How to Make Money in A Bear Market Learning from Futures And Options Traders joint work with Roel Oomen Deutsche Bank and Alex Stremme Warwick Business School 1
investigating asset pricing anomalies market timing is a very common investment strategy many newsletters are devoted to trying to make this strategy successful academic evidence suggests that few investors retail or institutional are successful market timers natural to ask if conditional strategies using predictive variables can lead to successful market timing 2
investigating asset pricing anomalies predictive variables considered so far have been business cycle or macro-economic variables our focus is different-we consider the positions of traders in futures and options markets the views and positions of these traders particularly in the options markets are forward looking and could thus have predictive power Our strategies are real-time and do not depend on variables such as the book-to-market ratio that were considered ex-post 3
investigating asset pricing anomalies our predictive variables are the VIX which has been found to be useful for stock-bond diversification and market timing we also consider hedging pressure which is defined as the fraction of long positions in each category of traders the CFTC Commitment of Traders report classifies traders as hedgers, non-hedgers and non reportable traders and we use hedging pressure for each of these categories non-hedgers are mostly large speculators while non-reportable traders are small speculators 4
methodology: we also examine conditional asset allocation: using predictive variables our dynamic strategies are unconditionally efficient with respect to the conditioning information (predictive variables) these were introduced in Hansen and Richard (1987) and then Ferson and Siegel (2001) and are studied in detail in Abhyankar, Basu and Stremme (2007) 5
related literature: Methodological Hansen and Richard (1987) fundmental paper that introduced conditioning information in meanvariance analysis Ferson and Siegel (2001) provide explicit portfolio weights for unconditionally efficient strategies show that these strategies exhibit a conservative response to extreme values of the predictive variables Abhyankar, Basu and Stremme (2007) provide a statistical test to measure whether return predictability expands investor s opportunity set study the empirical performance of several strategies 6
related literature: Predictive Variables DeRoon, Nijman and Veld (2001) study the effect of hedging pressure in both futures and asset markets Connolly, Stivers and Sun (2005) show the the VIX helps for stock-bond diversification Copeland and Copeland (1999) Investigate market timing strategies using the VIX 7
data: base assets and predictive variables Data is at a weekly frequency Base assets are the S&P500, 10 Year T Bond and 1 month T Bill as the risk free asset Data obtained from the CBOE www.cboe.com Predictive variables: Level of the VIX obtained from CBOE Hedging Pressure obtained from CFTC www.cftc.gov 8
data: Hedging Pressure Hedging Pressure: For a given category of traders it is defined as the fraction of long positions in that category the CFTC classifies large traders (positions in excess of a 1000 contracts on the S&P) as commercial or non-commercial commercial traders use the futures for hedging and non-commercial use it for speculation and arbitrage CFTC continually monitors the activities of traders and has re-classified traders from commercial to non-commercial In addition to this there is the non-reportable category which consists of small speculators 9
unconditionally efficient strategies are dynamic strategies that optimize with respect to unconditional moments Conditionally efficient strategies optimize with respect to conditional moments Unconditionally efficient strategies are conditionally efficient but not vice-versa (Hansen and Richard (1987)) Unconditionally efficient strategies perform tactical asset allocation with a strategic objective: they are less responsive to short term fluctuations in predictive instruments, unlike conditionally efficient strategies We consider out of sample performance of maximum return strategies estimated using a rolling window of 6 months in order to trade off between incorporating changing market conditions and allowing the strategy to capture the true predictability. 10
Overview The hedging pressure variables that work are commercial hedging pressure and non reportable hedging pressure-positions of large hedgers and small speculators Large speculator s positions are not informative-too diverse perhaps VIX alone does not work-sends strong signals to move in and out of market but no directional ability Strategies using commercial hedging pressure are positively correlated with it while those with non reportable hedging pressure are negatively correlated with it Adding the VIX to hedging pressure leads to more variability in weights and generally improves performance We first use the T bill as a riskless asset then time against a 10 year T bond and finally consider both the bill and bond S&P weights are circles, T bill is light dotted line. Results are net of transaction costs of 5 bps 11
results: market timing using the short rate doesn t work 12
results: market timing with commercial hedging pressure 13
results: timing with non reportable hedging pressure 14
results: timing with commercial hedging pressure and VIX 15
results: timing with non reportable hedging pressure and VIX 16
results: timing with commercial and non reportable hedging pressure and VIX 17
results: timing with 10 year bond : commercial hedging pressure 18
results: timing with 10 year bond : non reportable hedging pressure 19
results: timing with 10 year bond : commercial hedging pressure and VIX 20
results: timing with 10 year bond : non reportable hedging pressure and VIX 21
results: timing with 10 year bond : commercial and non reportable hedging pressure and VIX 22
results: timing with 10 year bond and 1 month T bill: commercial and non reportable hedging pressure and VIX Minimum Variance Strategy 23
results: timing with 10 year bond and 1 month T bill: commercial and non reportable hedging pressure and VIX Maximum Return Strategy 24
results: how do the strategies work a composite signal The VIX provides a signal to change the level of investment while the hedging pressures provide directional signals We then multiply each of these with the VIX to create directional variables (VCHP, VNRHP) 25
results: timing with VCHP and VNRHP (T bill) 26
results: timing with VCHP and VNRHP (T bill and 10 year Bond) 27