Alpha Cloning - Following 13F Fillings

Out-of-sample strategy implementation in QuantConnect (chart, statistics & code)

Source paper

Cohen, Folk, Silli: Best Ideas

Abstract: We examine the performance of stocks that represent managers' "Best Ideas." We find that the stock that active managers display the most conviction towards ex-ante, outperforms the market, as well as the other stocks in those managers' portfolios, by approximately 1.6 to 2.1 percent per quarter depending on the benchmark employed. The results for managers' other high-conviction investments (e.g. top five stocks) are also strong. The other stocks managers hold do not exhibit significant outperformance. This leads us to two conclusions. First, the U.S. stock market does not appear to be efficiently priced by our risk models, since even the typical active mutual fund manager is able to identify stocks that outperform by economically and statistically large amounts. Second, consistent with the view of Berk and Green (2004), the organization of the money management industry appears to make it optimal for managers to introduce stocks into their portfolio that are not outperformers. We argue that investors would benefit if managers held more concentrated portfolios.

Other papers

• Myers, Poterba, Shackeford, Shoven: Copycat Funds: Information Disclosure Regulation and the Returns to Active Management in the Mutual Fund Industry

  Abstract: Mutual funds must disclose their portfolio holdings to investors semiannually. The costs and benefits of such disclosures are a long-standing subject of debate. For actively managed funds, one cost of disclosure is a potential reduction in the private benefits from research on asset values. Disclosure provides public access to information on the assets that the fund manager views as undervalued. This paper tries to quantify this potential cost of disclosure by testing whether "copycat" mutual funds, funds that purchase the same assets as actively-managed funds as soon as those asset holdings are disclosed, can earn returns that are similar to those of the actively-managed funds. Copycat funds do not incur the research expenses associated with the actively-managed funds that they are mimicking, but they miss the opportunity to invest in assets that managers identify as positive return opportunities between disclosure dates. Our results for a limited sample of high expense funds in the 1990s suggest that while returns before expenses are significantly higher for the underlying actively managed funds relative to the copycat funds, after expenses copycat funds earn statistically indistinguishable, and possibly higher, returns than the underlying actively managed funds. These findings contribute to the policy debate on the optimal level and frequency of fund disclosure.

• Wermers: Is Money Really 'Smart'? New Evidence on the Relation Between Mutual Fund Flows, Manager Behavior, and Performance Persistence

  Abstract: Mutual fund returns strongly persist over multi-year periods - that is the central finding of this paper. Further, consumer and fund manager behavior both play a large role in explaining these long-term continuation patterns - consumers invest heavily in last-year's winning funds, and managers of these winners invest these inflows in momentum stocks to continue to outperform other funds for at least two years following the ranking year. By contrast, managers of losing funds appear reluctant to sell their losing stocks to finance the purchase of new momentum stocks, perhaps due to a disposition effect. Thus, momentum continues to separate winning from losing managers for a much longer period than indicated by prior studies. Even more surprising is that persistence in winning fund returns is not entirely explained by momentum - we find strong evidence that flow-related buying, especially among growth-oriented funds, pushes up stock prices. Specifically, stocks that winning funds purchase in response to persistent flows have returns that beat their size, book-to-market, and momentum benchmarks by two to three percent per year over a four-year period. Cross-sectional regressions indicate that these abnormal returns are strongly related to fund inflows, but not to the past performance of the funds - thus, casting some doubt on prior findings of persistent manager talent in picking stocks. Finally, at the style-adjusted net returns level, we find no persistence, consistent with the results of prior studies. On balance, we confirm that money is smart in chasing winning managers, but that a "copycat" strategy of mimicking winning fund stock trades to take advantage of flow-related returns appears to be the smartest strategy.

• Verbeek, Wang: Better than the Original? The Relative Success of Copycat Funds

  Abstract: We construct hypothetical copycat funds to investigate the performance of free-riding strategies that duplicate the disclosed asset holdings of actively managed mutual funds. On average, copycat funds are able to marginally outperform their target mutual funds, but their relative success has increased after 2004, when SEC regulation imposed all mutual funds to quarterly disclose their holdings. We find a substantial cross-sectional dispersion in the relative performance of copycat funds. Free-riding on the portfolios disclosed by past winning funds and the funds that disclose representative holdings generates significantly better performance net of trading costs and expenses than the vast majority of mutual funds. The results indicate that free-riding on disclosed fund holdings is an attractive strategy and suggest that mutual funds can suffer from information disclosure requirements.

• Amir-Ghassemi, Papanicolaou, Perlow: Aggregate Alpha in the Hedge Fund Industry: A Further Look at Best Ideas

  Abstract: Our paper addresses the portfolio selection of hedge fund firms as a measure of abnormal skill. It further decomposes this skill through the lens of canonical `Best Ideas'. Both are achieved through the application of regulatory mandated position-level data from the SEC, colloquially known as 13F data. The approach aims to reduce common biases associated with traditional return database analysis while unlocking position-level portfolio analysis. Across a composite of hedge fund managers and twenty years of analysis, we find historically abnormal excess return associated with their security selection. However, there has been a significant decline in this abnormal return after the 2008 financial crisis. We examine this out-performance through portions of each manager’s portfolio, using ex ante methodologies to elicit Best Ideas. We find no significant difference in return characteristics between these Best Ideas relative to the aggregate portfolio positions of these hedge fund managers. These findings are broadly in contrast to similar studies conducted on mutual funds, demonstrating differences in portfolio behavior across the two classes of fund managers.

• Fleiss, Alexander and Kumaar, Amrith and Rida, Adam and Shin, Junsup and Lai, Xinying and Fang, Vivian and Chen, Jialiang and Li, Ang: Deep Reinforcement Learning & Feature Extraction For Constructing Alpha Generating Equity Portfolios

  Abstract: The ambition of this paper is to catch hidden information inside the Securities and Exchange Commission’s (SEC)13F public holding data in order to construct an equity portfolio that maximizes returns. The 13F ling data give us the quarterly stock trading decisions of included funds, but we’re not given any insight on how they made their decisions or if information has been shared between funds. To remedy this lack of knowledge, this paper used feature extraction in order to lter out the best performing funds through several criteria. We propose a method employing powerful machine learning techniques (Deep Reinforcement Learning) to try to catch the missing pieces of information behind the decision process and use them as a prediction tool to construct quarterly equity portfolios. This approach reached an annualized return of 21% with a sharpe ratio of 1.8 outperforming the S&P 500 both in returns and stability through historical backtesting.

• Schroeder, J and Posch, Peter N.: Outperforming the Market: Portfolio Strategy Cloning from SEC 13F Filings

  Abstract: Can mirroring the investment strategies of institutional managers lead to market-outperforming returns? Our findings demonstrate that cloned portfolios in the top quartile, derived from SEC EDGAR Form 13F filings, replicate the funds' performances and exceed the SP500 index by 24.25% on an annualized risk-adjusted basis. Analyzing over 150,000 portfolios between 2013 and 2023, we compare original versus replicated strategies across 12 metrics, such as Alpha, Sharpe and Sortino ratios, various return rates, annualized volatility, and maximum drawdown. Through Wilcoxon signed-rank tests applied to delta distributions, we reject the null hypothesis across all metrics, except for annualized volatility, maximum drawdown and tracking error, and demonstrate that cloned portfolios balanced on the disclosure date of filings (rather than quarter-end), successfully mirror the performance of the original funds, including both market-underperforming and -outperforming funds.