Oct 16, 2018

What is the correct benchmark for trend following?


"What is the correct benchmark for trend following?" [1]
This is the question asked by Alpha Architect here today. I've seen this kind of thing before. I love their post and their asking of the question but I find, for myself, that the answer to this question is really really easy.   For me the answer, not to drag it out, is that the proper benchmark is one's own current portfolio design and the expectations that come from that.  I know this because I have been invested in as well as rolling my own trend following systematic alt risk program for close to 10 years and that is what I do. 

I have looked at a million ways to evaluate the choice and its marginal impact. I always come back to the: (1) do nothing or keep doing the same thing (current portfolio) vs. (2) do something different (add trend following) framework.  And in that framework it's easy.  It is always do something different (add trend following) because it is additive to the do nothing scenario.   

I say that because, as a retiree, I am not an infinite-life single-period no-consumption portfolio manager. I am a many-period short-lived consuming retiree.  With kids. And goals.  Trend following, with a similar-return lower-vol vibe, provides benefits to me that I cannot get elsewhere.  I won't prove it in this post. Just google some work by Andrew Clare on trend following and perfect withdrawal rates, or a ton of work by NewFound Research on trend following*. You can even search this blog. It (trend following) is clearly additive for people like me when we use our experience from the past data (standard disclaimer on past not predicting future...).  

So, if one were to have the wherewithal to track the month over month performance of one's own portfolio in a time-weighted fashion (to account for in and outflows) and net of any trend following positions, then I think that is the proper measure. An easier way is maybe a proxy for a portfolio similar to one's own base. If that is 60/40, for example, or close, then some index/fund based on 60/40 might work, say an ETF like AOM or maybe Vanguard VSMGX or something. Maybe there are others. But you'd have to be careful to make sure things like dividend flows and issues like advisory fees are factored in.  Bon chance!


*
-------------------------------
Risk Ignition with Trend Following, Newfound Research 4/23/18
Using Trend-Following Managed Futures to Increase Expected Withdrawal Rates, A Miller 2017


-------------------------------
[1] I didn't realize Tadas was going to pick this up in abnormalreturns.  This was pretty a pretty casual post.  There are more formal and complex ways to express this that I have not bothered with.  The base portfolio is probably assumed to be single-period mean-variance efficient on a frontier using expected forthcoming arithmetic returns.  In addition, since we are actually multi-period retirees we also probably contemplate a geometric mean frontier that anticipates the volatility effects on returns which may create inflection points in the geo frontier at higher vol so we are somewhere between min-var and inflection on that geo frontier.  Then we also probably think about the covariance and vol-reduction considerations of adding/reallocating an nth asset such as trend following and whether it nudges the frontier out, which it often does in multi-period mode.  That's why I am looking at the baseline as a benchmark and then incrementally I'm not really looking at trend following in isolation but more likely the incremental portfolio changes after the reallocation. My guess is that it is more often than not a constructive move, a point I was being casual about above.  The question is does it help my portfolio by making the change, hence the baseline as benchmark.  I can be coached on this if I am way off...







No comments:

Post a Comment