Tuesday, October 7, 2014

A Trend Model anomaly?

I had a fair number of comments on my post last week about the investment results of my Trend Model (see Trend Model September report: +40.1%). Undoubtedly it got attention because of its outsized realized returns for one year.


A puzzling trend following question
I did, however, get a thoughtful response from one reader that I couldn't answer. Specifically, I pointed out that the Trend Model account achieved its returns when the US stock market had risen steadily in the last year with no major corrections. Pullbacks were brief and shallow.

The underlying model is mainly based on the application of trend following techniques to commodity prices and global equity prices. Moreover, trend following models are notorious for subpar returns in trend-less markets which subject them to getting whipsawed in and out of positions. The chart below shows the actual (not back-tested) Trend Model signals and showed periods of whipsaw.

Trend Model Signal History

I wondered out loud that the best stress test for this trading strategy would be when stocks were to undergo a volatile corrective period like 2010-2012.



Is the Trend Model long or short volatility?
Here is the question that I couldn`t answer. Trend following strategies are generally thought of as having a long-volatility characteristic (via Futures Magazine):
Simply stated, trend following is at its core a long-volatility strategy. In other words, it makes money when volatility expands (i.e. during trending moves). Conversely, it suffers frequent but small losses during non-trending periods in exchange for such infrequent but large gains. During non-trending periods the strategy attempts to tread water through the judicial use of stop loss orders until some market movement provides a large outlier move in which the strategy can profit.
Is it ironic that my Trend Model is performing well in an environment where pullbacks are minor and stock market volatility, as measured by VIX, is low, but I wished to see it stress tested in a corrective period when volatility, as measured by VIX, is high? Given the market action in the past few days and the most recent Trend Model market call (see A Yom Kippur bottom? Or just more volatility?), I may get that stress test very soon..

Does the Trend Model have a long or short volatility exposure? Is it behaving in accordance with other trend following systems? What am I missing?


Trend Model performance vs. CTAs
Trend following models have had a difficult time in the last few years. The BarclayHedge CTA Index shows how poorly CTAs did starting in late 2010, though they recovered late last year. The actual performance of the Trend Model account was stellar in the last year (much like the BarclayHedge CTA Index). On the other hand, the above chart of the actual Trend Model signals starting in 2009 were quite good as well (unlike BarclayHedge CTA).


Is there a disconnect here? Is this an anomaly in the characteristic of my Trend Model? If anyone has any thoughts, please either post it in the blog comments or email me at cam at hbhinvestments dot com.

5 comments:

Cameron Fisher said...

Cam: I've been reading your blog daily. Regarding the Trend Model: Is there some reason why you have not back-tested the model? Is historical data not available? - Cameron Fisher

Cam Hui said...

I have backtested the model, but backtests always show great returns. There is no substitute for live results. That's why I show the following on the blog:
1) actual trading results; and
2) actual paper trading signals.

Anonymous said...

https://www.scribd.com/doc/242217890/A-Century-of-Evidence-on-Trend-Following-Investing

Gary Hart said...

Cam, I think Schwager's comment about volatility and trend primarily applies to commodities not the stock market. For the stock market the VIX tends to be low during up trend moves. The VIX is a good gauge for uncertainty. When the stock market is trending up the uncertainty is relatively low. When there is lots of uncertainty and hence a high VIX (high expected volatility) there tends to also be trend moves in commodities. Great evidence of this is the improved performance of trend following CTA's recently.

The question then becomes which causes which? Does high volatility cause trend moves in commodities, or do trend moves in commodities increase volatility in the stock market? Or, does uncertainty cause both, or.....?

Anonymous said...

I agree with Gary above. They, CTA's, are long the mini or spx but that is such a small % of their capital allocation that they need the bigger portion of their portfolio to start moving, which is what they've got with the dollar and oil lately. Investing in ctas would be a good bet right now as they are in a drawdown and we all know what happens after a drawdown on a robust, simple, system like trend following. The funny thing is that I bet all their clients are screaming at them to allocate more to es and spx etc