((Note: this the fourth update to a post which first appeared in 2019. It is instructive in describing the approach and priorities which I bring to my strategy development process.)

"The most recent incremental upgrades were released in mid-2019 and are described in considerable detail below together with the resulting outcomes. In each case, the focus has been on improving control over the upper bounds of portfolio volatility, improving the consistency of annual returns, and on reducing fund turnover."

## The Cliff Notes Version (or TL;DR)

- Adaptive Global was introduced in mid-2018. Minor adjustments were made in mid-2019 to improve volatility management at the cost of a slight decrease in performance. The Performance Report for the Adaptive Global strategy reflects near-real-time performance since mid-2018 and real-time performance since July 2019.
- The Performance Report for the Adaptive Income strategy reflects real-time performance since August 2019. The 2019 update incorporated a major redesign of the strategy.
- Two full market cycles have occurred since the last changes were incorporated into the TAAStrategies during 2019.
- I demonstrate a trade-off between developing strategies for long term safety versus short term returns.

## A Short History of the TAAStrategies

Tactical Asset Allocation (TAA) has been a passion since I began studying TAA in 2011. Early in my research, I was content to study and experiment with strategies developed by others. However natural curiosity coupled with decades of software development experience led me to begin tinkering with spreadsheets. Tinkering was soon followed by the construction of ever more sophisticated TAA strategies.

I began using tactical strategies for a small portion of my family portfolio in 2012 and my own strategies in 2014. I gradually added to our investment as my confidence grew stronger based on observable results.

In early 2014, I completed my first large-scale Tactical Model which provided a platform for developing and testing increasingly sophisticated tactical strategies. A year later, I developed the first Market Conditions Model and began integrating it with the Tactical Model.

I began publishing Tactical Global Core and Satellite strategies in early 2016. Ultimately, the Core and Satellite strategies would be combined into a more sophisticated Adaptive Global strategy in 2018.

A third generation Tactical Model was completed in mid-2018. The centerpiece was Adaptive Dynamic Momentum, an innovative and powerful momentum strategy which required nearly 5 years to develop. Adaptive Dynamic Momentum uses sophisticated statistical techniques to dynamically discover and validate security trends for each fund in our baskets, each period. While requiring significant computing resources; the improvements in both higher return and reduced risk is significant.

Adaptive Dynamic Momentum was incorporated into the flagship Adaptive Global Strategy in mid-2018.

Adaptive Dynamic Momentum, which is particularly adept at recognizing sustainable shifts in momentum, also provided the foundation for the development of Adaptive Income, our most conservative strategy, released during the summer of 2018.

Early-2019 saw the development of a second generation Adaptive Income strategy with an entirely new fund basket and major improvement in performance metrics.

The upgrade of Adaptive Income was followed by the introduction of the Adaptive Innovation strategy which provided exposure to the leading edges of innovation without the high volatility. Adaptive Innovation was discontinued in late 2022 when the risk:reward became less favorable when compared to the Adaptive Global strategy.

The most recent incremental upgrades were released in mid-2019 and are described in considerable detail below together with the resulting outcomes. In each case, the focus has been on improving control over the upper bounds of portfolio volatility, improving the consistency of annual returns, and on reducing fund turnover.

I also demonstrate a trade-off between developing strategies for long term safety versus short term returns.

## Strategy Testing

The Adaptive Global and Adaptive Income strategies were developed using bear/bull datasets from 2007 through 2017. Out of sample validation of risk and return metrics using bear/bull datasets from January 2000 through September 2007 were statistically comparable to the 2007-2017 period.

The fund baskets for the TAAStrategies are constructed from indexed Exchange Traded Funds (ETFs) with just two exceptions, an Open End Fund and a Closed End Fund, both with long history. A number of the ETFs we use were not created until 2007+. In each case, we infill using predecessor Open End Funds (OEFs) for which the indexing and/or subclass is substantially similar to the ETF.

I have been asked if it is possible to extend backtests to the 1970’s. While a few publishers attempt this; I believe it is not possible to produce credible results for any but the most basic TAA strategies using a handful of classes/sub-classes due to the lack of funds with substantially similar indexing and/or classification. Doing so would force me to stretch the term "substantial" far beyond my comfort level.

## Tactical Model upgrades

Upgrades to the Tactical Model add to the capabilities of the strategy development and testing platform. They do not affect the strategies.

**Limited Portfolio Volatility Weighting**

Summary: I developed an enhanced version of Volatility Weighting which allows us to specify a limit for the expected portfolio volatility. When expected portfolio volatility exceeds the cap, allocations are shifted from high volatility funds into lower volatility fixed income funds until the limit is met. This has the effect of lowering the volatility of monthly returns as well as reducing the depth and duration of drawdowns.

- The Performance Summary included with each performance table includes two important measures of volatility. The first is the “Standard Deviation of Monthly Returns” which measures the degree to which monthly returns deviate from the average return. The lower the number, the more consistent are the monthly returns.
- The second measure is the “Ulcer Index of Daily Drawdowns” which is a measure of the length and severity of downside (unfavorable) volatility. “
*Ulcer Index measures the depth and duration of percentage drawdowns in price from earlier highs. The greater a drawdown in value, and the longer it takes to recover to earlier highs, the higher the UI. Technically, it is the square root of the mean of the squared percentage drawdowns in value. The squaring effect penalizes large drawdowns proportionately more than small drawdowns.*” I began including this some months ago at the suggestion of a subscriber and it has quickly become one of my goto measures of downside risk. As a point of reference an Ulcer Index of 5 and below is considered to be very low. - Fund volatility, as measured by annualized Standard Deviation (StDev), is not constant. For example, MDY, a mid-cap index fund used in Adaptive Global, has a full cycle average 20 day StDev of 21.8% which ranges from a weekly low of 4.9% to a high of 107.0%.
- The distribution of volatility is heavily lop-sided across 1210 weeks of history with 777 weeks (64%) below the average and 433 weeks (36%) above the average.
- As a general rule, volatility declines in strong trends and rises when trends weaken or change direction.
- Significant periods of low volatility in MDY include 09/15/17 through 01/26/18 and 11/08/19 through 01/24/20 where the 20 day StDev dropped below half the average. Both periods show a high rate of low-risk appreciation in MDY.
- Significant periods of extreme high volatility in MDY include the periods from 09/19/08 through 04/24/09 and 03/13/20 through 07/02/20 where 20 day StDev was more than twice the average. (The high of 107% occurred during the Covid Crisis.)
- Clearly, we could pull the average volatility significantly downward if we could lop off some of the occurrences which are above the average 21.8% and we can do that by filtering out volatility which is trending higher.
- Two of our Strategies (Adaptive Global and Adaptive Income) employ Volatility Weighting which allocates portfolio weights inversely to volatility. Once the required number of funds are selected based on momentum, the funds with the lowest volatility get the largest allocations (within limits). This moderates the expected volatility of the strategy.
- I developed the Limited Volatility Weighting algorithm to further moderate portfolio volatility while maintaining returns. First, the fund set is selected based on momentum. Next, the initial allocation is performed using Volatility Weighting and the resulting portfolio volatility is calculated. Finally, if the estimated portfolio volatility exceeds the target percentage specified for the Strategy; the algorithm begins shifting allocations away from the highest volatility funds into lower volatility fixed income funds until the target is met.
- Limited Volatility Weighting is employed only when high levels of volatility in one or more selected funds causes expected portfolio volatility to exceed the cap. The effect is similar to that of putting a speed governor on a vehicle. The cap for each Strategy is commensurate with the Strategy objective and is selected based on the mid range of compromises between lowering volatility and lowering returns which are a function of both advances in price and reduction in the depth and duration of drawdowns.
- Limited Volatility Weighting has been adopted for all of our Strategies because it lowers both the StDev of Monthly Returns and the Ulcer Index. This improves the consistency of returns at the expense of a slight reduction in returns.

**Position Optimization for LTCG**

- Fund selection based purely on momentum selects the best performing funds … period. However, there are times when a fund we already own is performing strongly but does not quite make the cut. We end up selling the fund we own to replace it with the higher performing fund.
- Position Optimization favors holding existing fund positions over trading into new positions when the existing fund meets high thresholds of momentum and confidence. The second generation algorithm incorporates improvements that can increase average holding periods.
- It should be noted that while optimization succeeds in extending holding periods, those periods are often not long enough to qualify as Long Term.

## Strategy Upgrades during 2019

Returns in the Tactical Adaptive Strategies lagged the market for much of 2018 and 2019. This was the result of severe chop in the equity market which failed to produce an enduring trend among the funds in our baskets. However, our drawdowns also lagged the market by half.

Adaptive Global already had a broad basket of 24 funds so there was no point in altering the basket. I did; however, take a look at what would have been required to improve Adaptive Global's return for 2018:

- Remove the market condition constraints
- Force the Adaptive Dynamic Momentum to use shorter momentum lengths
- Use Equal Weighting instead of Volatility Weighting

Had we made these changes, we would have seen a 4.20% return for January through October instead of 3.44%. However, the cost of juicing the performance in 2018 would have been devastating across a full market cycle: CAGR droped from 15% to 11%, the Ulcer Index rose from 3.7% to 8.1%, the StDev Of Monthly Returns rose from 9.5% to 12.5%, Max Report Drawdown rose from 8.7% to 20.6%, and the Up/Down efficiency dropped from 229% to 156%.

I don’t believe that is the type of long term performance and safety any of us are seeking. Even the best of investment strategies has times when it underperforms. Changing strategies to chase performance is seldom rewarding but at least we can play "what if?".

**Adaptive Global (July-2019)**

The primary benefit of minor changes to the Adaptive Global strategy was reduction in both Standard Deviation of Monthly Returns and the Ulcer Index:.

**Fund basket and condition eligibility**: no change was made to fund baskets or condition eligibility**Fund selection**: no change to fund selection method (Adaptive Dynamic Momentum)**Weighting of selected funds**: Volatility Weighting was replaced with Limited Portfolio Volatility Weighting which is employed to cap the expected total portfolio volatility when high volatility funds (especially equities and commodities) are used. This slightly reduces both risk and return.**Long Term Position Optimization**: Upgraded which slightly improves holding periods and performance.**Results**: CAGR decreased slightly, Max Monthly Drawdown decreased, and the Up/Down Ratio decreased fractionally. The Ulcer Index declined and Standard Deviation of Monthly Returns declined reflecting improved consistency of returns.

**Adaptive Income (August-2019)**

The primary benefit of changes to the Adaptive Income strategy was a broadening of the fund basket and reduction in both Standard Deviation of Monthly Returns and the Ulcer Index:

**Fund basket and condition eligibility**: the number of fixed income ETFs was increased from 5 to 6 to broaden the basket to include an investment grade ETF.**Fund selection**: no change to fund selection method (Adaptive Dynamic Momentum)**Weighting of selected funds**: Adaptive Income used a single selection with a 100% weighting to the best performing fund. Limited Portfolio Volatility Weighting is now employed to cap the expected total portfolio volatility when higher volatility funds (especially high yield) are used. This forces the allocation across a second, lower volatility fund when the cap is exceeded.**Long Term Position Optimization**: Upgraded which slightly improves holding periods and performance.**Results**: CAGR increased slightly, Max Monthly Drawdown increased, by 1%, however Max Daily Drawdown remains unchanged, and the Up/Down Ratio rose sharply. The Ulcer Index declined slightly and Standard Deviation of Monthly Returns declined sharply.**Note**: the increase in Max Monthly Drawdown is directly attributable to the addition of the investment grade ETF. Because the Max Monthly Drawdown remains exceptionally low, this appears to be a reasonable trade-off given the size of other improvements.

## View Across Full Market Cycles

We can now measure performance across four market cycles including two market cycles following the changes incorporated into the TAAStrategies during 2019.

## Conclusion

The experience I acquired over many years of investing in a variety of managed portfolios and hedge funds taught me the importance of evaluating risk and reward over a full market cycle rather than chasing recent performance. This provides a more balanced perspective on risk and reward.

## Earl Adamy

## Ready to subscribe to the TAAStrategies?

*"I believe in my experience, methods, and approach strongly enough that if you aren't entirely happy within the first two months, I'll return the entire Subscription Fee."*

## Prefer to learn more about the TAAStrategies?

Exceptional results are due entirely to the complementary strengths of our Market Conditions Model and our Tactical Model.