Macro Trend Rotation

As of January 1, 2019, I am adding a new model portfolio to the publication. It is called Diamond-Zircon Rotation model portfolio. Over the last several months, many subscribers have shifted to Zircon from Diamond in response to my comments about the negative Macro (long-term) trend in stock prices. Both Diamond and Zircon are based on the same set of signals (which I call “D5”), but Diamond is more aggressive than Zircon; its gains and losses can be expected to be larger.

While the original intent of the Focused 15 Investing model portfolios was that each subscriber would select a model portfolio with a risk profile that is appropriate for their risk tolerance and time horizon, many subscribers have shifted model portfolios over time. This response from subscribers is perhaps inevitable. I have been open about the risks I see in the market and it is easy to rotate from one model portfolio to another. Given this usage of the publication, I sought to develop a more disciplined and robust way of rotating between Diamond and Zircon based on the larger market trends.

The Diamond-Zircon Rotation model portfolio uses a new signal, what I call the Macro Rotation Signal. To create the Macro Rotation Signal, I adapted a subset of the existing algorithms to focus on just the long-term price trend. By doing this, I was able to use algorithms that have been used real-time for about 10 years and have roughly 100 years of history.

The Macro Rotation Signal can be applied to various pairs of more aggressive and less aggressive model portfolios, like Diamond and Zircon. I am also reviewing the advantages of applying the Macro Rotation signal to Sapphire and Onyx (which consists of low volatility sector ETFs). 

The testing suggests that rotating between Diamond and Zircon can reduce weekly losses. Rotation provides the comfort factor of not using the aggressive ETFs (“DDM” in particular) in a negatively trending stock market.  It also enables a single model portfolio to have a wider range of risk exposures without having to trade a large number of ETFs. 

The result of this analysis is the Diamond-Zircon Rotation model portfolio (sg227). The individual existing Diamond and Zircon model portfolios have not changed and can still be used as before.

Now is a good time to introduce the Diamond-Zircon Rotation model portfolio because a key price trend measure (mentioned below) for the DJIA recently turned negative, which indicated that a shift to Zircon would be appropriate. I have spoken to subscribers to notify them about the availability of the Diamond-Zircon Rotation model portfolio.

For those following Diamond-Zircon Rotation, there may be some reduced return over the short-term should market prices begin a long-expected bear-market rally. Yet, the bottom of the current market cycle does not appear to be at hand and future market declines can be sufficiently large to justify the shift to the less-aggressive Zircon at this time.

The Diamond-Zircon Rotation model portfolio can be expected to produce slightly lower returns than Diamond over the long-term (by about 1.5 percentage points per year, annualized) but the variability of returns is also slightly lower. Again, a key advantage is reduced aggressiveness when the market is in a long-term vulnerable trend.

I tested the Macro Rotation Signal over three different time horizons. Because of data limitations, I could not do all tests over the longest time horizon. The first analysis covers 100 years of price history for the DJIA. This analysis focuses on the benefit of using the Macro Rotation Signals to move out of the market during times of vulnerability.

The Macro Rotation Signal is based primarily on two elements, a) the status on the Macro MRI, and b) the overall price trend of the DJIA using a technique similar to a widely used technical analysis tool called Moving Average Convergence Divergence (MACD). The MACD-related element is what turned negative over the last few weeks. The table below shows the performance statistics for the 100-year period for a simple model portfolio consisting of holding the DJIA in positively trending markets and holding cash in negatively trending markets.

These statistics are based on price change only; they do not include dividend return. Nor do they include any return for cash that might be held. The DJIA – Buy-and-Hold is the benchmark for this analysis; it is what would be achieved with no Macro Rotation.

(1918 – 2018)
Rate of Return (annualized, higher is better)
Variability (annualized standard deviation of weekly returns, lower is better)
Ratio of Rate of Return to Variability (higher is better)
DJIA – Buy-and-hold
DJIA using Macro Rotation Signal to get out of market and hold cash

Thus, the Macro Rotation Signal improves returns and reduces variability – both are positive changes. This result is seen in a higher Ratio of Rate of Return to Variability measure.

The average number of rotations per year is 1.8 over the 100 year period. Of course, these are not spread evenly over the period. Instead, they tend to be clustered. The figure below shows the DJIA price (log scale) in brown. The green line reflects the price change of our simplified rotation portfolio.

The better return for the simple model portfolio using the Macro Rotation Signal is based on identifying the negative trends in the market and avoiding losses. Subscribers may recognize this chart from the initial descriptions of the investment approach.

The following chart is of the same series but with a shorter time span, beginning November 1931, at the bottom of the stock market during the Great Depression. This time span highlights how Macro Rotation Signal works during different market environments.

The approach does best when there are market losses. It does not keep up with the buy-and-hold DJIA during strong trending markets, which are circled below.

Over this shortened time period, the following are the statistics for rate of Return / Return Variability (higher is better).

  • For the DJIA (buy and hold):       0.41
  • Using the Macro Rotation signal: 0.67

During the late stages of an ascending market the rotation signal produces a shallower slope than the buy and hold. This indicates that the Macro Rotation Signal reduces risk prematurely. Subscribers will recognize the tendency of the MRI-based approach to underperform in the late stages of an ascending market, which we see in the figure above. I adjust for this in the design of the model portfolios (e.g., Diamond, Zircon). The most challenging period marked “A” on the graph above.

The table below compares simulated returns for the Diamond-Zircon Rotation model portfolio to Diamond and Zircon over the period 1966 through the end of 2018, which includes period A indicated above. These simulations do not include dividend return or any return to the cash held in portfolios because of data limitations over this long time horizon.

1966 - 2018
Rate of Return (annualized, higher is better)
Variability (annualized standard deviation of weekly returns, lower is better)
Ratio of Rate of Return to Variability   (higher is better)
Simulated Diamond (D5 Signal Set using ETF “DDM”)
Simulated Diamond-Zircon Rotation
Simulated Zircon (D5 Signal Set using ETF “DIA”)
DJIA Buy-and-Hold

Zircon’s annualized rate of return is slightly higher than the DJIA Buy-and-hold, but the variability of returns is less than half. This boosts the ratio of return to variability to a high level.

One can see that the Diamond-Zircon Rotation model portfolio has much better performance than Zircon and slightly lower performance than Diamond. For many subscribers with longer time horizons, Diamond might be the right choice. Yet, for those with shorter time horizons or uncomfortable holding aggressive ETFs in negatively trending markets, the Diamond-Zircon Rotation portfolio might be the right choice.

On the three charts below, I show the Macro Rotation Signals for the period 1980 to the end of 2018. One can see the major tops and bottoms of the market, which are indicated. It is at these point that the Macro Rotation Signal adds value.

One can see from the figure some minor tops and bottoms. The Rotation signal is less likely to add value for these. This is particularly true during period A (from 1985 through 1994). Macro Rotation shows considerably greater effectiveness in the period of 2000 to 2018 than than during period of A. Going forward, we want to be prepared for many environments so it is important to consider the success of the approach during the difficult period A.

One can also see in the charts that some of the Macro Rotation shifts are clustered in time; a rotation in one direction is reversed just a few weeks later. To date, I have not identified how to reduce this clustering further without reducing returns.

Trading in the Diamond-Zircon Rotation portfolio is simple and, I believe, mitigates this issue. It holds only two ETFs, ETF “DIA” for the DJIA with no leverage. ETF “DDM” for the DJIA at two times leverage. It does not hold ETFs for long and short-term bonds. Instead, cash is simply help in one’s account. At this time, bond ETFs do not have high return so excluding them does not result in a big negative impact on returns.

As mentioned, I am reviewing applying the Macro Rotation Signal to other pairs of model portfolios. These might have superior performance characteristics while still being easy to implement.

Of course, subscribers are free to use any of the model portfolios. Please review the statistics and the various risks and select the model portfolio appropriate for you.