Blog Post

I'm indebted to Covestor member fallond for bringing this to my attention. I couldn'[t resist writing the folks at CXO about their recent report on the use of stops, and to my utter amazement, got a super reply back overnight. Talk about reponsive! Anyway, here's the note I received, and below that, the note that I sent.

Don,

Thanks for your comments.

Please note that the three studies reported in http://www.cxoadvisory.com/blog/external/blog8-11-08/ and http://www.cxoadvisory.com/blog/external/blog8-04-08/ are done by others. Prior to these studies, I had not noticed any formal studies on stop losses for some time. The authors of the studies are generally academics at universities, although I believe Andrew Lo ("When Do Stop-Loss Rules Stop Losses?") is probably well-grounded in real hedge fund activities. I assign highest credibility to the study in which he participated. As noted, this study concludes that “in the presence of momentum and/or regime-switching, stop-loss rules can add value.”

I do not have the resources to test all the permutations you suggest, but I can offer some general observations derived from reviewing hundreds of academic papers and practitioner claims:

 - Academic researchers tend to use somewhat simplified and rigid rules in their studies. These limitations may stem from data availability constraints, programming complexity or lack of theoretical arguments to support refining assumptions (related to a desire to avoid data mining/snooping bias). They sometimes focus on abstract statistical outcomes difficult for practitioners to interpret or implement.

 - Practitioners often rely on small samples and/or poor sampling methods. They tend to be very empirical in approach, incorporating substantial hindsight bias/data mining bias into their analyses (exacerbated by importing data mined rules from others). Many practitioner analyses do a poor job of addressing trading frictions. Some say that what they do is “more art than science.” The “foundational” books by Taleb and Aronson listed in the right margin of CXOadvisory.com are antidotes.

 I’ll take a look at the study you cite. You may be interested in the 30 items on trend following at http://www.cxoadvisory.com/blog/internal/blog-momentum-investing/.

Best regards,

Steve LeCompte

CXO Advisory Group LLC

(And here was my not to CXO)

Dear CXO, Thanks a bunch for your recent research on whether stops work. Although I'm not in a position to challenge the validity of your findings, as presented, I was really surprised at the results. I wonder if I might suggest adding a couple of variables to the mix? I'm a trend follower, with a fully automatic system. It has done fairly well over virtually all test periods, and in actual use (Don't ask about my discipline in sticking to it religiously!:) There are several ingredients that I believe are essential to a good system, and if stops are combined with those, I believe the very same parameters that you tested for would look better if whipsaw trades are filtered out at the onset. For instance: 1. If I only enter a trade on a high volume (say, at least 1 StDev above average) high range (ditto), with my own system, that has made a significant improvement. 2. I got the impression from your article that your stop would be moving up steadily as a move progresses. Does it stall out and go flat during brief consolidations? If not (mine does), I believe some simple rules to enforce that would also greatly improve the score. Years ago, as you probably know, Welles Wilder's Parabolic was all the rage, but a great drawback to it was that it automatically raised the stop into a parabolic curve, without regard to the stock or future's price and volatility. By modifying that, increasing the acceleration factor only if there are new highs (up to the max of .20), and decreasing it for every day that there is not a new high (down to a minimum of 0.00), made a great difference for me and my commodity clients. It greatly improved their score. 3. Throwing in any kind of long term filter also made a great difference. I'd really like to see your test rerun with some of these ideas. Thanks a bunch! PS: have you looked at this super report using a 10 ATR stop?

http://www.blackstarfunds.com/files/Does_trendfollowing_work_on_stocks.pdf