Dynamic Global Rotation Portfolio — Global Equities

Philosophy of a Trend-Based Approach

Equity investment strategies may be based on either (1.) macroeconomic (or company) fundamentals, (2.) a contrarian approach of buying beating down stocks, or (3.) a technical trend-following approach. Empirical evidence demonstrates that lags in economic data releases, or the numerous difficulties involved in firm valuation, make a fundamentals-driven approach almost irrelevant for an active investor, even given a long-term horizon. Indeed, the stock markets anticipate with several months lead time the future state of the economy while it can take a long time for a company’s value to be reflected in the market — an “attractive” company can just keep getting more “attractive”. Similarly, a technical contrarian or deep-value investor approach, which relies as much on luck as on the analyst’s skill, produces erratic results with an overall trade success rate rarely statistically different from 50/50.  From these observations, we have concentrated on developing a stock trading strategy based purely on market data. In efficient markets, prices should reflect all relevant information in the valuation of a stock. Moreover, of the three approaches, a trend-following strategy should always produce the most consistent portfolio performance, as equity index and stock price movements tend to persist. Recognizing the futility of trying to pick a market bottom or top, our model is designed to seek confirmation of new up trends before proposing a buy-signal (or confirmation of a new down trend before issuing a sell signal).  Maximizing fund performance involves both avoiding a precocious entry into a stock still in a corrective phase as well as not abandoning a position still in an accumulation phase.  Our experience confirms that trends (both up and down) often remain in place much longer than anyone would anticipate due to the presence of irrational, emotional investors.

Strategy Objectives

The strategy seeks to outperform a global equity benchmark index (MSCI All-World) both in bull markets and in bear markets. The investment universe is global equities, with no restrictions across countries and a non-benchmarked allocation. The strategy invests in both national indexes and in sector indexes from all regions (U.S., European, Asian and Emerging Market). Our universe contains around 200 global indexes with an associated “investable” ETF (dollar-denominated, high daily volume, tight bid/ask spread). This universe is chosen to minimize the problem of asymmetric information inherent in trading individual or less liquid stocks, where knowledge of the company and/or insider information is critical. The strategy primarily invests in long equity positions with a maximum holding of 100% of assets (no leverage, no short positions). Remaining fully invested is necessary in order to outperform the indexes in bull markets. The cash balance is discretionary and is a reflection of the overall risk environment. While the objective is to maintain a minimal cash holding (100% long equities), a deterioration of risk conditions signaled by the models allows the manager to raise cash in line with the risk indicator. The strategy depends on a dynamic rotation into outperforming nations or sectors.

In employing the strategy we attempt to remain invested in positions that are both in absolute up trends and are outperforming the global benchmark index. Given the use of index funds, which are diversified by definition, the maximum position size is generally 8% of fund value (or about 12 holdings). Past performance suggests that the average position is turned (or rotated) between 8 to 15 times per year. The fund’s base currency is U.S. dollars. While the strategy does not take outright foreign exchange positions, global investing entails assuming currency risk. Both local currency and dollar-denominated prices for each index are analyzed in our models in order to detect currency weakness in foreign stock positions. For example, the strength or weakness of the peso versus the dollar will be reflected in our dual analysis of the Bolsa Index and the Dow Jones USD Mexico Index. In light of this method, currency hedging (or use of a currency-hedged ETF) is only undertaken if there is a compelling reason to invest in foreign stocks whose underlying currency is negatively-oriented.

Construction of Indicators

The DGR Macro Trading Model, upon which the strategy’s allocation is determined, consists of two components : an absolute indicator (based on market prices) and a relative indicator (based wholly on the WMA prices). The absolute indicator is simply the nominal index level or stock price. The relative indicator is an adjusted price level created in-house which reflects both the over/underperformance of the index or stock in relation to our multi-asset benchmark return (in our models, a dynamic weighting of MSCI All-World Index and government/corporate bonds) and the risk environment. Each indicator is in turn composed of a collection of trend-following technical indicators, including various popular simple or exponential moving average crosses, MACDs, Trender®, Williams %R, RSI, Stochastics, etc, in addition to several of our own proprietary indicators. The individual indicators are separated into two trading horizons and given weights based an optimization procedure using the R statistical software’s performance attribution package. Each individual indicator then contributes its weighted score, based on its current state, to either the Short-Term horizon indicator (STI) or the Medium-Term horizon indictor (MTI). The STI (whose score reacts to price data up to 3 weeks) is much more sensitive to price movements than the MTI (whose score reflects price data over the past 1 to 3  months).

Separately, a risk indicator is computed to evaluate the overall risk environment. The objective of the risk indicator is to determine if the markets are moving towards a period of “risk-on” (where investors are eager to buy risky assets) or a period of “risk-off” (when most investors are risk-averse). To do so, the indicator first studies the relationship between over twenty asset pairs. Examples of binomial asset pairs that are used include the S&P Consumer Discretionary Index vs the S&P Consumer Staples Index; Philadelphia Semiconductor Index vs the Dow Jones Telecom Index; the JP Morgan High-Yield Bond Index vs US Treasury Bond Prices or the Australian Dollar/Japanese Yen cross rate. When the first index in each pair outperforms the second index, based on our trend indicators, the Market Risk Indicator becomes more favorable to risk-taking. Inversely, when the second index begins to outperform, the risk indicator esteems markets are moving into a risk-averse phase. In addition to index pairs, the risk model evaluates various market indicators, including Advance-Decline lines, Volatility, New Highs/New Lows, and Volume. as well as credit spreads. The readings of each individual component are added to give an overall risk index level. Based on the total risk score and sub-total of a short-term version of the risk score (which gives an indication of the current evolution of the risk environment), our risk indicator reports five possible states:

<————————————————————————————————————————————–>

High Risk                                                                Balanced Risk                                                             Low Risk

———————————>                                         <——————————————–

High Risk, Improving Market                                             Low Risk, Deteriorating Market

An assessment of our risk indicator determines the percentage allocation of the fund. A low risk environment necessitates a significant allocation to equities, while a high risk environment requires cash to be raised. The cash allocation during periods of changing risk environment, or a high risk environment, is discretionary and is determined by the availability of indexes in up-trends. If the model finds very few indexes on a buy signal, for example, the cash allocation will necessarily be larger. In general, a significant deterioration in our composite risk indicator entails a greater consideration for capital preservation and an increase in cash holdings.  

Implementation of the Strategy

The first step is the selection of investable indexes or stock from our 250 index universe. Scanning the DGR Macro Trading Model output provides a short-list on candidates for investment. Outperforming indexes/ETFs, as well as improving indexes (moving from underperformance to outperformance), are candidates for investment. Given multiple signals of outperformance across several indexes or individual stocks, we attempt to respect the principles of diversification. For example, the strategy will  avoid holding simultaneous positions on a financial sector  index, an insurance index, a regional bank index et a REIT index, even if the model generates positive signals on all four, as these sectors are all particularly sensitive to interest rate changes. The second step is confirm the entry/exit point by running the selected indexes/ETFs in the WMA Trend Model to verify the absolute trend readings. A short-term composite reading and a medium-term composite reading is produced for each asset reflecting “Over-Weight” to “Under-Weight signals.  Full positions or half positions for a given asset may be taken based on confirmation/non-confirmation of the model’s signals over the two horizons. Stop-losses are not used as the model imposes a strict sell discipline. Indexes/ETF positions which lose their relative outperforming status or who flash sell signals on all horizons of the composite indicator are liquidated upon the market close. Conversely, profit-taking occurs when the relative outperformance of a position begins to weaken, even if the absolute trend remains up.