德银-全球-量化策略-2019年全球量化策略会议的收获-2019.6.17-24页_583kb

2019 Deutsche Bank Global Quantitative Strategy Conference Summary

Core Content

The 2019 Deutsche Bank Global Quantitative Strategy Conference, held in New York on May 9th, brought together over 250 buy-side investment professionals. The conference focused on bridging the gap between academic research and practical applications in quantitative investing. It featured nine presentations from leading academics and industry experts, covering topics such as machine learning, portfolio construction, risk analysis, and traditional factor research. The goal was to provide actionable insights and innovative strategies that can be implemented in real-world investing scenarios.

Main Presentations and Key Ideas

1. World of Quants: The Scientific Approach towards Investing

• Presenter: Nitish Maini, General Manager, Virtual Research Center
• Key Points:
  • WorldQuant uses a collaborative model between portfolio managers and researchers to generate positive risk-adjusted returns.
  • The "triple axis" diversification strategy uses ideas, regions, and performance parameters to find untouched sources of alpha.
  • The backtesting architecture is designed to test new strategies efficiently and effectively.
  • Human input is essential for generating ideas and interpreting results, while machines provide speed, scale, and accuracy.

2. What Happens When the World Goes Passive?

• Presenter: Dr. Ronnie Shah, Head of US Quantitative Strategy
• Key Points:
  • Passive assets have grown at an annualized rate of 25% and now account for over 20% of equity ownership.
  • ETF flows (inflows and outflows) are a strong predictor of future stock returns.
  • A simple long/short ETF flow strategy generates a Sharpe ratio of 1.3 over the past 12 years.
  • ETF flow is largely uncorrelated with traditional quant factors like Value, Momentum, and Low Vol.
  • Index reconstitution is the main driver of ETF flow returns.

3. Fundamental Analysis Redux

• Presenter: Prof. Richard Sloan, University of Southern California
• Key Points:
  • Fundamental analysis remains crucial for capital markets and resource allocation.
  • Formulaic value investing has failed to consistently outperform due to the misinterpretation of fundamental-to-price ratios.
  • Multi-factor strategies work well in-sample but underperform out-of-sample.
  • Case studies show how fundamental analysis can uncover issues missed by formulaic approaches.

4. Trading Without Regret Using Machine Learning

• Presenter: Prof. Michael Kearns, University of Pennsylvania
• Key Points:
  • A "no-regret" learning framework is a simple model that doesn't require statistical assumptions.
  • It involves maintaining a dynamically weighted portfolio through trading bots.
  • The framework aims to approximate the best possible signal in hindsight.
  • High number of signals and sideways markets can lead to portfolio concentration and higher risk.
  • Prof. Kearns suggests incorporating risk metrics like daily PNL to manage exposure.

5. Hedging Risk Factors

• Presenter: Prof. Bernard Herskovic, UCLA
• Key Points:
  • Standard risk factors can be hedged at low or no cost.
  • Hedging macro factors like industrial production and unemployment reduces exposure to bad times.
  • Hedge portfolios perform well during downturns and have low betas on key factors.
  • Minimum variance portfolios that avoid expected returns can produce large alphas.

6. A Fundamental Factor Model

• Presenter: Prof. Stephen Penman, Columbia University
• Key Points:
  • A new fundamental pricing factor is constructed based on accounting information and consumption-based theory.
  • The factor (ER) has low correlation with the market and hedges against bad times.
  • It captures most of the returns explained by existing models in a single factor.
  • The model includes Earnings-to-Price, Growth in Net Operating Assets, and Accruals.

7. Earnings Quality on the Street

• Presenter: Prof. Shiva Rajgopal, Columbia University
• Key Points:
  • An analysis of a research firm's earnings quality data was conducted, identifying 121 metrics to detect aggressive accounting practices.
  • Companies flagged by the RF tend to be larger, have lower arbitrage barriers, and higher M-score and F-score.
  • The RFSCORE metric was developed to improve earnings quality modeling.
  • The ROC analysis showed the RFSCORE has predictive power, though the dataset was limited and potentially subject to data-snooping bias.

8. Explanation for the Dispersion Anomaly

• Presenter: Prof. Paul Irvine, Texas Christian University
• Key Points:
  • The dispersion anomaly refers to the lower returns of stocks with high analyst forecast dispersion.
  • Analysts are more optimistic in high-dispersion groups, leading to negative forecast revisions.
  • This optimism results in negative cash flow shocks, explaining the dispersion anomaly.
  • Removing forecast bias from dispersion results in a positive predictive measure for returns.

9. Factor Investing in Credit Markets

• Presenter: Jose Gonzalez, Deutsche Bank
• Key Points:
  • Factor investing in credit markets is under-researched due to high transaction costs and asset diversity.
  • The CDS market provides a standardized and flexible way to analyze credit risk.
  • Six investable factors are identified: Quality, Low Duration, Value, Low Beta, and two types of Momentum.
  • Factor strategies are aggregated to reduce turnover and improve performance.
  • The multifactor portfolio is used as an overlay to a corporate bond benchmark.

Key Information

• The conference emphasized the importance of combining academic research with practical applications.
• ETF flow is a powerful signal for stock returns and is largely uncorrelated with traditional quant factors.
• Fundamental analysis is still vital for understanding company-specific risks and opportunities.
• Machine learning offers a no-regret framework for portfolio construction, though challenges like concentration risk exist.
• Hedging strategies can reduce exposure to macro and credit risk factors without sacrificing returns.
• The dispersion anomaly is explained by analyst optimism and subsequent forecast revisions.
• Factor investing in credit markets is a promising frontier, but transaction costs and complexity remain hurdles.

Conclusion

The 2019 Deutsche Bank Global Quantitative Strategy Conference provided a comprehensive overview of current trends and innovations in quantitative investing. It highlighted the importance of diversification, the role of machine learning in portfolio management, the need for fundamental analysis, and the potential of factor investing in both equity and credit markets. The insights presented offer actionable strategies for investors seeking to improve risk-adjusted returns in a rapidly evolving financial landscape.