Implementing Point and Figure RS Signals

Transcription

1 Dorsey Wright Money Management 790 E. Colorado Blvd, Suite 808 Pasadena, CA John Lewis, CMT August, 2014 Implementing Point and Figure RS Signals Relative Strength, also known as Momentum, is one of the premier investment factors and has proven to be extremely robust over time. Using momentum as a selection criteria means you are looking for securities that have outperformed the broad market or the rest of the securities in the universe. Momentum looks for strength, and it is a trend following methodology. There are numerous studies showing momentum works well as a selection criteria within many different markets (U.S. equities, foreign equities, commodities, etc ) and across markets (asset allocation). As long as there is dispersion in the universe, momentum tends to work well over time. 1 The most common definition of momentum is using a time based window. For example, the trailing 12 month price return for each security in your universe would be calculated, and then the securities with the highest trailing returns would be selected. The momentum anomaly exists at the intermediate term time horizon so most successful strategies will use a time window of about 3 to 12 months. At intervals less than 3 months, momentum models are affected too much by short term market noise. Very long term intervals, like 3 to 5 years are affected more by mean reversion than momentum. There are several ways to calculate relative strength or momentum. One way we have found to be very successful over the years is using Point and Figure (PnF) analysis to determine momentum. We can classify stocks into baskets based on their intermediate and longer term momentum characteristics using PnF signals. We discussed this methodology in our June, 2014 whitepaper titled, Point and Figure Relative Strength Signals. In that study we demonstrated that holding a basket of stocks on PnF buy signals and in a column of X s (positive long and intermediate term relative strength) was superior to any other PnF configuration, the S&P 500, and an equal weighted return of all the securities in our universe (please see Appendix 1 for details). Point and Figure relative strength signals are a solid method for determining momentum rankings within an equity universe. One major difference between PnF relative strength and traditional momentum methods is the element of time. As mentioned above, momentum is an intermediate term factor. PnF, on the other hand, strips time out of the equation and looks only at the volatility of the securities. In our July, 2014 whitepaper titled, Point and Figure Relative Strength Box Sizes, we explored using different box sizes for PnF relative strength analysis. Using a small box size allows smaller movements to affect the PnF chart, while using larger box sizes means larger moves must occur before there is a change to the 1 The relative strength strategy is NOT a guarantee. There may be times where all investments and strategies are unfavorable and depreciate in value.

2 chart. Much like a time based momentum measure, there is a sweet spot in the box size that picks up the intermediate term trend while filtering out short term noise. Appendix 2 details the returns from using various box sizes for the same universe of securities. For a universe comprised solely of domestic equities, the best returns were generated using box sizes in the 6.5%-7.5% range. As the box size got smaller or larger, the returns decreased. The results from these previous studies used a universe of U.S. common stocks that were ranked in the top 1000 by market cap. Both of these studies provided good guidelines on how to use PnF as a tool to create momentum rankings, but they didn t shed any light on what happens within each basket. From a practical standpoint, it is difficult for most investors to purchase a basket of stocks that contains 250 or more securities. Does taking a small subset of the highest ranked securities perform well over time, or does the investor need to purchase everything in order for the strategy to work? To answer that question, we created numerous portfolios that purchased securities at random instead of purchasing the entire basket. The methodology to create the random portfolios is very straightforward. We used the same overall universe of securities from the previous studies and classified each security into one of four baskets (BX, BO, SX, SO) based on PnF relative strength. On the start date of the test, we drew 50 securities at random from the stocks with the best PnF relative strength ratings (BX). The next month, any of the 50 securities that remained on a BX configuration remained in the portfolio, and all of the other securities were sold. For each security that was sold we drew one security at random from the BX basket that wasn t already in the portfolio. The portfolio was always fully invested with 50 securities, and each month those 50 stocks were equal weighted. We created 100 different portfolios using this methodology so we wound up with 100 different equity curves. The securities were different from portfolio to portfolio, but each portfolio held only stocks on a PnF buy signal and in a column of X s. Before discussing the results of the simulations, it is valuable to look at the length of time securities remained eligible for the portfolio. If securities only remain in the BX basket for a couple of weeks or months, the turnover might be too high to implement the strategy. The table below gives a summary of all the PnF baskets over the course of the entire test. The performance listed in the table is relative to the S&P 500 Total Return Index so it didn t matter if the security was in the BX basket during a bull or bear market. Stocks in the BX basket outperformed the broad market by about 5% and they are held 280 calendar days on average. Obviously, there is tremendous variation in both of these numbers, but the averages indicate that using PnF signals to form momentum portfolios is very implementable. It is also interesting to note that as the momentum ranks get worse, so do the relative performance numbers. Signal Column Rel Perf Days Buy X 5.32% 280 Buy O 0.45% 164 Sell X 0.31% 132 Sell O -1.48% 175 Exhibit 1 contains the summary data for the 100 trials of randomly drawn portfolios. The green dot on the chart signifies the return of the S&P 500 Total Return Index. The small red bar is the average re-

3 Exhibit 1: Random Portfolio Returns The performance information presented is the back-tested performance of non-investable indexes. Investors cannot invest directly in an index. Indexes have no fees. Back-tested performance is hypothetical (it does not reflect trading in actual accounts) and is provided for informational purposes to illustrate the effects of this strategy during a specific period. Back-tested performance results have certain limitations. Such results do not represent the impact of material economic and market factors might have on an investment advisor s decision making process if the advisor were actually managing client money. Back-testing performance also differs from actual performance because it is achieved through retroactive application of a model investment methodology designed with the benefit of hindsight. Examples presented herein are for illustrative purposes only and do not represent past recommendations. Past performance is not indicative of future results. Potential for profits is accompanied by possibility of loss.

4 turn for the 100 portfolios for the given year, and the box denotes the upper and lower quartile bounds of the returns. The whiskers on the end of each box extend out to the minimum and maximum return for each year. The first thing to notice is that all 100 portfolios outperformed the broad market over the entire test period. The S&P 500 Total Return was up about 802% while the lowest returning trial was up 1,364%. Even the most unlucky portfolio outperformed the broad market. If you had an average performing portfolio you did significantly better than the broad market at about 2,324%. This clearly demonstrates the robustness of using PnF relative signals to form momentum portfolios. Even drawing securities at random from a basket of highly ranked momentum stocks delivers outperformance over long periods of time. One of the reasons we were able to generate outperformance on all 100 trial runs is the discipline of the implementation. Each month, securities that didn t qualify were sold and new securities that qualified were added to each portfolio. A momentum strategy is constantly pushing the portfolio toward strength. There is never a time when a weak security is held because we were waiting to get back to even or some other reason. There is a tremendous opportunity cost in not constantly cutting out weak securities. The process also never excluded any security because it was too volatile, had gone up too much, or some similar reason. Strong stocks were added to the portfolio in a systematic and disciplined fashion based solely on the momentum ranks. There were plenty of securities that declined significantly or blew up while we held them. Looking at the long term results of our random trials it should be clear that blow ups don t destroy the returns of a momentum process. The process is robust enough to take care of those situations as long as you remain disciplined. The real problems come when you lose focus and allow stocks with poor momentum characteristics to remain in the portfolio. Looking at the data in Exhibit 1, it is also important to keep in mind that while these strategies outperformed over the entire test period there were certainly periods where the momentum factor didn t perform as well as the broad market. These periods generally occur when there are major changes in leadership, or there is a lack of definable leadership. These periods are not uncommon, and any momentum based strategy will go through them. The tests we ran did nothing to attempt to combat this problem. We simply kept the system in place and waited for the new leadership to be rotated in to the portfolios. There may be ways to alleviate some of the performance lag experienced by trend following strategies whenever there are changes in leadership, but those regime switching models are very difficult to implement. While it is very uncomfortable for investors, we have found that patience and adhering to the strategy are the best ways to deal with periods of underperformance. One additional major point many investors fail to realize is the tremendous amount of difference in the returns of the 100 portfolios in the same year. In a year such as 1995, about 50% of the portfolios outperformed the broad market and 50% underperformed. All 100 trials used the exact same momentum process and universe of securities so it is hard to believe that is possible. There is an element of luck in the outcomes of the process. Is

5 one portfolio better than the other because it performed better for a year or two? Clearly not. The focus should be on the process because the outcomes are so difficult to control. There may be things you can do to narrow the range or improve the outcomes, but there will always some variation in the returns. It is important to keep this in mind when looking at any historical testing. If we ran just one test that happened to be the highest returning test of the 100, our strategy would look much better than it really is. That is why examining the overall process and the possible range of outcomes is so important. Small variations in historical performance are often the result of chance rather than one system being superior to another. The disciplined application of the method is what drives the returns over the long haul. Momentum is a very powerful investment factor. Point and Figure relative strength charts can be used to determine which securities have superior momentum. There will always be uncertainty in the outcomes of a momentum process in the short term, but a disciplined application of the strategy gives investors a high probability of outperformance over time. When using a PnF relative strength model, you always want to purchase securities with the best momentum characteristics and sell underperforming securities very quickly. It is very possible to do this when selecting a small subset of high momentum stocks from a much larger basket. Disclosures Nothing contained herein should be construed as an offer to sell or the solicitation of an offer to buy any se-curity. This report does not attempt to examine all the facts and circumstances which may be relevant to any company, industry or security mentioned herein. We are not soliciting any action based on this document. It is for the general information of clients of Dorsey, Wright & Associates, LLC ( Dorsey, Wright & Associates ). This document does not constitute a personal recommendation or take into account the particular investment objectives, financial situations, or needs of individual clients. Before acting on any analysis, advice or recommendation in this document, clients should consider whether the security or strategy in question is suitable for their particular circumstances and, if neces-sary, seek professional advice. Dorsey, Wright & Associates, its officers, directors, partners and/or other associated persons may own, hold options, rights or warrants to purchase some of the securities or assets mentioned in this report, or close equivalents. Even if Dorsey, Wright & Associates does not currently hold the asset, it may in the future. Dorsey, Wright & Associates may elect to buy or sell these assets or change its opinion without regard to this report, and without prior notice. The price and value of investments referred to in this document, if any, and the income from them may go down as well as up, and investors may realize losses on any investments. This report is based on public information that we consider reliable, but we do not represent that it is accurate or complete, and it should not be relied on as such. Opinions expressed herein are our opinions as of the date of this document. We do not intend to and will not endeavor to update the information discussed in this document. The projections or other information in the attached document regarding the likelihood of various investment outcomes are hypothetical in nature, do not reflect actual investment results and are not guarantees of future results. Results may vary with each use and over time. The outcomes in this document are based on historical market data and are presented to illustrate the relative strength strategy. These outcomes cannot be relied upon in the face of changing market conditions, and should be re-evaluated frequently. The performance information presented is the result of back-tested performance. Back-tested performance is hypothetical (it does not reflect trading in actual accounts) and is provided for informational purposes to illustrate the effects of relative strength during a specific period. The relative strength strategy is NOT a guarantee. There may be times where all investments and strategies are unfavorable and depreciate in value. Relative Strength is a measure of price momentum based on historical price activity. Relative Strength is not predictive and there is no assurance that forecasts based on relative strength can be relied upon. Back-tested performance results have certain limitations. Such results do not represent the impact of material economic and market factors might have on an investment advisor s decision making process if the advisor were actually managing client money. Back-testing performance also differs from actual performance because it is achieved through retroactive application of a model investment methodology designed with the benefit of hindsight. Dorsey, Wright & Associates believes the data used in the testing to be from credible, reliable sources, however; Dorsey, Wright & Associates makes no representation or warranties of any kind as to the accuracy of such data. All available data representing the full platform of investment options is used for testing purposes.

6 Appendix 1: PnF Box Size Returns Date BX BO SX SO SPXTR Univ EQ Wt 12/29/ /31/ % -9.76% % % -3.10% % 12/31/ % 27.72% 44.97% 43.41% 30.47% 46.73% 12/31/ % 11.23% 17.23% 23.87% 7.62% 15.44% 12/30/ % 18.46% 7.99% 14.86% 10.59% 14.62% 12/30/ % -3.57% -2.61% -2.67% 0.85% -3.68% 12/29/ % 32.04% 20.95% 29.73% 37.58% 32.47% 12/31/ % 17.12% 10.18% 16.02% 22.96% 18.65% 12/31/ % 26.64% 18.34% 21.62% 33.36% 26.92% 12/31/ % 12.09% 6.61% 4.31% 28.58% 10.75% 12/31/ % 20.99% 5.57% 5.55% 21.04% 17.23% 12/29/ % -3.47% -9.84% % -9.10% -4.90% 12/31/ % 7.21% -3.74% -1.07% % -2.59% 12/31/ % % % % % % 12/31/ % 42.38% 56.32% 69.01% 28.68% 41.06% 12/31/ % 17.67% 20.81% 10.87% 10.88% 19.14% 12/30/ % 6.55% 5.13% 5.48% 4.91% 10.21% 12/29/ % 15.20% 8.77% 16.17% 15.79% 16.29% 12/31/ % 3.15% 10.13% -1.87% 5.49% 8.34% 12/31/ % % % % % % 12/31/ % 38.20% 61.16% 78.71% 26.46% 44.08% 12/31/ % 28.19% 19.28% 25.84% 15.06% 24.95% 12/30/ % -0.69% -7.04% -7.47% 2.11% -0.27% 12/31/ % 21.01% 12.31% 16.16% 16.00% 16.69% 12/31/ % 35.23% 40.88% 33.03% 32.39% 36.31% Cum % % % % % % Ann 14.45% 10.81% 7.94% 8.38% 9.44% 11.39% St dev 20.49% 19.27% 24.33% 28.25% 18.65% 20.39% Sharpe (Rf=0) All Performance numbers are based on the back-tested performance of non-investable indexes. Investors cannot invest directly in an index. Indexes have no fees. Please see the final page for important disclosures regarding back-tested performance. Examples presented herein are for illustrative purposes only and do not represent past or present recommendations. Past performance not indicative of future results. Potential for profits accompanied by possibility of loss.

7 Appendix 2: PnF Box Size Returns Point And Figure Box Size Date 1.50% 2.50% 3.50% 4.50% 5.50% 6.50% 7.50% 8.50% 9.50% 10.50% 12/29/89 12/31/90-9.1% -3.1% -2.3% -2.2% -0.1% -0.7% -2.4% -3.9% -4.5% -4.7% 12/31/ % 47.3% 50.9% 52.7% 53.3% 55.5% 55.8% 55.4% 54.9% 44.5% 12/31/ % 12.8% 13.2% 14.6% 14.3% 13.3% 12.1% 11.8% 10.5% 8.2% 12/30/ % 11.7% 12.7% 13.3% 13.9% 14.6% 14.5% 13.6% 13.9% 10.1% 12/30/94-6.6% -5.8% -6.7% -5.4% -5.0% -4.6% -5.0% -5.6% -5.3% -6.3% 12/29/ % 30.9% 33.9% 34.8% 36.1% 37.6% 37.3% 37.7% 37.5% 38.1% 12/31/ % 14.3% 17.9% 21.2% 21.3% 24.6% 24.1% 23.4% 23.5% 23.8% 12/31/ % 23.9% 29.2% 32.3% 32.5% 35.7% 34.2% 34.1% 33.2% 36.8% 12/31/98 5.7% 7.5% 14.7% 16.9% 20.4% 20.6% 20.1% 19.3% 20.3% 23.9% 12/31/ % 33.6% 38.6% 36.6% 42.0% 45.9% 49.4% 46.9% 45.9% 36.8% 12/29/ % -4.3% 2.8% 6.9% 3.2% 2.0% 1.7% 1.3% 3.0% 0.1% 12/31/ % -16.0% -14.7% -14.0% -13.2% -12.4% -12.6% -11.8% -13.1% -12.7% 12/31/ % -14.6% -14.0% -14.6% -14.4% -13.6% -12.7% -13.6% -13.1% -13.5% 12/31/ % 32.9% 31.5% 33.0% 34.7% 33.6% 33.6% 33.9% 33.6% 31.5% 12/31/ % 18.3% 18.7% 19.2% 19.9% 21.2% 21.1% 21.6% 22.3% 19.1% 12/30/ % 11.7% 12.6% 13.0% 13.7% 12.5% 13.3% 13.4% 13.3% 15.4% 12/29/ % 15.8% 15.7% 16.1% 15.7% 17.1% 17.4% 16.9% 16.3% 15.1% 12/31/ % 19.3% 18.2% 17.8% 16.0% 16.7% 16.8% 16.1% 16.5% 16.6% 12/31/ % -38.3% -34.9% -33.2% -33.3% -33.8% -33.4% -34.2% -35.3% -37.7% 12/31/ % 31.9% 29.2% 29.6% 28.9% 29.2% 28.0% 27.0% 26.8% 24.9% 12/31/ % 24.0% 23.3% 24.2% 24.3% 23.7% 24.7% 24.3% 24.8% 22.3% 12/30/11-0.2% 2.4% 2.6% 2.4% 3.2% 2.4% 2.5% 2.8% 3.0% 3.2% 12/31/ % 12.8% 13.9% 13.0% 13.5% 15.1% 15.1% 15.5% 15.2% 16.3% 12/31/ % 37.4% 38.2% 38.7% 38.2% 36.4% 36.9% 35.2% 35.0% 34.2% Cum 542.3% % % % % % % % % % Annual 8.1% 10.9% 12.6% 13.5% 14.0% 14.5% 14.4% 14.0% 13.8% 12.5% St Dev 20.1% 19.6% 19.6% 19.6% 19.9% 20.5% 20.7% 20.6% 20.6% 19.7% All Performance numbers are based on the back-tested performance of non-investable indexes. Investors cannot invest directly in an index. Indexes have no fees. Please see the final page for important disclosures regarding back-tested performance. Examples presented herein are for illustrative purposes only and do not represent past or present recommendations. Past performance not indicative of future results. Potential for profits accompanied by possibility of loss.