Monday, October 15, 2012

The TLT / XLF Pair Trade

Regular long time readers are well aware of Project Z, a BZB research project that has extended over several years and which endeavors to capitalize on the relative change in variance between pairs of stocks and ETFs that have a positive correlation.  Clicking the link will show a list of approximately 40 Project Z related posts.

Without getting too technical... Project Z examines the log ratio of each side of a pair's standard deviation from a norm...with the expectancy that underlying forces will tend to drive price in each stock or ETF return to the norm.  The actual algorithm that drives Project Z is rather complex so I won't delve into its various nuances.  However, the TradeStation 2000i performance reports of both sides of the TLT/XLF pair trade (in Excel format) are available to readers upon request (15 pages each).

Why am I even talking about this?  Because pair trading reflects the true essence of market neutral trading and investing may surprise you to learn... some of the biggest proprietary trading shops make most of their money with this strategy.

I'm going to devote today and next Monday to explore the TLT / XLF setup.  There's a lot of information to examine within this trade and its important to understand the dynamics behind why this pair is unique.

One thing that's different about this version is that it uses the TrendX price in lieu of the daily close.  The TrendX is simply a floating pivot value (High+Low+Close/3) that typically reveals an intrinsic momentum factor not apparent when simply looking at raw closing prices.

To get a full picture of ways to manage a pairs trade we need to examine each side of the trade independent of the other. This week we're looking at the XLF side because there's nothing that says you have to trade both sides of the pair simultaneously/  Let's just find where the sweet spot of the pair trade resides and try and farm those easy dollars. 
The returns for the XLF side are pretty impressive given that XLF is notoriously difficult to forecast with most traditional technical indicators.  There are an equal number of long and short trades and both sides of the trade yield the 92.86% profitability.  The maximum consecutive loser trades at 1 makes this an attractive setup in my book.
These returns are based on a 16 month lookback and the design parameters are only marginally optimized.  For backtesting and performance updating I use a floating 16 month test period, updated monthly.  This procedure helps to identify when a system is losing its efficacy and either needs rethinking or discard.  The TLT/XLF trade has seen periods of stagnancy during the past 16 months but the seldom falls below the equity line.
Next Monday we'll look at the TLT side of the trade.