20160914-法国巴黎银行-Introducing_MarFA__BNP_Paribas_Market_Factor_Analysis_13页_967kb

MarFA™: BNP Paribas Market Factor Analysis Summary

Core Content

BNP Paribas has introduced MarFA™ (Market Factor Analysis), a new cross-asset model that identifies correlations between assets and the main market and macroeconomic factors driving them. The model provides two strategies for generating trading signals: MarFA™ Trading (short-term signals lasting 1-2 weeks) and MarFA™ Macro (longer-term signals lasting 1-2 months).

MarFA™ monitors 75 assets across equities, fixed income, credit, FX, and commodities using Principal Component Analysis (PCA). It calculates the correlation between assets based on the last three months of data and identifies the Market Factors and Macro Factors that drive these assets.

Main Points

• Market Factors:

  • The 1st Market Factor is primarily driven by low implied volatilities in equities and FX, and negative US real yields.
  • The 2nd Market Factor is associated with longer-term real yields.
  • The 3rd Market Factor is influenced by USD performance and xccy funding.

• Macro Factors:

  • The 1st Macro Factor reflects declining long-term inflation expectations.
  • The 2nd and 3rd Macro Factors are linked to economic activity such as data surprises, economic surveys, and employment data.

• Trading Signals:

  • MarFA™ Trading indicates that Japanese, French, and Canadian yield curves have steepened too far, while high yield credit spreads and Australian equities are too low.
  • MarFA™ Macro highlights that Chinese equities are overvalued, Mexican equities are undervalued, and French, German, and Japanese 2s10s curves have steepened too far.

• Fair Value Estimation:

  • The model calculates fair values for assets based on Market and Macro Factors.
  • Assets that deviate from their fair values by 1-2 standard deviations are likely to reverse, providing trading opportunities.

Key Information

• Asset Correlations:

  • The 1st Market Factor explains 80-95% of the variation in equity and credit indices.
  • Fixed income and FX assets show weaker correlations with the 1st Market Factor.

• Z-Score Deviations:

  • The most overvalued assets (according to MarFA™ Trading and Macro) include JP 2s10s, USDCAD, EURSEK, France 2s10s, and MSCI China2.
  • The most undervalued assets include EM credit, spread, UK HY, spread, MSCI Australia, GBPUSD, and MSCI Mexico2.

• Backtesting Results:

  • MarFA™ Trading has a Sharpe ratio of 2.6 over the period from June 2010 to August 2016.
  • MarFA™ Macro has a Sharpe ratio of 0.9.
  • The strategies perform best for equities and fixed income, with hit ratios generally above 50%.
  • The average trade duration is 1-2 weeks for MarFA™ Trading and 1-2 months for MarFA™ Macro.

Asset Class Performance

• Equities:

  • Sharpe ratio: 1.6 for MarFA™ Trading, 0.8 for MarFA™ Macro.
  • Most overvalued: MSCI China2 (2.1 z-scores), JP 2s10s (3.6 z-scores).
  • Most undervalued: MSCI Australia (3.3 z-scores), GBPUSD (3.5 z-scores).

• Fixed Income:

  • Sharpe ratio: 2.1 for MarFA™ Trading, 0.7 for MarFA™ Macro.
  • Most overvalued: US 10yr yield (2.59 Sharpe ratio), Japan 10yr yield (1.73 Sharpe ratio).
  • Most undervalued: Canada 10yr yield (0.74 Sharpe ratio), France 10yr yield (0.57 Sharpe ratio).

• Credit:

  • Sharpe ratio: 0.65 for MarFA™ Trading, -0.8 for MarFA™ Macro.
  • Most overvalued: EZ HY, spread (4.4 z-scores).
  • Most undervalued: UK HY, spread (3.2 z-scores), EM credit, spread (2.8 z-scores).

• FX:

  • Sharpe ratio: 0.5 for MarFA™ Trading, 0.0 for MarFA™ Macro.
  • Most overvalued: USDCAD (3.0 z-scores).
  • Most undervalued: USDZAR (2.4 z-scores), EURSEK (2.6 z-scores).

• Commodities:

  • Sharpe ratio: 0.2 for MarFA™ Trading, 0.1 for MarFA™ Macro.
  • Most overvalued: Gold (3.6 z-scores).
  • Most undervalued: Copper (3.0 z-scores).

Summary of Key Findings

• MarFA™ is a quantitative tool that helps identify market and macroeconomic factors influencing asset prices.
• It provides trading signals based on the deviation of assets from their fair values.
• The model uses PCA to identify the most significant factors driving the market.
• The backtesting suggests that MarFA™ has predictive power, with positive returns and high hit ratios.
• The Z-scores indicate overvaluation and undervaluation of specific assets, which can be used to formulate trading decisions.

Conclusion

MarFA™ is a robust framework for cross-asset analysis, offering insights into market and macroeconomic drivers and generating trading signals for both short and long-term strategies. It provides a systematic approach to identifying mispriced assets and has shown positive backtesting results, indicating its potential for profitable trading.