A cura di Tommi Johnsen, Alpha Architect
- Daniel Giamouridis (Co-editor of FAJ)
- Financial Analysts Journal
- A version of this paper can be found here
- Want to read our summaries of academic finance papers? Check out our Academic Research Insight category
What are the research questions?
- How will systematic (“coordinated”) investing affect prices?
- What is the risk of increasingly coordinated holdings (crowding)?
- How does coordinated investing affect market microstructure and optimal execution?
- Is factor timing possible?
- What are the issues in the design of factor-based strategies?
- What is the role of data science and machine learning?
What are the Academic Insights?
- The author defines “coordinated investing” as buying stocks in “baskets.” For example, when an investment in an ETF is made, the ETF sponsor buys the entire basket of constituent stocks at the same time. Theoretical predictions have yet to be extensively tested with data, while the empirical evidence that is published on this question is mixed. While ETFs facilitate efficient pricing, it has been established that market prices have become noisier. Higher levels of systematic trading risk in ETF constituent stocks compared to non-constituent stocks have been documented. These observations have implications for diversification, trading and cross-asset dependence of impact cost and stock selection (alpha).
- There are too many investors and strategies chasing too few factors (remember August 2007?). An accurate measure of crowding is needed. The measure should include bottom-up measures of constituent liquidity, market impact, and should take into account the tendency that investors have to “time” factors.
- Systematic strategies, by design, have a natural order flow, leading to coordinated portfolio trade lists. The execution of these trade lists increases covariances and correlations of intraday returns and volume, both contributing to variability of observed execution costs.
- Academic evidence suggests that reducing tracking error (active risk) during times of high volatility in the market is an effective strategy for avoiding “bad” times in factor investing. Increasing interest from practitioners on this topic should spur research on the market positioning of factor portfolios, identifying factor exposures to macro risks, and factor risk concentration in portfolios.
- Questions remain as to how a specific factor product performs relative to its stated objective. What is the specific factor strategy’s return performance relative to the actual factor premium? How is the actual factor premium to be measured? As an example, do all Value products perform equally at capturing the Value premium? How is the Value premium defined in an applied setting?
- The issue of whether or not “machine learning” is actually a new paradigm for systematic strategies has not been determined. It is undeniable that quants have always used “big data” and statistical methods which are the commonly accepted hallmarks of machine learning. Can other machine learning approaches such as pattern recognition, outperform the traditional application of linear regression to factor investing and trading? The empirical evidence with respect to the application of machine learning to all aspects of investing: alpha, beta, risk management, trading and execution is yet to be explored.
Continua a leggere su Alpha Architect