Trend Following in Retirement Portfolios - another quick look

I started to take a look at trend following in retirement portfolios here and here.  This kind of thing has been covered elsewhere before so this idea is not new (see references). I am just trying to get my head around it on my own in a way that I can understand.  So far, it, trend following, seems accretive.  Today, however, I am going to look at something else a little bit different.

IF we can say (not sure we can) the following:

1. Trend following, either as an allocation or as an overlay on an asset structure, has a convex pay-off structure, and
2. The convex pay off structure (return smoothing) enhances retirement consumption capacity, and
3. Trend following is not strictly a return premium for taking on risk but rather an exploitation of a behavioral anomaly, and 
4. Non-real-premia strategies can sometimes suffer from crowding and decay  

THEN it might be fair to ask the following

1. Is trend following suffering any decay these days, and
2. If there is decay, does it matter, or 
3. Are there different ways to look at the accretiveness of trend following? 

--------------

This is not a tutorial on trend following methods or theory nor does it differentiate between cross sectional or time series momentum or any other flavors of how people attack the behavioral anomaly at play here.  For that, use Google or SSRN or other resources and entities like AQR or Newfound Research that are among the many that have come at this topic.  Here, however, are some quick riffs...

1. Thoughts on "IF" #1, the convex payoff

Here is a stylized depiction of the payoff structure of different assets from a NewFound Research blog on No Pain No Premium (see references). I'm less proving anything here than I am just asserting it.

And here is my rendition of some AQR data on time series momentum where I geo-chain TSMOM excess returns over an arbitrary and cherry-picked and annualized 18 month hold and compare to the SnP total return calculated in a similar fashion. I have not convinced myself that this is a proper way to do it.

So, convex.


2. Thoughts on IF #2, Trend following can enhance lifecycle consumption profiles

From the same Newfound post:

"By combining different asset classes and payoff functions, we may be able to create a higher quality of portfolio return. For example, when we overlay a naive trend strategy on top of U.S. equities, the result converges towards a distribution where we simply miss the best and worst years. However, because the worst years tend to be worse than the best years are good, it leads to a less skewed distribution. "

Which they illustrate thusly:

And about which I opined in a previous post that ditching the left tail for a higher quality and more "normal looking" return distribution can pay dividends by way of a higher "horizon spend capacity."

Again we haven't proved anything here, just illustrated the idea. It's a pretty good idea though.[1]


3. Thoughts on IF #3, Trend is currently an exploitable anomaly

No stats are offered on how exploitable or how much it adds in value but here is a visualization of the trend anomaly concept, again from NewFound in another blog post (NewFound 2/25/2019), just to see it:

Theory of Trends from Newfound Research

4. Thoughts on IF #4, Strategies sometimes decay if not a real source of risk-based premium and thoughts on THEN #1 "is there decay?"

This trend anomaly is seen across multiple asset classes and geographies and has been persistent over time though it exhibits, like everything else, periods of under-performance.  Whether under-performance ever becomes un-diagnosed strategy decay or something that has been crowded out is the trading question of the new century.  For now all we can say is that it is a recognized and acknowledged source of return. I'll add more links here or in another post on the history and expected future of trend following. I just have to find them. I know I have referenced this kind of thing in the past on this blog.  Again, I'll point to AQR and NewFound for background for now.

The question for me -- since I am personally invested in trend following strategies and practice it myself in capital I have allocated to myself, and since the outcomes since 2009 have kind of sucked -- is whether "trend is dead."  I started to look at this general idea myself using the AQR data set I used for time series momentum above.  Here was my first attempt using monthly return data grouped into box quartile plots by year:

Doesn't look that great to me and matches my personal experience but this is statistically inconclusive so I asked Corey H over at Newfound. He was tied up in a conference but sent me this (https://qoppac.blogspot.com/2018/11/is-trend-following-dead.html?m=1)  a blog post by Rob Carver.  First of all, I didn't realize that asking if "trend is dead" is now the third rail of trading.  His first sentence goes like this:

I get asked this question at least once a week. As those of you that have met me IRL ('in real life') will know I have limited patience and I'm easily bored. I'm definitely bored of answering this question. This post is the last time I'll answer it.

He goes on to subject trend following to statistical tests more rigorous than I can pull off (yet... TBD. I'm feeling too old to learn) and comes to his own conclusions. To jump to the punch line, Rob says this: 

"Is trend following dead?" I don't know. Probably not. Now leave me alone and let us never speak of this again. The next person who asks me this question will get a deep sigh in response. The one after that, a full eye roll. And with the third person I will have to resort to physical violence.

Let's call that a no for now.  

5. Thoughts on THEN #2 and #3 - Does it matter and is there another way to look at this?

As always there are statistics and probabilities...but there are also payoffs and consequences.  We have only looked, in item 4, at return statistics.  But we are not interested in trend just for return, we are also interested because we want to either spend more or enhance portfolio longevity in the presence of multi-period consumption and for that return is not the only game in town to which we alluded above in looking at PWRs and left tail mitigation.  

So, let's say that the payoff is generally convex on both sides of the scatter chart. Let's also say that positive returns have been awful since 2009. Do we abandon this strategy yet as a dead strategy. As Rob says "probably not." That's because there are two sides to the convexity and the left side is what matters a lot for ret-fin.  

First let's look at the time series of AQMIX. This is not exactly what I do personally but as a big publicly-available managed futures fund it has to stand in for trend following, otherwise, what are they?  I do have a position in this, btw. 

2010 is as far is it goes back. Doesn't look great lately and it has not been a fun ride. Time to jump, right? Not so fast.

Now let's take the AQR TSMOM data from above and isolate 2009+ which, in my experience has been horrible and I want to jump ship every day I think about it. Blue plus red is the exact same as the chart above of all blue dots. Red alone below is 2009 +

Yeah, that looks pretty shit. But the thing I like is that the convexity is still there on the left side. This still looks like a good hedge.  In fact, this should look awfully familiar.  To my amateur eye, this (red) looks like a type of put option.  As a refresher, here, from google images, is the pay-off structure for a long put: positive premium cost above strike guarantees certain negative payoff and the option value below strike rises with decreasing price. Basic stuff. Standard hedge. 

But the red dots above, though put-like, are not a standard hedge. It's still a hedge though. Let's re-render the red dots (2009+) in the chart above -- while calling zero return in the SnP the "strike" -- as a box plot to the right of the strike and scatter below. There is no real reason to do this other than to illustrate a point.  

This makes it look like some exotic option with a "negative, uncertain premium" and a positive payoff structure below the (arbitrary) strike. I like it: it's a hedge that might pay me if I'm wrong (in this case market goes up i.e., against the hedge) and will likely pay me if I'm right. Decay or no decay, this looks like it still might be a winner assuming stability in the strategy returns and convexity...and that is a big if.

Why not just buy a put? Well first of all, no one is going to ever give you a negative premium deal.  Second, in Clare (2017), they do a mini literature review on the use of puts vs trend following and cite Israelov (2017), Ilmanen (2016), Asvanunt et al (2015), Strub (2013) to come to the conclusion: 

...divesting (albeit temporarily) offers a much better solution to reducing [long/slow or fast] drawdowns (i.e., managing tail risk) than either buying puts systematically or trying to time their purchases using conditioning information. ...[and]... Unless one knows when a 'fast crash' is about to occur and can time the option purchase cycle to good effect, then switching to cash via a trend following rule appears to be the best solution.

So, as painful as my ownership of trend following has been of late, I will probably hang in there for a little bit. We'll see. Always good to at least think about it.  

---------------References ---------------------------------------------

AQR data series 2018, Time Series Momentum: Factors, Monthly

Carvers, Rob [blog], (2018, Nov). Is trend following dead?

Clare, Seaton, Smith and Thomas (2017), Can Sustainable Withdrawal rates be Enhanced by Trend Following?

Hoffstein, C. [blog], ) 2/4/2019, Feb), No Pain, No Premium, NewFound Research 

Hoffstein, C. [blog], ) 2/25/2019, Feb), Three Applications of Trend Equity, NewFound Research

---------------Notes ---------------------------------------------

[1] I forgot to mention, Hoffstein (2/4/2019) shows that allocating to bonds does not mitigate the left tail because the blended payoff is dominated by the kurtosis and skewness of equities.