Sunday, September 11, 2016

Making Adjustments To Our Pair's Trading Model


We had started a new pair’s trading model several weeks ago and started to build positions. The market had been very quiet with little volatility so we were waiting for a day like Friday to stress-test several components of the model. Given the rather robust equity sell-off it allowed us to monitor the positions and get acquainted with the various nuances that appear in any new investment model.

We suffered losses in the portfolio but we also found the weak points in our execution. The good news is that ten of the fifteen pair’s trades that we initiated were net gainers on the day, meaning the long positions did not fall as much as the short positions. That success rate of 67% in a market that was down almost 3% on the day is encouraging.

What did not work in our favor was the method of allocation we placed on each side of the trade, i.e. we were exposed too aggressively to the downside due to our error in weighting the positions.

We went to work this weekend to figure out what needed improvement and to put new procedures in place to assure this flaw in the system was eliminated.

Where We Miscalculated

Going into these trades we were basing our net long/short exposure on weightings derived from the last six months’ beta between the two securities in question. We understand now that recent pricing sensitivity has very little correlation to future pricing sensitivity and the method of beta calculation that we were using was skewed too far in one direction to be effective. The chart below shows one example of a pair’s trade that we initiated last week, long PFF / short SPY.





The red triangle represents the short term average (six month) beta profile that we used for this specific trade. The blue square is the long-term average (since 2007) beta profile and the green circle is the long-term median point within the range of beta between the two instruments.

Using the red triangle as our beta profile we saw on Friday that PFF was down 1.2% on the day. The SPY was down 2.4% on the day. That left us with a net loss on the pair’s trade of approximately -0.6%. Had we used the longer term average beta profile our net loss on the trade would have been -0.2% on the trade. Had we used the median figure our net gain on the trade would have been +0.5%.

Our relative performance z-score lets us know when there is an obscure pricing opportunity, i.e. pricing that resides near those protruding arrows in the figure above.  We trade on the assumption that those points on the outer boundaries will make their way back to the center and profit from the spread. By taking such an aggressive long/short ratio as we did in our current trade we have minimized our profit scenario to PFF rising much faster than SPY or falling much less than SPY as opposed to a reasonable mean reversion of the two prices. The basis of our relative price performance z-score is much more effective using the median beta weighting.

That happened across the board on Friday. Our long VOO / short XLK resulted in a net pair’s loss of -0.3% when technology clearly underperformed the S&P 500. Using the median approach would have resulted in a gain of +0.2% on the trade for the day. The long PCY / short SPY trade resulted in a net loss of -0.4% vs. +0.6% using the median method. The long EMB / short SPY resulted in a net loss of -0.5% vs. what could’ve been a net gain of +0.6%. Short XLI / long VOO resulted in a smaller net loss of -0.1% vs. a potential gain of +0.3%. We were down a net -0.9% in short XLE vs. long VOO but could’ve net a gain of +0.4%.

The list goes on as most of our pair’s moved in the right direction for us even though we lost money on most of them.

Corrective Action

We are of the opinion that the recent market weakness seen Friday has the potential to continue over the next several weeks and months. This is a call that we have been making for the last few weeks. We took profits from the Brexit lows in our core portfolio over the last two weeks and attempted to set up defensive positions as pair’s trades to mitigate suspected market weakness.

Given our belief of further weakness we are taking the following corrective actions in our pair’s portfolio.

(1)    We will rebalance our existing pair’s trades to better reflect the median beta profile we discussed. We anticipate this could help us recoup a majority of Friday’s losses and at the very least decrease the volatility that we experienced last week.

(2)    All pair’s trades going forward will be done using the new trade balancing methodology as discussed.

(3)    Over the next few weeks and months, as our macro market outlook improves and we start to build positions in our core portfolio we will use that as an opportunity to peel off the newly inserted shorts used to rebalance the portfolio. This will allow us to recapture the cost of our market lesson on Friday.

(4)    Stay vigilant and alert to additional nuances within our pair’s strategy.



Joseph S. Kalinowski, CFA

Email: joe@squaredconcept.net

Twitter: @jskalinowski

Facebook: https://www.facebook.com/JoeKalinowskiCFA/

Blog: http://squaredconcept.blogspot.com/

Web Site: http://www.squaredconcept.net/


No part of this report may be reproduced in any manner without the expressed written permission of Squared Concept Asset Management, LLC.  Any information presented in this report is for informational purposes only.  All opinions expressed in this report are subject to change without notice.  Squared Concept Asset Management, LLC is a Registered Investment Advisory and consulting company. These entities may have had in the past or may have in the present or future long or short positions, or own options on the companies discussed.  In some cases, these positions may have been established prior to the writing of the particular report. 

The above information should not be construed as a solicitation to buy or sell the securities discussed herein.  The publisher of this report cannot verify the accuracy of this information.  The owners of Squared Concept Asset Management, LLC and its affiliated companies may also be conducting trades based on the firm’s research ideas.  They also may hold positions contrary to the ideas presented in the research as market conditions may warrant.

This analysis should not be considered investment advice and may not be suitable for the readers’ portfolio. This analysis has been written without consideration to the readers’ risk and return profile nor has the readers’ liquidity needs, time horizon, tax circumstances or unique preferences been taken into account. Any purchase or sale activity in any securities or other instrument should be based upon the readers’ own analysis and conclusions. Past performance is not indicative of future results.












No comments:

Post a Comment