100 likes | 202 Views
Independent Study Project A Market-Neutral Strategy Lewis Kaufman, CFA Fuqua School of Business, ‘03 lewis.kaufman@alumni.duke.edu Faculty Advisor: Campbell R. Harvey November 12, 2014. Agenda. Performance, 1979-2001 Strategy Data Collection Factor Analysis Risk Optimization
E N D
Independent Study Project A Market-Neutral Strategy Lewis Kaufman, CFA Fuqua School of Business, ‘03 lewis.kaufman@alumni.duke.edu Faculty Advisor: Campbell R. Harvey November 12, 2014 1
Agenda • Performance, 1979-2001 • Strategy • Data Collection • Factor Analysis • Risk Optimization • Analysis of Results • Further Study • Conclusions 2
Performance, 1979-2001 Hedge, In-Sample Hedge, Out-of-Sample S&P 500 3
Strategy • Use intuition about drivers of stock price performance to develop a quantitative, market-neutral strategy • Focus on the S&P 500 where liquidity is greatest and transaction costs are minimized • Employ risk-optimization techniques to maximize returns for a pre-specified level of risk • Explore other potential enhancements to the base strategy 4
Data Collection • Identify appropriate time horizon • Obtain fundamental data for each stock in the S&P 500 dating back to 1979 • Sample period includes several different market environments • Test out of sample (1996-2001) to verify that the strategy is robust • Focus on quarterly intervals • Limited access to data • No access to expectational data (i.e. no EPS estimates) • All fundamental factors are therefore on a trailing basis • Companies with negative book value or net income excluded • EPS, NOPAT annualized each quarter 5
Factor Analysis • Identify key factors at the outset • Important not to mine data for best result – use intuition • Key factors EPS, P/B, P/S, ROE, ROIC • Create three fractile portfolios for each factor • Perform sort each quarter based on each factor • Group stocks into top, middle, bottom portfolios • Track performance of each fractile for coming quarter • Rebalance each quarter • Evaluate each pre-specified factor both in, out of sample • For each portfolios, perform diagnostics for all three fractiles • Select top three factors based on returns, risk 6
Risk Optimization • Risk optimization performed under supervision of Professor Campbell R. Harvey • Optimal portfolios weights determined • Scoring system established for individual security selection • Additional information available upon request 7
Analysis of Results • In-sample results: 1979-1995 • 28.6% compounded annual return, 11.5% for S&P 500 • 0.04 quarterly beta, implying negligible systematic risk • 6.0% quarterly standard deviation of returns, 7.0% for S&P 500 • Outperformed S&P 500 16 of 17 years • Outperformed 53% of time during up quarters for S&P 500 • Outperformed 100% of time during down quarters for S&P 500 • Out-of-sample results: 1996-2001 • 27.2% compounded annual return, 10.1% for S&P 500 • -0.41 quarterly beta, implies negative correlation with market • 10.3% quarterly standard deviation of returns, 7.6% for S&P 500 • Outperformed S&P 500 4 of 6 years • Outperformed 21% of time during up quarters for S&P 500 • Outperformed 90% of time during down quarters for S&P 500 • Out-of-sample analysis: A robust strategy • Positive absolute returns in all but one year despite a difficult market environment • -22.5% in 1999: peak of the bubble, fundamentals irrelevant • 2000 and 2001 were fantastic years: 65.6% and 78.9%, respectively • 1999 would not have put us out of business; 2000, 2001 more than compensate 8
Further Study • Incorporate expectational data • Inherent limitations to relying on past accounting data • Does not capture market expectations • Use other indicators to enhance strategy • Incorporate macro factors such as interest rates, dollar • Consider using leverage • All returns assume no leverage • Develop security weighting system • Examine value-weighted returns 9
Conclusions 10