Systematic vs. Discretionary: A deeper dive

November 18, 2017 03:44 PM
Research indicates that the proper answer to the question of whether to allocate to systematic or discretionary traders, is both.

Last month we discussed the differences between systematic and discretionary trading strategies comparing broad index categories. Here we look at styles through global macro indexes.

The Société General (formerly Newedge) Indexes provide another avenue of comparing systematic vs. discretionary styles. Specifically, the SG Macro Trading Index – Quantitative and the SG MTI -Discretionary.  The data period is January 2000 through May 2017. There are 35 constituents in the Quantitative Index and 57 in the Discretionary Index.  The delineation between the two indexes is similar to BarclayHedge’s delineation between their Systematic and Discretionary Indexes.  The SG MTI – Quantitative Index include constituents where both trade generation and execution are quantitatively based, or systematic. These indexes are strictly in the global macro sector and have a shorter history, but still provide additional insight.

As a side note, we contend that the global macro sector contains the largest groupings of both systematic and discretionary styles of any single strategy.

There is variation in how the different global macro strategies are classified. The breakdown in Wikipedia is not uncommon:

“Neil Azous, the Founder and Managing Member of Rareview Macro LLC. explains that Global Macro Trading is grouped into three strategies: Discretionary, commodity trading advisor (CTA)/Managed Futures, and Systematic. Azous states:

Discretionary macro [traders] execute their strategies by deploying directional positions at the asset class level to express a positive or negative top-down view on a market. Of all of the strategies, discretionary macro provides the most flexibility, including the ability to express either long or short views, across any asset class, and in any region.

CTA/managed futures [traders] use products very similar to those that discretionary macro managers trade. However, the methodology by which they arrive at those long or short positions is very different. CTAs apply priced-based trend-following algorithms to the trading of futures contracts.

Systematic macro is a hybrid between discretionary macro and CTA/managed futures. The signals that are used to enter into positions are based upon fundamental analysis, similar to discretionary macro, but the deployment of those trades is based on a systematic, or model-driven process, as is the case with CTAs.

This categorization is similar but does not line up with what is the norm in the managed futures space or our initial discussion above.   The description of discretionary macro lines up with the added caveat that technicals are frequently used for entry and exit.  However, CTA/managed futures places all CTA/managed futures in one group – systematic. In fact, it actually groups all CTA/managed futures strategies in the systematic, diversified trend followers category. This is a good generalization since systematic, diversified trend follower is the predominant CTA strategy by assets under management, and why we used it in our second exploration into the systematic vs. discretionary styles.

However, there are a number of discretionary global macro strategies within the managed futures space. The last grouping called systematic macro is a hybrid between the other two.  In our delineation, this group would fall in our discretionary group with the term “systematic” merely adding to a level of confusion.  However, these categorizations are common in the hedge fund industry.  In fact, it would seem that SG named its SG Macro Trading Index (MTI) – Quantitative sub-index just that as to avoid any confusion.

Our experience has been that systematic global macro strategies tend to trade a substantially greater number of markets than discretionary global macro strategies and typically seek to capitalize upon momentum return drivers.  The same reasoning applies here as did to the diversified vs. physical commodities sectors. Discretionary global macro strategies typically focus primarily on the fundamentals, though in the financial markets rather than the physical commodities markets. There are numerous strategies that seek to capitalize on carry, relative value and momentum return drivers and are generally thematic and opportunistic in nature (see “SG quantitative vs discretionary,” below).   

As in our previous comparisons, there is minimal difference between the quantitative and discretionary returns. The risk-adjusted returns of the Newedge MTI – Discretionary Index are modestly better. The kurtosis and skew numbers are much smaller than the last two pairs of indexes analyzed and more typical of individual strategies. The kurtosis of the Discretionary Index is much higher than the Quantitative Index, indicating much fatter tails.  There is an immaterial difference in the skews - both positive and indicating a moderate propensity for positive surprise.

The correlation between the two indexes is 0.26, which indicates minimal correlation. The return of the 50/50 portfolio landed right in the middle of the individual indexes. The significantly better risk-adjusted returns came by way of reduction of the volatility (see “Better together,” below).

Again, there are no major differences between the Société General (Newedge) Quantitative and Discretionary indexes.  The drawdowns of the Quantitative Index are slightly higher and with longer recovery times. Also, the run-ups in the Discretionary Index are moderately higher on average and slightly longer in duration. The 50/50 portfolio significantly reduced drawdown depth but surprisingly lengthened the average duration relative to the two individual indexes due entirely to the current drawdown that is more than four times longer than the other top five drawdowns.

So Who Wins?
In our examination of the systematic and discretionary trading styles, we analyzed three pairs of indexes.  We started with, and place the greatest weight upon, the Barclay Systematic Traders and Discretionary Traders Sub-Indexes.  They have the longest performance history, the largest number of constituents, and are the broadest based.  We found that there was minimal difference in the compound rates of return, but that the Barclay Discretionary Index has significantly higher risk-adjusted returns as reflected by the Sortino Ratio.  The better risk-adjusted returns of the Discretionary Traders Index stemmed from lower downside volatility.  Further, the kurtosis and skewness statistics indicate both have extremely fat tails with a strong propensity to surprise in positive returns, with the Discretionary Traders Index having the most extreme numbers for positive surprise.  This shows up in the Discretionary Index run-ups, which show a much wider spread between the largest couple of run-ups vs. the fourth and fifth largest relative to the Systematic Index.   Broadly speaking of the drawdown and run-up analyses, one could describe the Discretionary Traders Index’s returns stream as slower, smoother and generally less volatile that the Systematic Traders Index.

To further support or rebut the conclusions/observations of this initial analysis, we next examined two sector indexes dominated by the two styles: The Barclay BTOP50 Index as representative of systematic, diversified trend followers and the Barclay Agricultural Traders Index as representative of discretionary, fundamental niche strategies.  The BTOP50 Index returns are somewhat greater with very similar volatility levels resulting in slightly better risk-adjusted returns.  Both indexes have positive kurtosis and skewness numbers.  However, the shape of the return distribution indicates that the Agricultural Traders Index has a significantly higher propensity for much larger surprises. The drawdown and run-up data show that the drawdowns of the Agricultural Index are longer than the BTOP50 and moderately deeper as well.  Run-ups are also greater and of slightly longer duration.

Lastly, we reviewed the SG MTI Quantitative and Discretionary Sub-Indexes.  There is minimal difference in the returns and the Discretionary Index shows only a moderately superior risk-adjusted returns.  The shape of the curve indicates that the Quantitative Index has fatter tails, while both have a moderate propensity for positive surprise.  Drawdowns are slightly steeper and of longer duration for the Quantitative Index.  The run-ups are greater for the Discretionary Index, while the run-up durations are similar.

Overall, there were insignificant differences in the rates of returns of the two styles – systematic and discretionary.  The discretionary style yielded moderately better risk-adjusted returns. The discretionary style produced larger run-ups but with wider dispersions. The drawdown analysis was more mixed, but the systematic style had modestly larger drawdowns of shorter duration. The net result is that the shape of the return distributions showed that the discretionary style has much fatter tails with a greater propensity for positive surprise.  Again, the discretionary style return streams could be described as slower, smoother and generally less volatile that the systematic style with the net result of yielding modestly better risk-adjusted returns.

Returning to the question of whether the benefits of diversification extend to styles, the answer is yes.  While the favorable impact varied between the three pairs of indexes analyzed, the 50/50 blend of the two styles improved risk-adjusted returns, smoothed the shape of the distribution curve, and produced the smallest drawdowns across the board.

So while we approached this research as a comparison between systematic and discretionary trading styles, the answer to the question of which style is superior, is both. What seems clear is that systematic and discretionary strategies earn returns through different drivers and are relatively non-correlated, so by combining the two approaches on a portfolio level investors can both improve returns and reduce risk.

About the Author

Mike Dancey, CAIA, is VP of Institutional Services and Head of Research at Managed Account Research Inc.  He consults on CTA selection, due diligence, portfolio construction and integration.