Objective research to aid investing decisions

Value Investing Strategy (Strategy Overview)

Allocations for March 2024 (Final)
Cash TLT LQD SPY

Momentum Investing Strategy (Strategy Overview)

Allocations for March 2024 (Final)
1st ETF 2nd ETF 3rd ETF

Survey of Recent Research on Factors, Regimes and Robustness

| | Posted in: Big Ideas, Strategic Allocation

Why and how should investors pursue investment premiums associated with factors that explain performance differences among related assets (like common stocks)? In the January 2015 version of his paper entitled “Better Investing Through Factors, Regimes and Sensitivity Analysis”, Cristian Homescu summarizes recent research on: (1) factor-based investing; (2) enhancement of factor-based investing via regime switching models; and, (3) strategy robustness testing. Factor investing means systematic targeting of premiums associated with factors that explain an exploitable portion of return and risk differences among securities within one or several asset classes. Based on recent streams of research, he concludes that:

  • Candidate factors should have strong research support and economic justification, significant and persistent premiums, return histories that include bad times and exploitability via liquid assets.
    • Some factors (value, momentum, carry) cut across asset classes, while others are class-specific.
    • Different factors work at different times, such that multi-factor portfolios can generate smooth performance while component factors still capture their respective premiums on average.
    • Multi-factor portfolios are likely to reduce portfolio turnover compared to separate factor portfolios, because some trades cancel.
    • Combining multi-factor investing with risk budgeting achieves diversification efficiently via focus on sources of risk (the factors themselves).
    • Portfolio factor weights and rebalancing rules should consider the trade-offs between portfolio agility and both liquidity and turnover (implementation costs).
  • Regime switching models quantify the tendency of financial markets to change abruptly from one persistent regime to another (for example, between low-volatility and high-volatility).
    • While identified analytically, regime changes often match shifts in regulation, policy and other secular changes.
    • Statistics defining regimes typically include average returns, volatilities, autocorrelations and cross-correlations for a set of assets. One can also define regimes directly with factor behaviors for use in timing factor models.
    • Regime switching models appear to work with all asset classes.
    • Allocation strategies based on regime switching should consider the trade-off between number of regimes (model precision) and implementation costs. 
  • Sensitivity (robustness) testing assesses which variables dominate optimization and evaluates basic sensitivities to parameter settings. Integrating sensitivity testing within optimization procedures is highly beneficial.
    • Backtesting is useful but subject to manipulation/self-deception. Realistic backtests exhibit consistent in-sample and out-of-sample performance, subject to out-of-sample data adequacy and purity (truly unknown and not iterated).
    • Testing more and more strategy variations on the same data increases the probability of choosing a bad variation. Moreover, researchers conducting sensitivity tests on the same data tend to publish only the good results, thereby presenting a biased sample of outcomes.
    • Stress tests should include both the largest observed shocks within a past sample (such as September 2008) and scenarios imagining extreme events that might happen.

In summary, investors may want to consider multi-factor portfolios, with timing enhanced by regime-switching models and confidence boosted by robust testing, as a means of capturing widely accepted risk premiums efficiently and smoothly.

Cautions regarding conclusions include:

  • Many methods covered are likely too complex for most investors. Delegating factor portfolio construction, testing and maintenance to others means management fees.
  • The study offers no specific strategies for consideration.

Some key citations on testing/simulation in this paper are summarized in:

 

 

Login
Daily Email Updates
Filter Research
  • Research Categories (select one or more)