Thursday, January 14, 2010

Project Z - Qs

This is the second post in the Project Z series.
Today we look at the QQQQ component of our Lazy Man trading model using the VXX lens.
Once again we're looking at a divergent pair and in response to a comment by Gary on yesterday's post, keep in mind that I'm really not interested in taking both sides of this trade . . I'm just using the VXX to gauge the relative volatility and linearity of the rotation model components.
Just as a point of reference the current beta of the Qs is 1.1o, while the beta of the EEM tested yesterday is 1.47.

It is therefore a little bit surprising that both EEM and the Qs optimize with an N day value of 7 and the Qs generate one more trade than the EEM during the 6 month backtest period.
Both the Qs and EEM have eerily similar equity curves and linearity values. The Qs produce 16% more P&L gain and this may be traced to my tightening of the N day stop to 7, while I let the EEM ride out to 9 days. The Qs have a habit of acting a bit more squirrelly than the EEM so I'm inclined to rein it in as much as possible. One result of this short stop is that whereas the EEM pair generated all long EEM signals, the Qs variation produces 6 longs and 2 shorts.
After we examine the metrics of all 4 components we'll check the alignment of the signal dates to see how much deviation from SPY momentum can be detected. This is the process I term "basket weaving" and was explored previously with the FXY PDQ model.
Tomorrow we'll look at the DBC.


Kristianson said...

Interesting! I assume the performance of this Pair model is tracking taking a position in both QQQQ and VXX?


bzbtrader said...

You're absolutely right and using the PDQ Dashboard I can see how much gain is the result of each side of the trade.
For Project Z the important thing to know is that the proportion of gains per trade for the QQQQ, DBC, EEM and SH relative to the VXX is very close.
As the Project proceeds I'll present a couple ideas for evaluating the relative performance of the 4 ETFs versus the VXX, which is really what this little study is all about.

gohara said...

Hi Bob,
Interesting project. Would it be more appropriate to generate the pairs Trade Report with the "Ticker-A Only Unpaired" box checked? This option gives a return of 8.81% instead of 96.27%.

bzbtrader said...

The Ticker A only approach is a logical conclusion from the posts to date. However, I'm on an entirely different track here and hopefully this will become more apparent by the end of the month when I plan to conclude the Project Z posts.

Kevin said...


I'm not following - why would you not take the VXX side of the trade?


bzbtrader said...

You could take the VXX side of the trade as per the VXX PDQ, but Project Z is trying to explore the utility of using the VXX to forecast relative momentum and strength in a rotational model basket.

Kevin said...

OK - thanks.