Look-through: what is really inside your ETFs

In depth8 min read · 6 sources

With a look-through, every fund in the portfolio is replaced by its holdings. This shows you how much Apple, US or technology exposure you really have, even when the shares are spread across several ETFs that overlap. From the same holdings you can derive metrics for concentration and risk that a glance at the fund names would never reveal.

Overlaps add up

The share of a stock, a country or a sector in the overall portfolio is the sum across all funds: the fund's weight in the portfolio times the share within the fund.

In the same way, countries, regions, sectors, currencies and asset classes can be summed across all funds. If the US share in the MSCI World ETF is 72.9% (holdings list of 22 October 2026) and 30% S&P 500 is added, the portfolio's US share is 0.7 × 72.9% + 0.3 × 100% = 81.0%. A “broadly diversified” portfolio of five ETFs can thus be mostly invested in a single country.

How much two ETFs overlap

A simple measure of the overlap between two funds is the sum of the smaller weights for each stock: what both hold only counts up to the lower weight.

Effective number of holdings

The effective number measures concentration, not the number of securities. With equal-sized holdings it equals the number of holdings; the more unequal the weights, the smaller it becomes. Two examples from holdings lists of 22 October 2026:

A world ETF with 1,200 stocks is therefore diversified roughly like 86 equal-sized holdings, because the large US technology stocks make up a considerable share. This is not a flaw in the index but a consequence of market capitalisation weighting. It is worth knowing, though, when people talk about “1,200 stocks”.

What the research says

That what matters is the overall effect rather than the individual building blocks is the core of the portfolio theory of Markowitz (1952): the risk of a portfolio depends on the weights and the covariances. If you buy two funds that largely hold the same stocks, you gain little diversification. The look-through makes this duplication visible before you have to infer it laboriously from correlations.

How important a deliberate country allocation is can be seen in studies on home bias: French and Poterba (1991) found that investors in the US, Japan and the UK hold almost exclusively domestic stocks, even though international diversification would reduce risk; Coval and Moskowitz (1999) observe, even within the US, a preference among fund managers for companies from their own region. Conversely, a market-cap-weighted world portfolio can today have a very high share in a single country. The look-through shows both.

Concentration is the norm in stock markets: Bessembinder (2018) shows that more than half of US stocks since 1926 returned less than money market instruments over their lifetime, and around 4% of companies account for all of the wealth creation. Broad indices are certain to capture these few winners and therefore inevitably carry large individual positions. For fund analysis, Daniel et al. (1997) use holdings to benchmark each stock against comparison portfolios with the same size, valuation and prior-year return. This is the holdings-based counterpart to style analysis.

For advanced readers: concentration is not the same as risk

The effective number only counts weights. Two equal-sized positions in stocks that move almost in lockstep provide hardly any diversification, even if the metric shows 2. Risk-based measures therefore also take volatilities and correlations into account. One example is the diversification ratio of Choueifaty and Coignard (2008): the weighted sum of the individual volatilities divided by the volatility of the portfolio. It is 1 when everything is perfectly correlated and rises with genuine diversification.

• Risk contributions: a position's share of portfolio risk depends on its weight, volatility and correlation, and can be well above its weight (see risk contribution).
• Bonds: for bond funds, maturities and issuers matter more than countries. From the holdings you can derive duration, key rate durations and spread duration.
• Equity beta: a regression of each holding against the global equity market gives a portfolio beta. This is more accurate than the rule of thumb “every euro in equities has a beta of 1”.
• Taxonomy: reporting the asset class (what you own), risk driver (what generates return and risk) and instrument (how the exposure is built) separately prevents double counting. A leveraged commodity ETF, for example, stays in the commodities asset class, and the leverage is recorded under the instrument.

Where the holdings come from and where their limits lie

• Holdings lists from fund providers as CSV or Excel files, usually updated daily or monthly. They are the most accurate source.
• Your own uploads for providers whose lists cannot be retrieved automatically.
• Rough breakdowns (top 10, sectors) when no complete list is available. A remainder then stays “unassigned”.

Three pitfalls: first, holdings lists are snapshots, often a few days old. Second, synthetically replicating ETFs (with a swap) hold a basket of securities that does not match the index. Their holdings list shows the basket, not the economic exposure. Third, providers write names differently (“APPLE INC”, “Apple Inc.”); only matching via ISIN and cleaned names makes the totals across funds correct.

Step by step: checking your own portfolio

1. Look at countries and regions: does the breakdown match your intention? A world portfolio with an emerging markets ETF has a different US share than one without. Also check how much remains “unassigned”.
2. Check the largest individual holdings: if a stock reaches 5% or more across all funds, that is a bet in its own right, even if it comes from two different ETFs.
3. Compare sectors: technology, financials and healthcare dominate many indices. An additional sector ETF can shift the weights more than its share would suggest.
4. Put the effective number and top 10 into context: if the effective number falls significantly after a purchase, the new fund has concentrated rather than diversified the portfolio.
5. Look at bonds separately: for the bond allocation, duration, key rate points, issuers and currencies matter more than countries.
6. Cross-check with returns: correlations in the Backtest show whether the diversification you found is also reflected in prices.

The look-through answers the question “what do I own?”. Whether this allocation is good depends on your goals: if you deliberately set a country or factor tilt, you can see here whether it worked; if you want broad diversification, you will find unintended clusters.

How Portfolio Lab calculates it

The Look-through replaces each fund with its holdings list, provided it covers at least half of the fund's weight; otherwise it uses the provider's breakdown or the available top holdings and shows the remainder as “unassigned”. Individual securities are matched across all funds by ISIN and cleaned name, and cash and derivatives are not counted as stocks. The effective number is calculated over the looked-through individual securities ((Σ share)² / Σ share²), together with the share of the top 10 and the coverage by data source. In addition, the page shows regions, countries, sectors, a treemap, the taxonomy by asset class and risk driver, and the risk factors. The look-through methodology describes the sources and rules in more detail.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.