Extracting Systematic 13F Hedge Fund Alpha

There’s an SEC regulation well known to any reader of this blog: institutions holding greater than $100 million in US listed stocks and related options are required to disclose their holdings in such securities to the Securities and Exchange Commission (SEC) no later than 45 days after the quarter-end, in a form known as a 13F filing which is made publicly available.

This regulation started back in 1978, and since then, filings have been a popular source of hedge fund intelligence. Professional and amateur investors alike are known to study 13F filings in an effort to extract portfolio information about otherwise secretive funds. The reality is, the data is voluminous, complex, and disparate. As such, it requires a lot of effort to collect, parse, and organize into useful insights.

In 2016, Novus and Barclays entered a partnership to develop an index to systematically track commonly held stocks from within the hedge fund universe, based on these public regulatory filings. Our goal was to mute the noise in the total mass of 13F data, and identify signals of aggregate wisdom.

The outcome of this partnership was the Novus Barclays Public Ownership Indices, with Novus acting as the data / algorithm specialist to develop data indices, and Barclays Investment Bank separately providing an investible vehicle.

I recently co-authored a QIS Insights piece with Barclays Investment Bank, which discusses the construction methodology and performance of 13F-based approaches like ours. I encourage those interested to check out the full article.

Designing our Study

A narrative we’ve recently been discussing frequently on this blog is that the “best ideas” of hedge funds can produce better-than-average returns. The recent research study with Barclays focuses on measuring whether such risk-adjusted returns can be considered statistically significant. Past research suggested to us that there was a good chance.

For our study, we adapted a test designed for mutual funds by Cohen, Polk, and Silli (2010). Tailoring our work to hedge funds required modifying the aggregation method they had used; more specifically, we looked for conviction and consensus signals in hedge fund portfolios. We also realized that to systematically identify hedge fund alpha in the 13F filings, we needed to select the right group of managers—in particular, those with longer-term fundamental approaches to stock picking.

The “Who” Matters

Thus, an important takeaway from our findings was the impact of deciding who to include in this portfolio – that is to say, which managers’ filings should be used to harvest data for the index. Novus maintains a Hedge Fund Universe (HFU) of managers, broken down by style. Among these are managers who have a longer-term view on equity investments based on fundamental analysis. We classify these funds as “Fundamental Equity Hedge Funds” (FEHF). Broadly speaking, these funds are equity long/short, equity market neutral, equity long-only, and equity event-driven funds.

Using this classification system, we are able to differentiate between the holdings of funds that specifically focus on picking stocks they expect to appreciate over time, and those that turn over their portfolios quickly based on short-term price fluctuations (e.g., quantitative funds). The former set of managers provide useful signals in relation to data derived from quarterly 13F filings disclosed on a 45-day lag, while the latter provides irrelevant noise that we therefore systematically eliminate from our index creation inputs, since the value of portfolio information relating to such high-turnover funds, if any, has long since expired by the time the data becomes available. To the best of our knowledge, we are the first in the academic literature to use qualitative fund classifications in this context, separating out those funds that tend to have longer-term views.

Outcome

In a similar spirit to Cohen, Polk, and Silli, we were able to show that the “best ideas” of certain such hedge funds, when indexed as a group, have historically delivered economically meaningful and statistically significant risk-adjusted returns that outperformed the S&P 500 data index.

More specifically, harvesting conviction and consensus stocks from our chosen set of managers resulted in outperformance of the S&P 500 by 3.80% annually on average, and delivered a Sharpe ratio of 0.75 over the period May 2004 to June 2019. Notably, we were able to show that our findings are also robust even given the lag between the filing date (the quarter-end) and the date on which the holdings become publicly available (45 days later).

These results validated of our intuition around Fundamental Equity Hedge Funds. We were able to confirm through this study that tracking the public positions of these managers in aggregate outperforms those of other classes of manager and has historically delivered a higher grade of systematic 13F hedge fund alpha. This is not to say that these hedge fund managers outperform other managers, with respect to their actual private funds’ performance; 13F public filings do not provide this data, but rather only a quarterly snapshot of the long equity positions and equity options of such funds. Many hedge funds can and will take short positions, hold other types of securities, and change their holdings intra-quarter. Managers also often aggregate the positions of all the funds they manage into a single filing data set. Therefore, our results speak only to the construction of data indices based on the publicly available data.

 For inquiries about the index product, please contact Barclays Investment Bank at NovusIndex@barclays.com.

This article is provided for information purposes only, does not constitute investment advice, or an offer to buy or sell securities, and should not be deemed to be a recommendation to make any investment. Novus is not an investment adviser, broker dealer, or financial institution and does not offer or provide advice regarding analysis of securities or effecting transactions in securities. 

‍