A Trend Following Debate

A recent email exchange:

Name: Reading “The Talent Code” by Daniel Coyle. A superb book, I wonder if you know it. On the subject of TF may I also suggest you to consider a podcast with Gary Anderson (The Janus Factor). His work is superb because he has developed a way on knowing when to use tf tools or when to be a mean reverting contrarian. We know that TF does not always work: to know WHEN it offer a different edge. I really believe you should talk with him.

Covel: So all those drawdowns taken by some very smart and successful TFs over the decades were all in vane? They could have avoided them all by switching out at the exact right time to another strategy?

Name: Not 100% but to a certain extent it would seem so. Before you dismiss the idea, have a look at his papers / work. You can measure momentum for leaders and laggards separately; then you calculate the spread in relative performance, some sort of relative strength between 2 equity lines: one for a portfolio with only leaders and one with only laggards. Say from the universe of all commodities you look at the top 20% and bottom 20%. TF has its own cycles, they are not of fixed length clearly but its feature, positive feedback, can be measured. This is another level of analysis not seen in the Turtle work: having a plan, having a broad universe of asset classes, normalizing risk, adjusting risk is all a way to mitigate the inherent volatility of trend following. Anderson’s work has, so far, since 2003, limited to stock but there is no reason not to apply to commodities. Consider this: TF works like a peach until it breaks, then it starts to work again. Maybe the WHY cannot be explained but the WHEN, in reasonable terms, can be calculated, plotted and, possibly, integrated in a TF portfolio. Can you remember how from March 2009 the market has gone up overall but it was led by the laggards? Applying TF to stocks would not have generated good profits or even losses (what I call a “dirty” trend, where reversals are deep and costly for TF). Nowhere in the TF work I have done I have seen this interesting concept. Food for thought.

Covel: At first blush if tomorrow can’t be predicted the idea of a money making system that shifts gears such as you propose … would be novel. I would wonder why all the big names over the years have not been available to invent such a fool proof system? And beyond academic efforts, does a track record exist to show this bi-polar trading system in action?

Name: I am working to see if there is a way to use this “technology”. Tomorrow cannot predicted but you can reasonably sure that it is not going to go very far from where you are. There is a gradual transition from TF to MR and that can be measured – the HOW one implements the idea is another matter. Anderson has been working on stocks. No there is no track record.Consider this: work on relative strength goes back to the 60s with Levi’s paper. Only now there is widespread acceptance of momentum. It took ages to dismantle the idea of EMH. Anderson is trying to show / describe why and when TF (momentum) is “shape shifting”. It is a fascinating subject. And Levi’s work was snubbed from academia if you remember. Original ideas take time to establish, to be accepted. In any environment. You are showing yourself that TF works with numbers and yet very few out there are adopting it. It is what someone called one of those “mysteries of life”.

Covel: Good luck here. I would read Taleb’s work, consult with TFs who have come before us, objectively analyze why TF works, and maybe not try to reinvent the wheel.

Name: First I did not say that one could exactly know when to switch from TF to MR. There is of course some lag but what is interesting, is the persistence of a particular “state” until it changes again. I cannot say that TF drawdowns were in “vane” because either they recognized the problem and accepted to go with it anyway by sticking with their strategy no matter what (a-la Dunn) or they have tried to reduce the impact of the rotation between positive and negative feedback by adjusting risk and market exposure by calculating correlation coefficients to change the mix. Also adding more and more markets, reducing exposure when all the markets are working too much too nicely (a sign that risk has gone up too much), or simply by decreasing risk during drawdowns. ALL these methods address the problem from a different point of view without necessarily asking the question: is TF working NOW, is there a way to see if the lack of performance in leaders and laggards from a TF can be distilled in an indicator to warn… I have added some of his work available on internet for you to see and ponder. It’s your call.

Covel: What is the problem? The idea of a drawdown is the problem? Losing any money is the problem? What do you mean by “impact of rotation between positive and negative feedback”? A loss? BTW Connie Brown dodged my podcast. MTA really needs to clean up the hocus pocus on the train!

Name: I agree with that 150% – you know I dislike “normal” TA. But it is YOUR fault! You are going to someone that a.) is not systematic, b.) she uses EWaves, c.) she makes “calls”. Basically your are asking the right questions to the wrong person. Unless you are deliberately an agent provocateur with technicians. My humble suggestion is to interview those technicians that are systematic like Kirkpatrick. Connie Brown is the exact opposite. But you already knew that.

Covel: Yeah, but if the people in MTA let that nonsense in, on their boards, etc. — what value is CMT/MTA?

Addendum for everyone: Yes, that last sentence may be controversial, but tell me how my thought goes in the wrong direction?

The Author