Tactical Adaptive Income Strategy Whitepaper

Updated February 12, 2020

(Incorporates Limited Portfolio Volatility, addition of AGG to fund basket, and extended backtesting to 2000)

TAAS offers three Tactical Adaptive Strategies to self-directed investors:

  • Tactical Adaptive Global provides broad exposure to domestic and international equities, fixed income, commodities and precious metals. Tactical Adaptive Global is intended for use as the primary strategy in a tactical asset allocation portfolio.
  • Tactical Adaptive Income invests across a diversified basket of fixed income subclasses designed to provide a high level of income coupled with low risk.
  • Tactical Adaptive Innovation provides reduced-risk, focused exposure to innovation including: big data and analytics, nanotechnology, genetics, medicine and biotechnology, networks, energy and environment, robotics, 3-D printing, bioinformatics, and financial services.

This paper describes the design and principles which underlie the Tactical Adaptive Income Strategy in considerable detail. It is written for subscribers and prospective subscribers who seek to more fully understand the strategy.

The Appeal of Fixed Income

There are at least two good reasons to consider a tactical fixed income strategy to complement a broad market tactical strategy:

  • Eating distributions - some investors would prefer to withdraw distribution income for living expenses than “eat the seed corn”.
  • Tactical diversification - it is exceedingly unwise to put all your eggs in one basket, even when that basket looks far better than all the other baskets on display.

Before moving on, let's set the stage with an explanation of Tactical Asset Allocation.

Tactical Asset Allocation Defined

Tactical Asset Allocation (TAA) is among the best investment tools available for navigating Full Market Cycles (see What Is A Full Market Cycle? Why Should I Care?). While TAA tends to lag in late bull markets, it offers opportunities for higher Compound Annual Growth Rates and lower Maximum Drawdowns across a full bull/bear market cycle. Among the greatest strengths of TAA is its mechanical, rules-based approach, which not only keeps the portfolio attuned with market conditions but reduces the anxiety of managing the portfolio.

Perhaps the single biggest distinction between Tactical Asset Allocation and Modern Portfolio Theory is that while Modern Portfolio Theory seeks to reduce risk by spreading it across several asset classes, Tactical Asset Allocation seeks to reduce risk by cutting it.

Please see What Is Tactical Asset Allocation? How Does It Improve Returns? for a more complete discussion of Tactical Asset Allocation.

Tactical Process For Adaptive Income

  • Calculate trend scores (TScores) for each fund in the basket. Adaptive Dynamic Momentum (see Adaptive Dynamic Momentum - How Does It Improve Trend Identification), the trend identification methodology used by our Tactical Model, provides the critical pathway to finding (and sticking with) winners while avoiding (and/or quickly discarding) losers. During the development process, the Tactical Model quickly identified the fact that a shortened primary trend length is most suitable for achieving the low risk objective of Adaptive Income.
  • Order and select the funds according to TScore. This is where we give priority to holding existing positions when conditions are favorable, and manage the holding and reentry requirements imposed by Open End Funds.
  • Allocate a portfolio percentage to each selected fund using Volatility Weighting

Building A Fixed Income Fund Basket

Fixed income is viewed by many equity-focused investors as the backwater of the markets although the bond and credit markets are in fact very challenging.

The Fed’s aggressive management of monetary and interest rate policies have wrought critical long term changes in the credit markets. While the credit cycle will continue to play out, the days of 4% to 6% Treasury yields are unlikely to return during our investing lifetime absent a resurgence of inflation. Adapting to reduced yields means returns must be earned by exploiting more corners of the credit markets as well as shorter interest rate cycles.

I went through a process of testing nearly 40 different  fixed income funds before settling on a basket of six funds including corporate high yield, municipal high yield, corporate senior loans, government mortgage backed securities, corporate investment grade, and short term Treasuries. Five are Exchange Traded Funds and one is an Open End Fund.

  • WHIYX (1997): Ivy High Income Fund (Class Y) - Actively managed OEF
  • HYD (2009): Market Vectors High-Yield Municipal Bond - Indexed ETF
  • BKLN (2011): Invesco Senior Loan ETF - Indexed ETF
  • MBB (2007): iShares MBS ETF - Indexed ETF
  • AGG (2003): iShares Core US Invest Grade Bond Fund - Indexed ETF
  • SHY (2007): iShares Short Treasury Bond Fund - Indexed ETF

Fund inception date and liquidity are among the important criteria used in fund selection. Ideally, each fund has a history which goes back into 2007 prior to the start of the Great Financial Crisis. This allows us to capture performance during a full market cycle. Even better is history back to 1999 allowing us to capture performance during two full market cycles.

Relatively few ETFs have history back to 2007 and even fewer back to 1999. Fortunately, in using indexed ETFs we are able to identify Open End Funds with same/similar indexing and/or characteristics. Two Open End Funds are used to supplement full market cycle history back to 2007: Invesco Senior Float Rate Bank Loan Fund and Nuveen High Yield Muni Bond Fund.

Market Conditions Model

I tested the Market Conditions Model to constrain eligible fund selections during Hostile periods. Doing so provided a tiny increase in return at the expense of a doubling in monthly maximum drawdown. While the Market Conditions Model focuses on longer term cyclical trends, the Adaptive Income Strategy requires short term trend identification.

This is the only TAAS Strategy which does not use the Market Conditions Model; however we include below performance statistics for all three market conditions to demonstrate the historical consistency of strategy performance.

Identifying The Most Sustainable Trend

Most tactical algorithms focus on identifying funds with the strongest momentum. Adaptive Dynamic Momentum focuses on identifying the trend for each fund with the best combination of momentum and sustainability. This is accomplished by performing a computationally intensive examination of trend momentum for each fund as well as the quality of the trend which is expressed as a TScore (short for trend score).

The research and development of Adaptive Dynamic Momentum (see Adaptive Dynamic Momentum - How Does It Improve Trend Identification) required nearly five years; however, it is a quantum leap in trend identification which opens up new avenues for development of tactical strategies. While our Tactical Model provides several algorithms for trend identification, it is Adaptive Dynamic Momentum which consistently outshines all others. It provides the critical pathway to finding (and sticking with) winners while avoiding (and/or quickly discarding) losers.

The Tactical Model quickly identified the fact that a shortened primary trend length is most suitable for our basket. Adaptive Income uses an average primary trend length1 of just 8 weeks.

1 The optimum primary trend length identified by the Tactical Model varies by fund by week across a broad range of possible lengths so the “average” is not indicative of the actual primary trend length used for any individual fund or week. For example, Fund A may have a primary trend length of 17 for the same week in which Fund B has a primary trend length of 4, Fund C has a primary trend length of 9, and so on. The Tactical Model also identifies and applies a secondary trend length by fund by week. It is the combination of the primary and secondary trends, together with some additional filtering, which ultimately determine the Final TScore for each fund each week.

Ranking and Selection

Once the TScores have been calculated for each fund, the funds are filtered for special conditions.

One of the newest features in our Tactical Model reduces the number of trades. Before replacing an existing position with a higher trend scoring fund, the Model checks to see if can hold the existing position without compromising expected performance. For the Adaptive Income Strategy, this slightly reduces trades while extending the average holding period for each fund by a little over 10%.

Open End Funds are also checked for restrictions on reinvestment.

Finally, the funds are ordered from best to worst and selected for inclusion in the rebalance.

Allocating Funds

Generally, when building tactical strategies, I prefer to work with a sizeable basket of funds and spread the allocation across several of the best performing funds for diversification. Extensive testing with the fixed income basket revealed that the best results are obtained by selecting a single fund. In fact, including the “2nd best” fund in a combined allocation actually reduced returns while increasing drawdowns.

I have also come to prefer Volatility Weighting over the commonly used Equal Weighting which spreads the investment equally over each of the selected funds i.e. 4 funds at 25% to each fund.

Volatility tends to decline when a trend is strengthening and rise when a trend is correcting or changing direction. Rising volatility accompanied by an upward change in trend is “good volatility” while rising volatility accompanied by a downward change in trend is “bad volatility”.

In studying the characteristics of fund volatility, I learned that volatility is skewed sharply to the low side. That is, for a given fund, there will be many more instances where volatility is below the average and relative few instances where volatility is above average. AGG for example has an average volatility of 4.8% but of 160 months, 132 (83%) are below 4.8% and only 28 are equal to or above 4.8%. Not surprisingly, the highest volatility at 34% was in October of 2008.

Volatility Weighting spreads the allocation inversely to volatility2 with limits to prevent a funds from taking an extreme allocation. The six funds in the Adaptive Income basket have average volatilities ranging from 1.4% to 6.9%. One would expect that the 1.4% fund would receive a large allocation and the 6.9% fund would receive very little. But higher volatility funds tend to deliver higher returns, have higher TScores, and are therefore selected more frequently. To take that one step further, funds with extremely high volatility are quite likely to have poor TScores due to poor momentum and sustainability.

Adaptive Income selects just one fund. So it is going to get 100% of the allocation regardless of volatility? Not quite. I developed an algorithm I refer to as “Limited Portfolio Volatility” to provide more even more control over volatility.

Limited Portfolio Volatility permits the setting of a limit on expected portfolio volatility. The algorithm calculates expected portfolio volatility by multiplying and summing each fund allocation by its volatility. If the expected portfolio volatility exceeds the limit, it begins shifting allocations from high volatility funds to lower volatility funds with high TScores.

In the case of Adaptive Income, which has just one fund selection, we introduce a second, high TScore, lower volatility fixed income fund to absorb enough of the allocation to reduce expected portfolio volatility to the target. This has the effect of dampening strategy volatility.

2 20 to 30 day annualized

Scoring The Results

The Tactical Model is run one full market cycle beginning October 2007 to calculate Compound Annual Growth Rate (CAGR), Maximum Monthly and Daily Drawdown, and a host of other statistics including monthly and annual returns.

By way of benchmarking the full market cycle, the Vanguard Total Bond Market Index Fund, which invests 70% in government bonds and 30% in corporate investment grade bonds, shows a CAGR of 4.0% and a MaxMD of 4.0%.

We are able to extend the backtest two full market cycles by using Open End Funds with same/similar indexing and/or characteristics. This not only provides us with how the Tactical Model and fixed income subclasses might have performed during an earlier cycle but uses entirely out of sample data.

Performance Across Market Conditions

As mentioned above, we do not use the Market Conditions Model because the Adaptive Income Strategy relies upon very short term trend identification. Never-the-less, we are able to break out strategy performance by market condition for the two full market cycles from January 2000 through January 2020:

  • Favorable condition: CAGR @ 9.8%, MaxRD (3.9%), Up/Down Ratio 530%
  • Balanced condition: CAGR @ 7.8%, MaxRD (3.6%), Up/Down Ratio 619%
  • Hostile condition: CAGR @ 12.0%, MaxRD (2.8%), Up/Down Ratio 479%

Withdrawals

The table’s statistical summary suggests a Sustainable Withdrawal Rate. How is this calculated?

  1. Assume that the Maximum Monthly Drawdown occurs on the day that investment in the strategy takes place. This is generally an investor’s greatest fear.
  2. Multiply the balance remaining after the MMD by the CAGR to calculate the expected return.
  3. The Sustainable Withdrawal rate equals 2/3 of the expected return with the remaining ⅓ available to cover inflation and a safety factor

The calculated Sustainable Withdrawal Rate is well above the yields on all of our funds. How is this possible? It is due to the large capital gains in the high yield funds during the early stages of the bull market (especially 2009). While capital gains are great, we want to identify a rate which is closer to the fund distributions.

I use distribution adjusted closing prices so the discrete distributions are unavailable to the Tactical Model but we can use the Tactical Model to provide more information by testing different withdrawal rates. We also use the past 5 years which avoids including the high returns of the early bull market.

The withdrawal rate which stabilizes the remaining balance is 5.2% during the past 5 years. This is nearly 25% higher than the yield provided by a High Yield fund like HYG with much lower downside risk coupled with the opportunity for significant bull market capital gains. That said, one should not lose sight of the fact that returns during the next 10 years are unlikely to match those of the past 10.

When it comes to withdrawals, I strongly prefer a fixed monthly amount which levels the portfolio volatility risk. Those with RMDs can always adjust or supplement the final withdrawal to meet requirements.

Conclusion

The Adaptive Income Strategy smoothly and effectively transitions between risk on and risk off while delivering exceptional returns.

It combines the use of low-cost, passive index funds coupled and one actively managed fund, with an active management strategy to reduce losses and improve returns. There is a large body of academic research which is both substantive and compelling in making the case for the use of Tactical Asset Allocation to manage all or part of an investment portfolio. With the growing possibility of a bear market in equities and increased volatility in credit markets, investors should consider using Adaptive Income for part of their portfolios to improve returns and reduce risks.

The Adaptive Income strategy presented here can be used as a standalone income strategy or it can be paired with other tactical strategies, such as Tactical Innovation, for portfolio diversification.

Strategy Notes

There are three issues attendant with mixing Open End and Exchange Traded Funds in the same tactical strategy. I have included an Open End fund in this strategy only because it significantly improves performance.

  • Early Redemption Fees: ETFs trade with no minimum holding period. Some brokers impose an Early Redemption Fee (typically $50) on purchases held for less than 90 to 180 days, particularly if purchased through a No Transaction Fee program. Solution: the Tactical Model optimizes fund allocations to extend holding periods wherever practical.; however it will exit if risk conditions warrant.
  • Repurchase Limitation: Some funds and/or brokers impose a 60 day limitation on repurchase following a sale. Solution: the Tactical Model will not repurchase an OEF position for a minimum of 60 days after exiting.
  • Settlement of T+1 versus T+2: ETFs, like all stocks, settle in two business days (T+2) while some OEFs settle in one business day (T+1). An issue arises when selling an ETF, which settles in T+2, to buy an OEF which settles in T+1.  Solution: enter the order to purchase the OEF Market On Close effective for the day following the rebalance, or margin may be used to cover the 1 day difference in timing.

Inclusion of a municipal bond fund has no bearing on taxable versus non-taxable accounts. It is the yield and performance of the municipal bond fund which determines selection and not after-tax characteristics.

All of our Strategies are rebalanced on the last trading day of each month. Rebalance letters are sent out no later than the evening prior to rebalancing. This allows adequate time for review and preparation of orders prior to the final trading day of the month.

I highly recommend the use of Market On Close (MOC) orders which can be submitted any time after the previous close and not less than 15 minutes prior to the close. MOC orders are executed by the exchanges in the closing rotation which is where ETF sponsors handle ETF creation/destruction at the close to insure supply and demand are matched with execution at Net Asset Value. Market On Close is a native exchange order type which is frequently employed by institutions. MOC is available to retail traders on at least one trading platform at  virtually all major brokerage firms. Market On Close generally have a type of "MOC" and a price of "Market".