Investing Expertise

Can individual investors make a living by day trading? In the June 2020 update of their paper entitled “Day Trading for a Living?”, Fernando Chague, Rodrigo De-Losso and Bruno Giovannetti analyze performances of all Brazilian retail investors who begin trading futures on the main Brazilian stock index during 2013 through 2015 and persist in this trading for at least 300 sessions. They use data for 2012 to identify those who begin trading in 2013, and they use data for 2016-2017 to extend performance evaluations for at least two years of trading. They consider performance both gross and net of exchange and brokerage fees, but they ignore income taxes and expenses such as courses and trading platforms. They employ subsamples and regressions to measure learning while trading. Using trading records for the specified index futures for 19,646 individuals as described during 2012 through 2017, they find that: Keep Reading

Will machine learning revolutionize asset management? In their January 2020 paper entitled “Can Machines ‘Learn’ Finance?”, Ronen Israel, Bryan Kelly and Tobias Moskowitz identify and discuss unique challenges in applying machine learning to asset return prediction, with the goal of setting realistic expectations for how much machine learning can improve asset management. Based on general characteristics of financial markets and machine learning algorithms, they conclude that: Keep Reading

Can machine learning (ML) algorithms improve stock picking? In the May 2020 version of their paper entitled “Stock Picking with Machine Learning”, Dominik Wolff and Fabian Echterling apply ML to insights from financial research to assess stock picking abilities of different ML algorithms at a weekly horizon. Their potential return predictor inputs include equity factors (size, value/growth, quality, profitability and  investment), additional firm fundamentals, and technical indicators (moving averages, momentum, stock betas and volatilities, relative strength indicators and trading volumes). Their ML algorithms include Deep Neural Networks, Long Short-Term Neural Networks, Random Forest, Boosting and Regularized Logistic Regression. They apply these algorithms separately and in combination (by averaging individual predictions) to historical S&P 500 constituents. They test a long-only strategy that each week holds the equal-weighted 50, 100 or 200 stocks with the highest return predictions. Their benchmark is an equal-weighted portfolio of all S&P 500 stocks. They assume a 3-month lag for all fundamental data to avoid look-ahead bias. Using Wednesday (or next trading day if the market is not open on Wednesday) open prices and fundamental data for the historical components of the S&P 500 during January 1999 through December 2019 (1,164 total stocks), they find that: Keep Reading

Robo-advisors provide investors automated financial advice with varying levels of sophistication and degrees of individual tailoring. In their December 2019 book chapter entitled “Robo-advising”, Francesco D’Acunto and Alberto Rossi catalog the main features of robo-advising with respect to personalization, discretion, involvement and human interaction. They consider robo-advisors designed to assist short-term and medium-term (active) trading and those designed to guide long-term (passive) investment/accumulation for retirement. They review prior research on effects of robo-advisors regarding investment choices and performance. Based on the body of information on robo-advising, they conclude that: Keep Reading

Are robo-analysts, who apply technology to mass-produce recommendations with limited human intervention, better stock pickers than traditional human analysts? In their January 2020 preliminary (and incomplete) paper entitled “Man Versus Machine: A Comparison of Robo-Analyst and Traditional Research Analyst Investment Recommendations”, Braiden Coleman, Kenneth Merkley and Joseph Pacelli compare distribution, revision frequency and performance for stock recommendations from robo-analysts versus traditional analysts. In measuring performance, they consider 3-factor (adjusting for market, size and book-to-market factors) and 5-factor (additionally adjusting for profitability and investment factors) alphas of daily rebalanced portfolios of buy or sell recommendations, with a lag of one day between recommendations and trades. Using 134,781 reports issued by seven prominent Robo-Analyst firms and by traditional analysts for 1,002 stocks covered by at least three analysts for at least five years during 2003 through 2018, they find that:

Keep Reading

Are there any investors who have compellingly beaten the market? In his December 2019 paper entitled “Medallion Fund: The Ultimate Counterexample?”, Bradford Cornell reviews performance of the Medallion Fund from Renaissance Technologies as a clear refutation of market efficiency. He focuses on gross returns (including portfolio trading frictions, but not management fees), because they reflect the value added by the fund manager. Using Medallion Fund returns from its inception in 1988 through 2018, he finds that: Keep Reading

Do differences in expectations between institutional and individual investors in stocks and bonds, as quantified in weekly legacy Commitments of Traders (COT) reports, offer exploitable timing signals? In the February 2019 revision of his paper entitled “Want Smart Beta? Follow the Smart Money: Market and Factor Timing Using Relative Sentiment”, flagged by a subscriber, Raymond Micaletti tests a U.S. stock market-U.S. bond market timing strategy based on an indicator derived from aggregate equity and Treasuries positions of institutional investors (COT Commercials) relative to individual investors (COT Non-reportables). This Smart Money Indicator (SMI) has three relative sentiment components, each quantified weekly based on differences in z-scores between standalone institutional and individual net COT positions, with z-scores calculated over a specified lookback interval:

1. Maximum weekly relative sentiment for the S&P 500 Index over a second specified lookback interval.
2. Negative weekly minimum relative sentiment in the 30-Year U.S. Treasury bond over this second lookback interval.
3. Difference between weekly maximum relative sentiments in the 10-Year U.S. Treasury note and 30-year U.S. Treasury bond over this second lookback interval.

Final SMI is the sum of these components minus median SMI over the second specified lookback interval. He considers z-score calculation lookback intervals of 39, 52, 65, 78, 91 and 104 weeks and maximum/minimum relative sentiment lookback intervals of one to 13 weeks (78 lookback interval combinations). For baseline results, he splices futures-only COT data through March 14, 1995 with futures-and-options COT starting March 21, 1995. To account for changing COT reporting delays, he imposes a baseline one-week lag for using COT data in predictions. He focuses on the ability of SMI to predict the market factor, but also looks at its ability to enhance: (1) intrinsic (time series or absolute) market factor momentum; and, (2) returns for size, value, momentum, profitability, investment, long-term reversion, short-term reversal, low volatility and quality equity factors. Finally, he compares to several benchmarks the performance of an implementable strategy that invests in the broad U.S. stock market (U.S. Aggregate Bond Total Return Index) when a group of SMI substrategies “vote” positively (negatively). Using weekly legacy COT reports and daily returns for the specified factors/indexes during October 1992 through December 2017, he finds that: Keep Reading

Can investors exploit information about hedge fund stock holdings in SEC Form 13F filings? In their October 2019 paper entitled “Systematic 13F Hedge Fund Alpha”, Mobeen Iqbal, Farouk Jivraj and Luca Angelini investigate whether carefully culled “best ideas” of equity hedge funds produce significantly beat the S&P 500 Total Return (TR) Index. Using quarterly Form 13Fs for U.S. equity long-short, equity market neutral, equity long-only and equity event-driven hedge funds, they measure: individual hedge fund manager conviction regarding a stock based on size of position; and, hedge fund manager consensus regarding a stock based on the number of funds holding it. Using proprietary data, they identify hedge funds exhibiting long-term investment approaches. They then 47 days after the end of each quarter (to ensure availability of Form 13Fs), reform a portfolio from among long-term hedge funds holding at least five stocks, as follows:

1. Exploit conviction by identifying all stocks comprising at least 7.5% of a fund portfolio.
2. Exploit consensus by buying the equal-weighted top 50 of these stocks in terms of number of hedge managers holding them. 

Using processed quarterly data from hedge fund Form 13Fs, the specified proprietary data on hedge fund investment approaches and returns for associated stocks during the first quarter of 2004 through the second quarter of 2019, they find that:

Keep Reading

Who beats the stock market and why? In his October 2019 paper entitled “The Five Investor Camps That Try to Beat the Stock Market”, William Ziemba discusses how different categories of investors succeed. For investors pursuing active strategies, he addresses broadly the means of getting an edge and betting well. Based on his academic work and practical experience, he concludes that: Keep Reading

What are asset class return expectations and associated portfolio allocations of very sophisticated U.S. investors? In their February 2019 paper entitled “The Return Expectations of Institutional Investors”, Aleksandar Andonov and Joshua Rauh analyze disclosures of expected returns across asset classes among U.S. public pension funds, which hold assets of about $4 trillion (see the first chart below), including fixed income, cash, equities, real assets, hedge funds, private equity and other asset classes. Taking into account past fund performance, they investigate how fund managers estimate future returns. Disclosures also reveal target allocations to asset classes (see the second chart below). Together, expected asset class returns and target allocations allow calculation of expected portfolio returns. Using annual disclosures for 228 U.S. state and local government pension plans during 2014 through 2017, they find that:

Keep Reading