Big Ideas

Do returns for “smart beta” indexes, constructed to exploit research on one or more factors that predict individual stock returns, reliably predict returns for exchange-traded funds (ETF) introduced to track them? In the June 2020 version of their preliminary paper entitled “The Smart Beta Mirage”, Shiyang Huang, Yang Song and Hong Xiang compare returns of smart beta indexes before and after listings of corresponding smart beta ETFs (see the illustration below). They then explore four potential explanations of differences: (1) offeror timing of ETF introduction based on underlying factor performance, (2) offeror timing of ETF introduction based on underlying index performance, (3) long-term trends in factor premiums and (4) data snooping bias. Using introduction dates for 238 U.S. single-factor and multi-factor equity smart beta ETFs listed between 2000 to 2018 and price data for matched smart beta indexes as available through December 2019, they find that: Keep Reading

Are published studies that predict higher returns for some U.S. stocks and lower for others based on firm accounting, stock trading and other data reproducible? In their May 2020 paper entitled “Open Source Cross-Sectional Asset Pricing”, Andrew Chen and Tom Zimmermann make available data and code that reproduce many published cross-sectional stock return predictors, allowing other researchers to modify and extend past studies. They commit to annual updates of their study. Defining statistical significance as achieving at least 95% confidence in predictive power, they include:

• 180 clear predictors that exhibit statistical significance in original studies and are easily reproducible.
• 30 likely predictors that exhibit statistical significance in original studies but are not precisely reproducible.
• 315 additional predictors covered in past studies that were not clearly tested or failed, or are variations of these predictors. They further extend this group by separately testing 1-month, 3-month, 6-month and 12-month portfolio reformation frequencies (1,260 total tests).

They compute all predictors on a monthly basis and create for each a long-short portfolio based on the specifications and the sample period of its original study. They check predictive power of each using data available at the end of each month to evaluate long-short portfolio returns the next month. They assume a 6-month lag for availability of annual accounting data and a 1-quarter lag for quarterly accounting data. They make no attempt to account for portfolio reformation frictions or to winnow predictors based on similarity. Using data and sample periods for U.S. firms/stocks as specified in original published studies as described above, they find that: Keep Reading

How should investors think about maximum drawdown (MaxDD) as a portfolio/strategy performance metric? In their April 2020 paper entitled “Drawdowns”, Otto Van Hemert, Mark Ganz, Campbell Harvey, Sandy Rattray, Eva Martin and Darrel Yawitch examine usefulness of MaxDD for portfolio/strategy performance evaluation. They first quantify how MaxDD relates to key return statistics based on 100,000 simulations of monthly returns for each variation. They then investigate use of MaxDD for detecting portfolio/strategy failure due to strategy crowding or other market changes. Finally, they assess MaxDD-based rules for portfolio risk reduction. Using pure simulations and simulations based on actual U.S. stock market monthly returns since 1926, they find that: Keep Reading

Is the concept of emergent social behaviors useful in investing and trading? In his January 2020 address to the American Finance Association entitled “Social Transmission Bias in Economics and Finance”, David Hirshleifer discusses social economics and finance, a new field that examines how social processes shape economic and financial behaviors. This field is distinct from: (1) information economics (some people know more than others); and, (2) behavioral finance (people make systematic mistakes). He focuses on social transmission bias, systematic modification of signals or ideas between sender and receiver, as the key element of the new field. He employs five “fables” (models) to illustrate the novelty and importance of such bias. Based on his long experience in behavioral finance and recent/current studies, he concludes that:

Keep Reading

Suppose ten stock market timing strategies out of 10,000 beat the market for ten years running. Are they true outperformers, or just lucky? Multiple hypothesis testing methods address that question by controlling for luck. What are these methods, and how should researchers use them? In their November 2019 paper entitled “An Evaluation of Alternative Multiple Testing Methods for Finance Applications”, Campbell Harvey, Yan Liu and Alessio Saretto:

1. Address the scope of the multiple testing problem in empirical financial economics.
2. Summarize multiple testing methods based on conventional (frequentist) hypothesis testing.
3. Simulate performance of different methods across a variety of testing environments.

Their goal is to provide a menu of choices to help researchers improve inference in financial economics. Based on theory and simulations, they conclude that: Keep Reading

How can investment strategy researchers best address the randomness inherent in market data and the ability of investors/markets to adapt to changing conditions? In his September 2019 paper entitled “Tactical Investment Algorithms”, Marcos Lopez de Prado reviews three methods for testing the performance of an investment rule:

1. Walk-forward (WF) tests a rule against an actual historical data series, implicitly assuming that market behaviors are neither largely random nor changing (that the rule being tested is “all-weather”).
2. Resampling (RS) addresses randomness in market behaviors by assuming that resampling of past observations can usefully generate possible future price paths. 
3. Monte Carlo (MC) addresses both randomness and adaptation in market behaviors by simulating possible future price paths based on models of price generation derived from theory and statistical analysis of actual data.

Based on his knowledge of financial markets and testing methods, he concludes that: Keep Reading

Is there a way that asset managers can share knowledge/data across proprietary boundaries with many researchers to advance development of investment strategies? In their September 2019 paper entitled “Crowdsourced Investment Research through Tournaments”, Marcos Lopez de Prado and Frank Fabozzi describe highly structured tournaments as a crowdsourcing paradigm for investment research. In each such tournament, the organizer poses one investment challenge as a forecasting problem and provides abstracted and obfuscated data. Contestants pay an entry fee, develop models and provide forecasts, retaining model ownership by running calculations on their own hardware/software. Based on this hypothetical tournament setup and their experience, they conclude that:

Keep Reading

The following list links to summaries of recent (since 2010) investment research using long data samples. These summaries may be helpful in developing strategic allocations and tactical wariness for long-horizon investments.

• “Commodity Futures Risk Premium Over the Long Run” – estimates the historical risk premium of dead and live commodity futures contract series as available during 1871 through 2018 (148 years).
• “Bond Returns Over the Very Long Run” – assesses returns to investment grade bonds (federal, state/local and corporate) in the U.S. during 1793 through 2013 (221 years).
• “National Election Cycle and Stocks Over the Long Run” – examines the behavior of the U.S. stock market across the U.S. presidential term cycle (years 1, 2, 3 or 4) from the end of December 1872 through August 2019 (almost 147 years).
• “Stock Market Earnings Yield and Inflation Over the Long Run” – investigates how U.S. stock market earnings yield interacts with the U.S. inflation rate during January 1871 through June 2019 (148 years).
• “‘Sell in May’ Over the Long Run” – tests whether conventional wisdom to “Sell in May” works for the U.S. stock market during April 1871 through April 2019 (148 years).
• “Asset Class Short-term Momentum Over the Long Run” – measures short-term return predictability within global samples for five asset classes (equity indexes, government bonds, treasury bills, commodity futures and currencies) in U.S. dollars as available during 1800 through 2018 (219 years).
• “Commodity Futures Strategies Over the Very Long Run” – evaluates momentum (nearest contract 12-month excess return), value (spot price change from one year ago to five years ago) and basis (12-month average ratio of nearest to next-nearest contract prices) premiums for commodity futures as available during 1877 through 2017 (141 years).
• “Global Factor Premiums Over the Very Long Run” – checks the reliability of six widely accepted factor premiums across four asset classes using data as available during 1800 through 2016 (217 years).
• “January Barometer Over the Long Run” – tests support for the belief that “as goes January, so goes the rest of the year” for the the U.S. stock market during 1871 through 2017 (147 years).
• “Trend Following with Intrinsic Momentum over the Very Long Run” – investigates whether time series (intrinsic or absolute) return momentum works for 67 assets as available during 1880 through 2016 (137 years).
• “(Some) Commodities over the Long Run” – examines whether commodity futures are attractive as available during 1877 through 2015 (139 years).
• “Illiquid Asset Returns over the Long Run” – tests whether illiquid assets (houses, farmland, art, stamps, wine, violins, gold, silver and diamonds) are competitive as investments with liquid financial assets during 1900 through 2014 (115 years).
• “Asset Class Price Momentum Over the Very Long Run” – examines reliability of price momentum within and across country equity markets, global equity sectors, government bonds, commodities, currencies and individual U.S. stocks as available during 1800 through 2014 (215 years).
• “Market Timing with Moving Averages Over the Very Long Run” – explores which moving average rules and measurement (lookback) intervals work best for the U.S. stock market during 1860 through 2009 (150 years).
• “Currency Carry Trade Over the Long Run” – tests whether the currency carry trade (financing short-term deposits in currencies with high interest rates with short-term loans in currencies with low interest rates) generates a reliably attractive return during 1900 through 2012 (113 years).
• “Earnings per Share Growth in the Long Run” – investigates what long-run real return investors should expect from the U.S. stock market and whether popular metrics reliably indicate when it is overvalued using data as available during 1871 through 2013 (143 years).
• “Trend Following over the Very Long Run” – tests whether prices exhibit persistently exploitable trends for commodities, currencies, stock indexes and bonds as available during 1800 through 2010 (211 years).
• “January Effect Over the Long Run” – assesses support for belief in the January effect (exceptionally strong performance by the U.S. stock market during the month of January) during 1871 through 2013 (143 years).
• “Stock Price Momentum Over the Very Long Run” – analyzes momentum in U.S. stock prices during 1800 through December 2012 (213 years).
• “10-Month SMA Timing Signals Over the Long Run” – tests usefulness of a 10-month simple moving average (SMA) trading rule on the U.S. stock market during 1871 through 2012 (142 years).
• “Moving Average Rules Over the Long Run” – examines performances of 10,800 moving average trading rule variants as applied to then-current components of the Dow Jones Industrial Average during October 1928 through December 2011 (83 years).
• “Reversion of Stock Markets to Value Over the Long Run” – investigates whether investors can count on stock markets reverting to some valuation benchmark using annual index levels during 1900 through 2008 (109 years).
• “Trading Frictions Over the Long Run” – compiles a history of U.S. stock trading frictions spanning 1900 through 2000 (101 years).
• “A Long Run Demographic Stock Market Outlook” – assesses whether the demographic mix of a set of U.S. investors affects aggregate demand for yield, and thereby total returns, from equity investments during 1900 through 2008 (109 years).

Another long run source is the annual update of the work summarized in Triumph of the Optimists (Chapter-by-Chapter Review).

Some general cautions regarding such studies are:

• Reconstruction of price series from, for example, old newspapers involves missing data and potentially inconsistent reporting approaches. In other words, quality of old data is suspect.
• The number of asset class price series available may be small in early parts of sample periods.
• Some studies may impound survivorship bias via omission of assets that were important in the past but are no longer tracked in source databases.
• For studies using Shiller data, monthly levels are averages of monthly values, blurring monthly statistics and modestly blurring annual statistics. Results based on end-of-month values may differ.
• Reported returns are gross, not net. Accounting for costs of maintaining a tracking fund for a portfolio/index of commodities would reduce returns. Also:
  • Studies involving shorting (such as factor premium analyses) typically do not address the cost/feasibility of shorting.
  • Costs of maintaining tracking funds may vary by asset class, by country and over time, confounding comparisons. For example, commodity futures indexes generally assume monthly rolling of many contract series.
  • Investment capacities of some assets may be especially limited early in sample periods.
  • Tax consequences of trading vary considerably across countries and over time.
• Historical timeliness of data collection/processing for periodic trading (for example, for portfolio rebalancing) may be especially problematic in early parts of sample periods.
• Economies and markets change over time, making it difficult to assess the relative importance of older versus newer data.
• Distant past availability of retrospectively constructed indexes may have altered contemporaneous market behaviors (induced market adaptations).

How does a shift in emphasis from active to passive investing affect the financial market risk landscape? In their September 2019 paper entitled “The Shift From Active to Passive Investing: Potential Risks to Financial Stability?”, Kenechukwu Anadu, Mathias Kruttli, Patrick McCabe, Emilio Osambela and Chaehee Shin analyze how a shift from active to passive investing affects:

1. Investment fund redemption liquidity risks.
2. Market volatility.
3. Asset management industry concentration.
4. Co-movement of asset returns and liquidity.

They also assess how effects are likely to evolve if the active-to-passive shift continues. Based on their framework/analysis, they conclude that: Keep Reading

Where can investors find a large catalog of trading strategies spanning asset classes? In the September 2019 update of their paper entitled “151 Trading Strategies”, Zura Kakushadze and Juan Andres Serur make freely available their book describing over 150 trading strategies, including over 550 formulas. These strategies encompass many asset classes and trading styles, including: stocks, options, fixed income, futures, exchange-traded funds, commodities, foreign exchange, convertibles, structured assets (collateralized debt and mortgage-backed), volatility, real estate, distressed, crypto, weather, energy, inflation, global macro, infrastructure and tax arbitrage. Some strategies incorporate machine learning. The paper does not include empirical studies, numerical simulations or backtests, but does contain annotated source code for out-of-sample testing, references and a glossary of acronyms and mathematical terms. Based on experience, they observe that: Keep Reading