Thematic ETFs and hype: what studies, costs and overlaps reveal

Basics10 min read · updated 9 October 2026 · 9 sources

Thematic ETFs promise a share in a big trend, such as artificial intelligence, robotics, cybersecurity, hydrogen or defence. They are often launched when a theme is on everyone's lips and the stocks have already risen sharply. A major study shows what happens on average after that. This article explains how thematic indices are built, what they cost, how much they overlap with a global ETF, why many disappear again and what to look out for before buying, without naming individual products.

What thematic ETFs are

A thematic ETF tracks a narrowly defined index that selects companies by their link to a trend: by the share of revenue from the theme, by keywords in annual reports, by patents or by the judgement of a committee. Unlike sector ETFs, they do not follow a fixed industry classification but cut across industries and countries. They often hold a few dozen stocks, equally weighted or with caps, many mid-sized and small companies and a high share of technology and growth stocks.

Ben-David et al. (2023) describe the evolution of the ETF market as follows: the first ETFs tracked broad indices and offered diversification at low cost; later products follow niche portfolios and charge high fees. Thematic ETFs are the core of this second wave. An overview of how ETFs are structured and work in general is given in the article ETF basics.

Hype and timing: launched after the rally

The most important study on the subject comes from Ben-David et al. (2023). Specialised ETFs lost around 30% on a risk-adjusted basis in their first five years. High fees or hedging demand do not explain this; what matters is that the stocks they contained were overvalued at launch. According to this interpretation, providers cater to the expectations of investors who extrapolate past price gains into the future, with ETFs on attention-grabbing themes.

This fits with what is known about attention. Private investors mainly buy stocks that are currently catching the eye: through news, unusually high trading volume or extreme daily returns (Barber and Odean 2008). When the number of search queries for a stock rises, its price rises in the short term, and later the move reverses (Da et al. 2011). And sharp price run-ups in entire industries significantly increase the probability of a subsequent crash, even though on average they do not predict poor returns (Greenwood, Shleifer and You 2019). So if you only buy a theme once everyone is talking about it, you usually buy at prices that already reflect the future (see speculative bubbles).

The fact that a trend changes the world does not mean that its stocks are good investments: measured over their entire lifetime, a few stocks account for most of the wealth creation in the stock market (Bessembinder 2018), and which ones they will be is not clear at the beginning of a trend.

Costs and concentration

• Ongoing charges: thematic ETFs are usually considerably more expensive than broad index funds (Ben-David et al. 2023); on top of that come the costs of index adjustments, which do not appear in the TER but in the tracking difference.
• Few holdings: with 30 or 50 stocks, a thematic index is a concentrate. A setback at one large holding or a regulatory change affecting the theme hits a large part of the fund at once.
• Factor bet: many themes consist of young growth stocks with high valuations and high interest rate sensitivity. If interest rates rise, such stocks can fall together. The theme is then secondary.
• Small funds, wide spreads: young thematic ETFs often have a small fund size. The bid-ask spread is wider, and the risk of closure rises.

Overlap with a global ETF: worked example

Many investors buy a thematic ETF in addition to a global ETF. How much does it really add to the portfolio that is new? One measure is the overlap: the sum of the smaller of the two weights for every holding the funds have in common (look-through). Illustrative figures: a made-up thematic ETF on semiconductors and AI holds 25 stocks at 4% each. Six of them are among the ten largest in the MSCI World, 14 are included there with a combined 2.5%, and five not at all. Weights in the MSCI World according to the factsheet of 30 September 2026 (MSCI World Factsheet 2026).

The portfolio's weighted TER only rises by 0.04 percentage points, because the thematic ETF is small. On the part that genuinely deviates from the world market, however, you pay an extra 0.49% a year. This is how to compare satellite positions fairly: an expensive fund with a lot of overlap can cost more for every euro of deviation than an even more expensive fund that invests completely differently. Look-through shows which individual holdings add up across all funds in your portfolio; how to manage satellite positions with rules and a risk budget is described in the article core-satellite.

Closures, delisting and survivorship bias

Many thematic ETFs disappear again. If the inflows fail to materialise, a fund is no longer worthwhile for the provider, and it is closed or merged with another fund (fund closure). This has consequences for you: on liquidation, you receive the proceeds and have to reinvest them. Under German tax law, a payout during the liquidation only counts as a tax-free return of capital to the extent that the redemption price is below your acquisition costs (InvStG § 17 2016). Accumulated gains become taxable, regardless of whether you wanted to sell.

If you look at rankings of the best thematic ETFs, you only see the survivors. Closed funds are missing from the statistics, and it is precisely the unsuccessful themes that disappear first. This is survivorship bias, which systematically makes returns in fund statistics look too good (Brown et al. 1992). The same applies to backtests of thematic indices that were only calculated after a theme had succeeded: an index that retrospectively selects the winners of a trend naturally has a brilliant past (see backtest overfitting).

For advanced readers: active share, tracking error and cost per unit of deviation

Because both weight vectors sum to 1, half the sum of the deviations is exactly one minus the overlap. The active share of a portfolio with a small satellite position is therefore simply the share of the satellite times its “new” part. For risk, what counts is the tracking error: if the thematic ETF has a tracking error of 15% relative to the global ETF, a 10% allocation adds 1.5% tracking error to the portfolio. The extra cost divided by the active share gives the cost per euro of deviation. With this metric, you can also compare a thematic ETF with an active ETF or a factor ETF.

Our assessment: in a factor regression, many thematic ETFs are likely to behave like a bundle of growth, technology and small-cap risk, often with a beta above 1. Whether a fund delivers more than these known factors is shown by style analysis or a factor regression. Check how much independent return is left over.

What the research shows

Ben-David et al. (2023) provide the most comprehensive evidence to date: specialised ETFs earned about 30% less on a risk-adjusted basis in their first five years, because their stocks were overvalued at launch. Providers launch what is in demand at the moment. The attention explanation is supported by Barber and Odean (2008), who show that private investors mainly buy attention-grabbing stocks, and by Da et al. (2011), in whose study more attention leads to higher prices in the short term and lower prices later. For entire industries after sharp rises, Greenwood, Shleifer and You (2019) find a significantly increased crash risk. For a general overview of the structure of ETFs and their effect on markets, see Ben-David et al. (2017). Important for context: these are averages across many funds. Individual themes have performed brilliantly; you just do not know in advance which ones.

Checklist before buying

1. Why now? How much have the theme's stocks risen over the last one to two years? Buying after the hype is the typical situation that studies show to be expensive.
2. What is inside? Look at the holdings list, the number of holdings, the weight of the ten largest, countries and sectors. Does the selection really match the name?
3. How much is new? Calculate the overlap with your own global ETF or check it in look-through.
4. What does the deviation cost? Compare the TER and tracking difference with your core holding and divide the extra cost by the new share.
5. How large and how old is the fund? Closure is more likely for small, young funds; the launch date also shows how much of the history shown is merely a back-calculated index.
6. How much is enough? Set a fixed maximum share, for example 5 to 10% of the portfolio, and a rule for when you will sell. Not investment advice.

Limitations and common mistakes

• Confusing a theme with returns: a trend can be real and the stocks already too expensive all the same.
• Mistaking back-calculated indices for real history: before the launch date, the curve is a backtest by the index provider, often built with today's knowledge.
• Collecting many themes: five thematic ETFs with 30 holdings each often overlap heavily with each other and with the global index. That gives you more funds, not more diversification.
• Starting small, letting it grow big: after a good run, the share grows by itself. Without a cap, a satellite position turns into a concentration risk.
• Not planning for closure: with small funds, liquidation can come just when the price is low. Then come tax on earlier gains and the need to reinvest.

How Portfolio Lab calculates it

In Look-through, Portfolio Lab merges the holdings of all funds: identical holdings from different ETFs form one row, the “Positions” column shows in how many of the portfolio's funds a holding appears, along with Top 10, effective number, countries and sectors for up to three portfolios. This shows you how much a thematic ETF really changes. For funds without a complete holdings list, only the largest positions are included. Under Securities, you will find the TER, fund size and launch date. In the Backtest, with the global ETF as the benchmark, you measure the tracking error, information ratio and beta of a portfolio with a satellite position; Style analysis shows which indices or factors a fund actually follows. The app does not calculate a separate metric for the overlap between two funds; the rows with several positions in look-through show the holdings in common. More in the look-through methodology.

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