Dynamic Futures Trend Portfolios Expanded

If you’ve watched the performance of Managed Futures and CTA for the past few years, you’ll realize that the large majority of Advisors are stuck in a sideways pattern of returns (see Newedge Managed Futures Index below). This is most likely caused by the lack of trends, or reversals in trend, in the interest rate markets. For the past 30 years, prices have been rising overall, but interest rates in particular (prices rise when yields drop). CTAs have allocated a large part of their portfolios to interest rates, for good reason. They have been the biggest winners month after month, year after year, but no longer. We’re not saying that they won’t be a source of good returns sometime in the future, but for now they are dragging down performance.

2 Newedge Futures Index

Combine that with the traditional method of allocating assets in a futures portfolio. The basic premise is that we can’t tell in advance which markets will trend or perform better than others, so we allocate a fixed amount of equal risk to all futures markets that have a long-term history of success. And, because prices have risen dramatically over the past 30 years, that’s most markets. Typically, a large portfolio would divide the investment as equally as possible into four or more sectors (subject to liquidity), for example, 25% interest rates, 25% equity index, 25% FX, and 25% commodities (metals, energy, some agriculture). The more aggressive managers may have allocated up to 50% in interest rates, which served them well for a long time.

This method of portfolio construction uses fixed allocations. That is, within the interest rate sector they would allocation, say, 5% to Eurodollars, 5% to 5-year Notes, 5% to 30-year Bonds, 5% to Eurobobls, and 5% to Eurobunds. Again, trying to equally weight the risk of all assets to maximize diversification and avoid guessing which one will perform best.

And all that is very sound and good financial practice, up to a point. The point is where it stops working, and it hasn’t worked since of 2008 (a spectacular year for most CTAs). There was also a difficult period from 2003 to 2005 that doesn’t seem to appear on the Index. But then there may be some ex post selection. When is it time to rethink this approach? We think the time is now, perhaps yesterday.

A New Approach

The object is to trade those market that are making money and avoid those that are not. Of course, that’s everyone’s goal, but they don’t seem to do that in Managed Futures. Our stock portfolio program to takes a similar approach out of necessity. If you track hundreds of stocks and you can only trade a small number based on your investment size, then naturally you’ll need to select them using some sort of algorithm. Remember, for KaufmanSignals, everything is automatic, so we can’t look at “value.”

Stock selection seems to work well when we rank by multiple performance criteria, that is, if the stock has a history of doing well using a specific strategy, and is doing well recently, then it qualifies as a candidate for the portfolio. Why not use the same method for futures? The only significant differences are the far fewer liquid markets (a few dozen instead of hundreds as in stocks), the need to recognize the strong correlations in some sectors and, of course, the leverage. To solve these issues, we took the following approach:

  1. Limit the number of markets that can be drawn from any one sector
  2. Don’t trade any markets in a sector if none of them qualify with respect to performance
  3. Don’t leverage up any sector that does not have enough markets allocated

While this may pose a more complicated implementation problem, it is certainty an achievable task. It does mean that, when markets are going nowhere, or in a losing phase, fewer markets will be selected for the portfolio. We don’t have a problem with that and expect that investors won’t either. It will also tend towards “concentration,” that is, we may have many more equity index markets selected than interest rates during some intervals. But those index markets will have equal risk and be diversified, although their combined performance will be more important than interest rates at that time. The key question is, does performance ranking work for futures as it does for stocks?

The answer yes. It works better than the fixed allocations and it’s working better now, which is just as important. Before looking at results, remember that futures portfolios are engineered to have a target volatility. That means each program accepts a certain risk in order to achieve a high return. To do that it uses leverage. The typical Managed Futures program targets about 14% volatility (effective leverage of about 15:1), although that can vary from 8% to 18%. “14 vol” (as it is called) means there is a 16% chance of a 14% loss over the performance history (1 standard deviation), and a 2.5% chance of a 28% loss. In exchange, expectations of return can be high. It’s always the case of accepting a risk to get a reward. You can reduce this risk simply by investing less or reducing the target volatility. When you reduce the vol by 50% you also reduce the potential returns by 50%. The good news is that, if the (simulated) performance history goes back 20 years or more, then the 14% loss usually occurred only twice during that entire period. But it could occur anytime in the future. Because of the inherent leverage, futures traders are more realistic about risk.

Performance

There is no standard for Managed Futures performance, although the Newedge Nelson report, Hedge Weekly, and CTA Intelligence post returns for a large number of managers. Barclay’s also posts daily average CTA returns and those of the BTOP 50, often used as a benchmark. Most of the results look the same as the chart at the top of this report.

Our dynamic allocation process seems to work during the good times, when interest rates were steadily rising, as well as more recently when futures markets have been erratic and show few trends. We applied this to the our two futures strategies, Trend and Divergence. Note that these have very different internal structures. The Trend program hold trades for a long time, scaling in and out as the trend gets stronger and weaker. The Divergence program holds trades for about 5 to 8 days and the size is determined on entry. The success of the same ranking method on two very different strategies is a welcome sign of robustness. Below are the long-term, simulated results of the dynamic portfolio method applied to these two strategies.

Trend Futures

Divergence futures

Divergence Futures US-EU

The Trend program results go back to 1988 and the Divergence to 1992. Each show three portfolios that differ in the maximum number of markets that can be selected. Each market is initially allocated $25,000, so a 10 market portfolio requires an investment of $250,000. Only the largest US and European markets are used. A much broader set of markets are available for Institutional portfolios. All NAVs are shown with a 14% target volatility, so that somewhere on the curve you’ll find a drawdown of 15% to 20%.

It is interesting that the two strategies have somewhat different performance patterns. When used together, the recent drawdown in the Divergence program is offset by the gains in the Trend program. Diversification is always a benefit.

Beginning June 2014, these dynamic portfolios will replace the traditional fixed ones for our futures traders. We track the daily returns and post them at the end of each month.

Copyright 2014, KaufmanSignals.com, 20140604

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top