So a cap-weighted global fund still holds the same top names at large weights, just diluted by about a third.
• Many non-US markets are concentrated in their own way: one stock is over 20% of the market in Taiwan and in South Korea, and Switzerland and France are more concentrated than the US (Lamont).

Global diversification pays off when leadership changes between regions, which no one can time. It adds currency risk and sometimes higher fund costs.

7. Option C: factor tilts

Factor funds weight or select stocks by characteristics rather than size alone. The common ones, and how they interact with mega-cap concentration:

Points to keep in mind:

• Not every factor diversifies away from mega-caps. Quality and momentum can hold the same names. Check holdings, not labels.
• Factors have their own concentration. A value fund can be heavy in financials and energy. Run the section 4 checks on any factor fund.
• Tracking error is the real cost. It is easy to accept in a spreadsheet and hard to live with in a bull market led by the stocks you underweight.

8. A decision checklist

• [ ] Do you know the top-10 weight and effective N of your combined holdings, looking through all funds?
• [ ] Have you grouped holdings by theme, and do you know your total exposure to AI and semiconductor-driven companies?
• [ ] If the top 10 fell 40% together while the rest of the market was flat, what would happen to your portfolio, and could you hold through it?
• [ ] For any alternative, do you know its fees, turnover, tax impact and the periods when it lagged, and would you hold it through five years of trailing the S&P 500?
• [ ] Have you written down the rule you will follow (target weights, rebalancing schedule) so the next move is not driven by the last headline?

Why this matters

• "Diversified" is a measurement, not a label. A 500-stock index with an effective N near 45 behaves more like a concentrated portfolio than its name suggests.
• Cap weighting concentrates by design. It rewards winners with weight, cheaply and automatically. That is its strength and the source of today's concentration.
• Records depend on the yardstick. The top-10 weight is at a modern record; on 100-year data, single stocks have been bigger before. Both are true.
• Every alternative is a different bet. Equal weight bets on the rest of the market, global on other countries, factors on characteristics. None is free of risk or cost.
• The main risk is unexamined exposure. Knowing what you own matters more than which structure you choose.

Final Thoughts

Index funds solved stock picking for most people by replacing judgment with a rule. The rule still works: low cost, broad, hard to beat after fees. But the rule produces a portfolio whose risk changes over time, and in mid-2025 it has produced one where roughly a third of every dollar sits in about ten companies driven by similar forces.

You do not need to predict whether those companies will keep leading. You need to measure your exposure, decide whether it matches the risk you intend to take, and if not, choose an alternative whose costs and failure modes you understand in advance. Compute your effective N. Then decide on purpose.

Sources

• Reuters (Lewis Krauskopf), US stock market concentration risks come to fore as megacaps report earnings (July 23, 2025, syndicated): top-10 weight 37.3%, January record of about 38% in S&P Dow Jones Indices data since 1975, Nvidia 7.83%, technology 33.9%, top-10 P/E about 26x vs. 20x; S&P 500 up over 60% since the end of 2022, more than double the equal-weight version.
• Fortune (Sasha Rogelberg), The S&P 500 is 'extremely concentrated,' Apollo economist says (July 25, 2025): top-10 share of about 40% of market cap.
• Goldman Sachs Global Investment Research, Top of Mind, Issue 133: Market Concentration: How Big a Worry? (November 25, 2024): top 10 at about 36% vs. a 20% average and a 25% peak in 2000; highest concentration since 1932 by largest vs. 75th percentile stock; 3% annualized 10-year forecast; equal weight ahead in nearly 80% of rolling 10-year periods since 1970; Owen Lamont on 1950s and 1960s concentration and other countries; US about 70% of MSCI ACWI, top 10 US stocks over 20% of the global index.
• MFS, The Disappearance of Diversification (December 2024), as summarized in AssetMark, On the Mark: Index Diversification (March 2025): effective number of stocks about 45, lowest diversification since 1973; top 10 at 23% and 18% five and ten years earlier.
• MSCI (Donald Sze, Martin Gyugyi), Feeling Deja Vu After 25 Years? (March 24, 2025): technology and communication services about 40% of US market cap in both 2000 and 2025.
• S&P Dow Jones Indices, FAQ: S&P 500 Equal Weight Index: launched January 2003, same constituents as the S&P 500, quarterly rebalancing.