Overnight Returns as a Market Timing Strategy
Beat the market simply by holding the positions overnight?
TLDR
This paper investigates the non-linear relationship between overnight and daytime returns in equity markets, focusing on the market timing strategies of close-to-open (CO) and open-to-close (OC) returns. Key findings include:
• Overnight Returns: Risk-adjusted overnight returns significantly exceed daytime returns, with CO returns showing a positive selectivity alpha, especially during down markets.
• Daytime Returns: OC returns exhibit negative selectivity alpha and positive market timing ability, contrasting with CO returns.
• Market Timing: The CO strategy demonstrates negative market timing ability, while the OC strategy shows positive market timing.
• Down Market Alpha: Alpha is significant during periods of negative realized market risk premium, highlighting the importance of market conditions in return dynamics.
Introduction
Background
The paper builds on prior research by Kelly and Clark (2011) and Cliff et al. (2008), who observed that nearly all equity market returns since 1993 have occurred overnight. Close-to-close (CC) and close-to-open (CO) returns are nearly equal, while open-to-close (OC) returns are close to zero. This phenomenon is illustrated in Figure 1 for the SPDR S&P 500 ETF and the Vanguard Total Market ETF.
Key Questions
The paper addresses several critical questions:
• How do risk-adjusted overnight returns compare to daytime returns?
• What is the role of market timing strategies in generating CO and OC returns?
• How do non-linear factors influence the relationship between market returns and timing strategy returns?
• What is the significance of alpha during down market periods?
Methodology
Market Timing vs. Stock Selectivity
The paper distinguishes between market timing and stock selectivity, drawing on the work of Treynor and Mazuy (1966) and Henriksson and Merton (1981). Market timing involves predicting the direction of the market, while stock selectivity focuses on individual stock performance.
Non-Parametric Test
The Henriksson-Merton (HM) non-parametric test is used to evaluate market timing ability. The test compares the number of times a strategy return is positive when the market risk premium is non-positive ((n_1)) and negative when the market risk premium is positive ((n_2)).
Parametric Regressions
The paper employs parametric regressions to test for non-linear factors in CO and OC returns. The HM measure, (\mathrm{yt}), is used to capture the non-linearity introduced by market timing strategies. Additionally, the Goetzmann factor (PTMKT and PTSPY) is incorporated to account for the periodicity of market timer forecasts.
Data
ETFs Analyzed
The analysis focuses on four major exchange-traded funds (ETFs):
• SPDR S&P 500 ETF (SPY): Represents large-cap U.S. stocks.
• Nasdaq-100 ETF (QQQ): Tracks the technology-heavy Nasdaq 100 index.
• iShares Russell 2000 ETF (IWM): Represents small-cap U.S. stocks.
• Vanguard Total Market ETF (VTI): Tracks the overall U.S. equity market.
Return Calculations
• Close-to-Close (CC) Returns: Calculated as ((Close_t + Div_t) / Close_{t-1} - 1).
• Open-to-Close (OC) Returns: Calculated as (Close_t / Open_t - 1).
• Close-to-Open (CO) Returns: Calculated as ((1 + CCreturn) / (1 + OCreturn) - 1).
Risk Adjustment
The risk-free rate is subtracted from CC and CO returns to create risk premium series. OC returns are not adjusted for the risk-free rate since positions are settled on the same business day.
Results
Henriksson-Merton Non-Parametric Test
The non-parametric test reveals significant market timing ability for CO returns, with positive timing indicated by negative z-statistics. Conversely, OC returns show negative market timing, with positive z-statistics.
Parametric Regressions
• CO Returns: The CO strategy exhibits negative market timing ability and a positive selectivity alpha, particularly during down markets.
• OC Returns: The OC strategy shows positive market timing ability and a negative selectivity alpha, especially in down markets.
• Down Market Alpha: Alpha is significant during periods of negative realized market risk premium, highlighting the importance of market conditions in return dynamics.
Half-Day Returns
When analyzing half-day returns, the CO strategy shows negative timing and positive alpha in up markets, while the OC strategy mirrors these results.
Discussion
Market Timing and Selectivity
The paper highlights the non-linear relationship between market returns and timing strategy returns. The CO strategy’s negative market timing ability contrasts with its positive selectivity alpha, while the OC strategy shows the opposite pattern.
Down Market Alpha
The significance of alpha during down markets underscores the importance of market conditions in determining returns. The CO strategy’s positive alpha in down markets provides downside protection, while the OC strategy’s negative alpha exacerbates losses.
Implications for Investors
Investors should consider the non-linear factors and market conditions when constructing portfolios based on CO and OC strategies. The findings suggest that defensive positioning may be warranted during periods of market stress.
Conclusion
Key Takeaways
• Overnight Returns: CO returns exhibit positive selectivity alpha and negative market timing ability, particularly during down markets.
• Daytime Returns: OC returns show negative selectivity alpha and positive market timing ability.
• Market Conditions: Alpha is significant during periods of negative realized market risk premium, emphasizing the importance of market conditions in return dynamics.
Final Thoughts
The paper provides a comprehensive analysis of the non-linear relationship between overnight and daytime returns, offering valuable insights for investors. By understanding the dynamics of CO and OC strategies, investors can better navigate the complexities of equity markets.
For the full analysis and methodology, refer to the original paper: Non-Linear Asset Returns: Overnight Returns as a Market Timing Strategy.
Keywords: Overnight returns, daytime returns, market timing, selectivity alpha, non-linear factors, down market alpha.