Monday, January 14, 2013

T2 LAB..Model AB tests...1.14.13

I'm continuing to focus on refining the new Model AB components for a 2013 defensive portfolio. 
The problem, of course, is that past results are absolutely no guarantee of future returns.
The best we can do then is build a historically correlated component mix and prepare to go to cash when the model dynamics reach certain thresholds of weakness...hence our RSQ and P6 indicators.
After running over 50 test portfolios this is about as good as it gets risk wise when holding a fixed portfolio of equal dollar allocations for a 6 component model.  A nice RSQ of .98 has kept the drawdown to a minimum although the equity line cross of the RSQ back in mid October would have argued for a CASH position.
If we use a top 2 approach with active weekly rotation our returns are cut almost in half, volatility is increased 300% and we incur more trading costs and maintenance effort.  This dismal situation may be traced to the failure of the momentum algorithm to provide an accurate forecast or other factors, including the whipsaws above and below the RSQ.  Whatever the causes, the fixed portfolio, in hindsight, looks like the path of least resistance...and risk.

Keeping with the mission statement of Mosaic, our next tasks include:
1. Refining the AB portfolio allocations using these components.
2. Using these and other components to build several shorter term trading models using the T2 and/or VTV momentum approach.