Trend Follower Agonistes

A recent Bloomberg News article (https://www.bloomberg.com/news/articles/2019-03-01/one-of-wall-street-s-most-popular-trading-strategies-is-now-failing) mourned the passing of a venerable quantitative trading strategy – Trend Following.  The central claim of the article is that the strategy is antiquated relative to the speed and variability of modern markets.  We have been doing this for 40 years – it’s not the first time we have heard this. There is no denying that trend following has struggled, but we think the article has the reason exactly backwards.

The investment world’s conception of trend following has changed over the years.  Practitioners of the craft were, in the 70’s and 80’s, viewed as highly skilled magicians, teasing returns out of the tangled chains of futures markets, worthy of fees supporting retirement owning Major League Baseball teams or living in mansions in Mayfair.  Starting in the late 1980s, investors began to think about trend following in risk premia terms, reproducible factor returns – more science less magic.  Like other risk premia, trend results are highly variable (see our view on risk premia here – https://blog.mtlucas.com/2019/02/28/looking-beneath-the-hood-of-factor-investing), with recent trend returns on the downside of that variability curve.  We thought we would take a closer look at what is driving those lower returns, and why and when they may change.

At Mount Lucas, we have been thinking about the trend risk premium since 1988, when our first client pressed us for a benchmark for his shiny new managed futures investment.  That led to the creation of the MLM Index, the first price-based metric of trend returns, and perhaps the advent of the notion of “alternative beta”.  For those unfamiliar, the MLM Index is a simple metric of the trendiness of a basket of futures contracts, employing a simple long/short trading algorithm to determine positions.  The Index uses a static position weighting scheme, as opposed to a volatility adjusted method, making it particularly helpful in understanding environmental changes through time.  For example, the Index position in natural gas at a given price will be the same now as in 1995, regardless of short-term changes in volatility.  The history of the broadest version of the MLM Index is shown in the chart below:
TF1

Recent returns have been poor for the Index, and for the broader CTA community as well.  Let’s dive a little deeper and see what’s driving those results.  To state the obvious, trend risk premia is driven by changes in the component markets.  Big changes in a broad swath of markets lead to the highest returns, so we need to establish how the current environment compares to the past.  At the core of the calculation of the MLM Index is the creation of a continuous price series for each constituent market that computes the result of buying futures contracts and rolling them to the next delivery as they reach expiration.  Using this series for each of the 27 markets (we excluded one short history market) in the broadest version of the Index, we calculate the 1 year percentage change in each market.  Daily from 1993, we summed the absolute value of the 1 year percent changes in each sector – commodities, FX, bonds and equities – then weighted the sector totals by the basket weights used in the MLM Index. The sector weights are proportional to the relative volatility of the sectors – commodities are more volatile than bonds, so weight up the bonds. Nothing fancy, just arithmetic, knowledge of differential equations not required.  This daily number will be high if lots of markets have moved substantially in the last year, and, conversely, low if changes and breadth are small.  That series looks like this, back to 1993:
TF2

Right away we see that this series is at the lows, and has been low for some time.  Markets are not moving.  Markets are not moving too fast, they are not moving at all!  If we compare our price change series with the 1yr returns to the MLM Index, the relationship is clear.
TF3

As predicted, the return, or risk premium, in trend is zero or negative.  Compare the recent experience with 2008, when trend returns were one of the few bright spots on the investment landscape.  Managed futures, and the broader and related macro space, will suffer if markets are not moving.

If we view managed futures correctly, as a portfolio element rather than a standalone investment, we should be pleased with this outcome because of what that should mean for investments in traditional risky assets.  Stability across a wide range of sensitive prices should be a wonderful environment for equity and credit.  In fact, the trailing Sharpe Ratio of the US stock market to September 2018 was one of the highest ever recorded! The bad news is the events that are likely to shake the futures market out the doldrums are unlikely to be positive for risk assets. The somewhat symmetrical ying and yang of this relationship is shown below.  When and how that change happens is anyone’s guess, but we will be watching closely.

TF4

All Data from Bloomberg and Internal Calculations

Past performance is not indicative of future results.  The performance shown for the MLM Index EV is based upon a historical index and is gross of any fees and expenses.  No investor has achieved the exact performance results presented herein.

The S&P 500 Index is a market-weighted basket of 500 large capitalization stocks as compiled by Standard & Poor’s Corporation.