A Predictive Contrarian- Momentum Strategy

1
A Predictive Contrarian- Momentum Strategy

• Sharpe Minds Asset Management
• Ray Jacobson
• Venkata Kodali
• Josh Rose
• Jim Sheehan
• John Watts

2
Objective and Methodology

• Develop a predictive contrarian-momentum
  strategy.
• Data based on top (momentum) and bottom
  (contrarian) five monthly performers from 27 SP
  industry sectors.
• Utilize logistic regression to trigger when one
  position is better than the other (Long Momentum
  minus Short Contrarian, or vice versa).
• Third alternative is investing in T-bills. Note
  that overall strategy is market-neutral.

3
Model Variables

• Dependent Variable Return from a Long
  Momentum-minus-Short Contrarian strategy.
• If return is positive, LM_SC 1.
• If negative, LM_SC 0.
• Independent Variables
• Lagged change in 30Y 5Y US Treasury bond yield
  spread (1 period)
• Lagged change in Baa Aaa corporate bond yield
  spread (1 period)

4
Other Variables Considered
5
Regression Stats
6
Regression Predictive Power
7
Strategy Execution

• Raw Predicted gt 0.6 Take position LM SC
• Raw Predicted lt 0.4 Take position LC SM
• 0.4 lt Raw Predicted lt 0.6 Take position T-Bills

8
Performance of SMAM portfolio
9
Cumulative Returns
10
Out of Sample Looks Good
11
Concluding Remarks

• Market-neutral strategy generated positive
  returns in excess of T-bills.
• May be better to just invest in T-bills (less
  volatility), especially given transaction costs.
• Financial variables (e.g. Book/Market, P/E,
  DivYield) failed to have predictive power.
• Credit spreads did well, possibly signaling when
  riskier Contrarian stocks should outperform.