We continue our short series of articles dedicated to the exploration of trading strategies that derive their functionality from the deep understanding of how Exchange Trading Funds (ETFs) work. In our first post, we discussed how we could use the ETF flows to predict subsequent daily ETF performance. In today’s article, we will analyze how we can use the information about the sensitivity of individual stocks to the ETF arbitrage activity to build a profitable equity factor trading strategy.
There is very little doubt that exchange-traded funds (ETFs) have made factors and indexes more accessible to market participants. One of the key design innovations of ETFs is the implementation of an explicit arbitrage mechanism called creation-redemption that makes an ETF and its underlying portfolio fungible. This arbitrage mechanism ensures that, at least for ETFs holding liquid securities, opportunistic arbitrageurs will keep an ETF’s price and net asset value (NAV) aligned since deviations lead to profitable, essentially riskless trading opportunities. Upon observing this riskless trading opportunity, arbitrageurs act to capture profits by trading in the underlying portfolio until the discrepancy between price and NAV is eliminated.
How do they do it? They capture risk-less profits by trading each constituent security based on its portfolio weight since that is the only way to obtain the fungible unit and utilize the risk-less arbitrage mechanism in ETFs.
An interesting research paper written by John J. Shim (2020) discuss more deeply the whole ETF arbitrage process. The most important finding of the paper is that the stock price movements predicted by ETF arbitrage eventually correct (and converge) over time. More generally, ETF constituent stocks most sensitive to arbitrage trading tend to co-move more with the market and have negative alphas. On the other hand, stocks that are the least sensitive to arbitrage trading tend to co-move less with the market and have positive alphas.
The idea is nicely capture in the following picture.
Why Might It Fundamentally Work?
The study suggests that arbitrage comovement distorts prices and factor loadings due to the following reasons:
1. Mistranslation of factor information: Arbitrageurs’ mechanical trading in ETFs can result in the mistranslation of factor-related information from ETF prices to constituent securities’ prices. This leads to non-fundamental comovement and distortion in prices of individual securities.
2. Influence on observed covariances: Arbitrage comovement affects the observed covariances between securities, as it introduces a significant non-fundamental, mechanical arbitrage component when ETF trading activity is high. This can distort the true fundamental exposures and risk assessments of individual securities.
3. Shift in price discovery: The paper also highlights a shift from a “bottom-up” approach, where price discovery occurs at the individual security level, to a “top-down” approach, where price discovery for factors happens at the ETF level and is transmitted to individual securities through arbitrage. This shift can lead to a distortion in the relationship between factor loadings and individual security prices.
Strategy Proposal Based on Findings
The investment universe consists of heavily traded U.S. equity ETFs (at least $100mm in average daily trading volume). The value-weighted portfolios are formed by sorting securities held by non-industry high-volume on arbitrage sensitivity, which estimate is calculated as from equation (10). We need ETF constituents data for that, and we can use data from the ETF Global dataset for the aforementioned variables. Finally, we formulate a final long-short portfolio (H-L) that goes long (buys) the high arbitrage sensitivity portfolio and short (sells) the low sensitivity portfolio. It is assumed to be rebalanced monthly. Value-weights for constituents are based monthly on the past month’s market capitalization.
For high-volume and large-cap ETFs, the long-short portfolio has an annual alpha of 7.79% and 7.43%, respectively. We would also like to mention that while counterintuitive, these anomalous effects are strongest for large-cap stocks, which is contrary to most asset pricing anomalies where the effects are typically stronger for small-cap stocks, butis consistent with the idea that the most actively traded ETFs (which tend to be large-cap ETFs) result in the most arbitrage trading in the constituent stocks.
Interested?Would you like to try it yourself?
ETF constituents data are just one example of so-called ETF Reference data – the dataset that contains important information related to individual ETFs. Reference data usually contain product information on expenses, underlying index, fund’s AUM, bid-ask spread, industry-specific information, classification details on market exposures, geographic exposures, industry exposures, fund’s constituent data, NAV, shares outstanding, risk factor scoring, etc.
Quantpedia recommends the ETF Global® dataset, an aggregated ETF Reference database of 3,100+ U.S. Listed Exchange-Traded Products (ETPs) that include Exchange-Traded Funds (ETFs), Exchange-Traded Notes (ETNs) and Exchange-Traded Commodities/Currencies (ETCs).
Would you like to learn more about the benefits of using ETF Reference Data?
Strictly Necessary Cookies
Strictly Necessary Cookie should be enabled at all times so that we can save your preferences for cookie settings.
If you disable this cookie, we will not be able to save your preferences. This means that every time you visit this website you will need to enable or disable cookies again.