Reader Followup: Combining Strategies

01Dec08

Below is an email I received from Monty Webb of Richardson, Texas building off of some of my previous studies: the energy (XLE), consumer discretionary (XLY), and semiconductor leader/laggard strategies, and daily follow-through (note: all of these indicators are part of the State of the Market report).

This my favorite type of feedback. I generally try to keep the ideas on this blog reasonably simple and accessible. It’s nice to see readers build something a bit more sophisticated out of them.

In a nutshell, Monty has optimized a weighted portfolio of these different strategies and then done a very brief 2-week walk forward test (another flavor of in-sample/out-sample testing). Enjoy. Great work Monty!

Michael, XLP, consumer staples also has a similar result to XLY.(when adjusted for trigger point. i.e. above or below 0 was not as good as above or below -0.4 for XLE)

I took the following “predictors” (lead or lag and trigger point were allowed to optimize over the last year): XLE, XLP, XLY, SOXX, and Follow-through day, and plotted the individual returns over the last year. (each separate predictor had the trigger point set by using a one year lookback)

I also went back two weeks and did two different optimizations. One was to use equal weighting with all 5 and the last was to do a weighting based on 1 year up to 2 weeks ago. I then used those weights and walked forward 2 weeks and both look nice. the green ~~10~Equity is the combined weighted return. The 5 equal gave a 407% gain and the optimized gave a 419% gain. The blue straight line is the time all were optimized and past that line is the walk forward. The numbers across the top are the % gain values at the blue line time, and the colored rectangles with numbers at the far right are the values as of last Thursday.

The curve fit weights were: XLE=4, XLP =0, XLY =1, SOXX = 1, and Follow Day =5, so the XLP did not add any improvements to the overall formula and was set to a weight of zero during optimization.

I enjoy your blog very much. It gets me into search mode.

Thanks
Monty

[click to zoom]