Situation Index Handbook 2014

The Situation Index Handbook
Contents Page
1) 2) 3) 4) 5) 6) 7) 8)
Introduction to Market Timing in South Africa The Market Situation Index Equity Curves Non-linear application of the MSI protecting against rare events Taking the MSI international the China Situation Index The land of the rising sun the Japanese Situation Index The Australian Situation Index The Korea Situation Index Combined Asian SI equity curve the benefits of diversification
2 8 17 24 27 31 37 41
Overview I have developed an equity market timing tool, the Market Situation Index, which provides a quantitative framework for improving the top-down or macro component of investment processes. I believe it will have particular value for balanced mandate asset allocation, management of net equity exposure in long/short equity hedge funds and as a profit source in its own right (beta timing as alpha) in macro hedge fund mandates. The MSI is an impartial score of the prevailing macro-economic environment made up of eight factors derived from live price data. The factors are a mix of fundamental, sentiment, technical & valuation-based time series, all with a common-sense causal link to equity market returns. The key question that the MSI score answers is What has been the historic equity performance and risk associated with the prevailing macro-economic environment. The MSI has been back tested over some 12 years of weekly data. The majority of local institutions focus their investment staff and resources on stock selection within the handful of liquid locally listed counters. However South Africa seems well suited to equity market timing processes, with macro-economic factors being very diagnostic of equity returns, and cheap and easy execution of trades in deep equity index derivative markets. For absolute return mandates the MSI may be best applied using an options-based strategy, which restricts the potential losses from rare events at all times. While the MSI is a product of 20+ years of experience in South African equity markets, I have used the technology under-pinning it to successfully develop similar models for Japan, China, Australia and South Korea.
1
Page
If you are interested in receiving more research in the future, please drop me an email at [email protected]
This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved

1) Introduction to Market Timing in South Africa
The dramatic difference between market returns & fund returns on the one hand and investor returns on the other can primarily be attributed to poor market timing, both by professional investment managers or advisors and by clients themselves. While buying and holding a balanced or multi-asset portfolio appears to theoretically satisfy investor needs with an actuarial elegance, the grim reality for the majority of investors is that of adverse equity entry & exit points and mediocre returns.
Conventional wisdom says market timing is a negative sum game The potential to add value from market timing in asset classes with a volatile return profile is obvious. While the FT/JSE Top40 index has delivered excellent total returns of over 17% pa for the last 12+ years, it suffered two massive drawdowns (-46% and -36%) within this period. When we view the chart on a log scale the extent of the 2002/3 downturn is put into perspective.
Page
However, conventional wisdom, backed by studies showing poor market timing decisions by investors in aggregate, is that timing should be avoided and investors should rely on passive diversification between asset classes to enhance risk-adjusted returns. Specific criticisms are that given the costs of trades, market timing is a negative sum game and in addition it engenders a speculative mind set inappropriate for most investment mandates. Note that similar studies on stock selection also show that in aggregate it is also value destructive, yet that does not stop pretty much every market participant from trying to generate alpha through picking stocks. This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved

Stock selection a dubious source of future alpha in SA The vast majority of domestic equity mandates are focused on adding value through stock selection. Given the short list of liquid stocks (maybe some 70 names), this is a formidable challenge for a house with a substantial asset base. At the extreme, the biggest institutions have hundreds of billions of domestic assets under management. To effectively implement stock selection decisions across their portfolios can take weeks and carry a huge active cost of trading. Despite this challenge, investors continue to back these mega portfolios in their attempts to deliver alpha by stock picking in the years to come. Passive diversification has issues too! A great many balanced or multi-asset mandates have a static (or in practice close to static) asset allocation. Diversification is generally promoted based on historic return data, which are used to derive expected returns for each asset class, and hence the appropriate asset allocation. However, with volatile asset classes such as equities, little regard is shown as to how achieving these historic returns has been entry point dependent, and what the probable return outcome would be from current or indeed even more adverse entry points. In addition, diversification models shun cash holdings as the worst returning asset class, but ignore the free optionality embodied in cash the ability to purchase other risky assets on a market set back (call it the dry powder value of cash). The market is full of closet market timers Unfortunately it gets much worse for investors, as many of those who claim market timing detracts from value and hence should be avoided are actually closet market timers: A huge number of investors claim to be buy and holders, but in fact they use a form of market timing, the ICSIA system (I Cant Stand It Anymore). Though this is probably the most widely used market timing system in the world, we dont recommend it. It relies on emotional reactions to market fluctuations. After a long period of market gains, this system induces many of its followers to finally jump into the market, usually against their better judgment, when they can no longer stand to sit on the sidelines watching other people making what looks like "easy money." When the market's in decline, this irrational system prompts its followers to remain invested, even as they continue to lose money, until they cannot stand the losses any more - and then to bail out when prices are very depressed (Ref.1) I believe the above quote applies to professional money managers & financial advisors as much as to retail investors. By not overtly addressing market timing they end up taking significant and unplanned timing risk after reaching a pain threshold and capitulating. The level of investment expertise and experience appears to offer little protection against this value-destructive behavior (by way of example I recall a due diligence I was involved in of a listed institution, where despite the CIOs passionate belief that their top-down process was key to their success, excellent stock selections in just two years explained some 150% of their benchmark out-performance over the previous decade. In fact their asset allocation appeared reactive to recent asset class returns and hugely detracted from value).
Page

Closet market timing and its erosion of returns is endemic throughout the investment world. For instance, the Bogle Research Institutes 20 year study of US mutual funds ending 2002 (graphic below) showed that even during the greatest bull market in recent history investor returns from equity funds were far below that available risk free from cash deposits or money market funds. Their study highlighted what they called the character penalty, caused by poorly timed subscriptions & redemptions and poor choice of sector or theme fund away from the index. Astoundingly, this character penalty was more than 2.5 times bigger than the impact of both below index performance & costs on fund returns in the study (8% pa v. 3.1% pa). John Bogle is the founder of index tracker giant Vanguard, so he saw this research as being a huge selling point for their products. Through buying and holding of index funds an investor could avoid the character penalty and hugely reduce the cost penalty. While the latter is certainly true, unfortunately it appears that it is not so easy for us to step away from behaviorally induced poor market timing. The Bogle Research Institute 20 year Mutual Fund Study
Just in case you believe things have improved recently, the Morningstar Dodge & Cox global equity fund study (August 2012), showed a 10 year global equity fund return of 9.6% per annum, but a 10 year investor return of -1.2% pa, implying in this case a timing penalty of 10.8% pa. Investors in the fund timed their subscriptions and redemptions so poorly that in aggregate they lost money over a 10 year period when the nominal price of the fund rose 128%! While we can argue the predictability of the market itself, these behavioral failings of market participants appear to repeat themselves like clockwork with every major price cycle. Ironically, it is perhaps the reliable occurrence of this very behavioral pattern that provides opportunities for market timing as prices and the underlying macro-economic factors diverge. Despite the huge potential for profits, market timing is largely ignored by most market participants, and in fact
Page

actively derided by many with actuarial backgrounds. This dearth of intellectual competition is in sharp contrast to stock selection (lots of very smart and well resourced people trying that). South Africa appears well suited for a successful market timing process with several key macro economic variables having a significant impact on risk asset returns. In addition SA has deep and liquid index derivative markets, enabling cheap execution via index futures and options with minimal market impact even for sizable trades. I strongly believe that market timing of South African equities has the potential to both enhance returns and avoid substantial risk for an investor. Of course, to successfully overcome the formidable behavioral hurdles of big, binary investment decisions, market timing exposures must be carefully planned, backed by effective tools and preferably applied through an objective process. The Market Situation Index (MSI) An objective score for market screening With an equity-centric background, when I was given my first balanced portfolios to manage (in 1996), I had little idea what to do about asset allocation. What proportion of my clients portfolios should be in equities, bonds and cash? After consulted with my vastly more experienced colleagues, I developed a weekly screening routine based on reviewing an extensive number of charted factors, primarily macro-economic variables, which provide some insight into current market conditions. As the years went by I developed some proprietary fundamental models that also became important factors. Greater experience allowed me to narrow down the factors I reviewed. Every factor was chosen for its common sense causal link with South African equity returns. Despite these efforts to create a structured process, I alarmingly found, however, that my own prior view tended to influence which factors I found noteworthy (if I was bearish, for instance, I would subconsciously focus on bearish charts that supported my view and downplay bullish charts). To compensate for this behavioral problem, I started calculating a screening score for the factors reviewed for each of several specific markets and sectors. Each factor would be calibrated around an inflection point to score either 1 (bullish) or 0 (bearish). This provided an objective scale, e.g. for the JSE Top40, the 8 key factors reviewed could score between 0 and 8. This was then totaled to score what I called the Market Score. Should I be feeling bearish, but the score is 7/8, I would need to review my thesis given the dissonance between my view (bias) and the score (impartial). The Factors used for the Market Score The Score was based on 8 factors, each selected due to their common sense impact on the South African market determined over many years of managing money. The factors include: Fundamental factors Technical factors. Sentiment factors Valuation factors
Page

The factors utilize both local and global macro data, and are all based off live market prices available at any time as opposed to delayed economic data releases. The exact nature, composition and calibration of the factors is of course proprietary. As an aside I also calculate situation indices developed in a similar fashion for several specific sectors and other markets. It must be said that I made my living picking stocks, the market score process was merely something developed to assist in what I viewed as an undesirable but necessary dilution of focus. Given this and the general industry skepticism towards market timing, it is perhaps not surprising that it took many years before I investigated the score as a potential source of profits. Back-testing the MSI In 2011 I came across a technical trading tutorial in which the practitioners used a Situation Index (calibrated by various technical studies), to determine whether to focus on bullish or bearish technical patterns. The similarity to my screening score resonated and later that year I began to think about the potential for the score as a trading tool. To that end I recreated a weekly historic database of the 8 factors that make up the market score, and christened it the Market Situation Index (MSI). The next step was to determine whether the MSI had any potential value as a market timing tool. Most market timing tools are created by statistical parameterization (i.e. testing a great number of possible factors and including those that statistically enhance the tools back-testing results irrespective of their logical or casual relevance). That every factor included in the MSI was there due to a clear casual relationship determined over many years of managing South African equity portfolios, must surely improve the odds of developing a robust predictive tool?
Market Situation Index & FT/JSE Top40 Index
800 8
700
600
500
400
300
200
100
Mkt Sitn Index RHS /8

Data ex Bloomberg
Top40 Index ATR 17.2%

0
2003/08 2004/08 2005/08 2006/08 2007/08 2008/08 2009/08 2010/08 2011/08 2012/08 2013/08
Page
2001/08
2002/08

The MSI (the blue area in the chart above) can be seen on an eye ball basis to be relatively high in periods of market strength and relatively low when prices are declining. But was there any true predictive ability? I have done an extensive weekly back-test over some 12.5 years (646 weeks to the end of 2013). The current weeks MSI score was utilized to categorize the following weeks market returns. I divided the MSI score into 3 zones (out of the possible range of 0 8): Sell Neutral Buy Score 0, 1, 2 Score 3, 4, 5 Score 6, 7, 8
It should be noted that if a factor is near a calibration inflection point, insignificant changes in a factor value can change the scoring. Hence having broad zones for each category is appropriate, as I want to avoid an illusion of precision in what is a probabilistic exercise. My review effectively chains together all the weeks over the back-test period that have buy ratings (i.e. the previous weeks MSI score is 6, 7 or 8), and evaluates the characteristics of that series of returns. The exercise is then repeated for all neutral and sell weeks.
Market Situation Index

Position based on Calibrated Score Number of Observations Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts SELL 89 14% -8.2 -25.5 NEU 213 33% 12.6 -4.6
Buy & Hold BUY Top40 344 646 53% 28.1 17.2 10.8
Backtest of 12.5 years of weekly data 17 August 2001 to 3 January 2014
Results of the MSI Back-test As shown in the table the index returns in weeks following MSI readings of Buy, Neutral and Sell were dramatically different. Buy weeks on average delivered returns of 28% pa, 10.8 percentage points pa above the average of the Top40 for the whole period. In contrast Sell weeks delivered average returns of -8.2% pa, 25.5 percentage points pa below the average of the Top40 index total return. While I believe statistics have no place in factor selection and calibration, they are entirely appropriate when evaluating back-test results. A quantitative evaluation of the MSI back-test (done by my appropriately qualified colleague Reza Khan) has confirmed its statistical significance and in addition determined that no one factor can explain the differences, indicating that the combination of factors is the key, to my mind implying a robust model. Of course the MSI does not predict market tops and bottoms. It is important to have a clear understanding of what question the MSI readings answer What has been the historic equity performance and risk associated with the prevailing macro-economic environment as represented by the MSI score?
Page

2) The Market Situation Index Equity Curves In the previous section Background to Market Timing in South Africa I described how I developed an equity market timing tool, the Market Situation Index (MSI). I suggest that reading that section first provides important background. This section examines the MSI in more detail over the 12.5 year back test period, looking at the returns and importantly volatility of returns. I then derive two equity curves using the back test data. The shape of the equity curves is very encouraging, providing strong evidence that the MSI tool can be an effective and robust component of a top-down macro process to determine asset allocation (balanced mandates), net equity exposure (long/short equity funds), and to generate timing profits (beta timing as alpha) in global macro mandates. The MSI back test The MSI was developed after many years of managing equity portfolios in South Africa (this is my 21st year in markets and 19th year of managing portfolios professionally). It consists of 8 factors across 4 categories - fundamental, technical, sentiment and valuation. Each factor is calibrated to score 0 or 1 around an inflection point, giving a total score range between 0 and 8. All the factors included in the tool have a common sense causal link with equity market returns - no statistical parameterization was done (i.e. testing a great number of possible factors and including those that statistically enhance the tools back-testing results irrespective of their logical or casual relevance). The weekly MSI scores were used to determine whether the model was showing a Buy, Neutral or Sell, and the following weeks market returns then categorized accordingly. All Buy weeks were then chained together, similarly for all Neutral and Sell weeks. The back test was for 646 weeks ending 3/1/2014 (some 12.5 years). The risk and return characteristics of these chained weeks are shown below.

Position based on Calibrated Score Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts SELL 14% -8.2 -25.5 36.9 15.2 NEU 33% 12.6 -4.6 21.7 0.1
Buy & Hold BUY Top40 53% 28.1 17.2 10.8 15.4 21.7 -6.3
Backtest of 12 years of weekly data
The MSI Buy periods have both significantly better returns and much lower volatility of return than the Top40 index. Encouragingly, the Sell periods had a negative return and a very high volatility of return (2.4x higher than the MSI Buy periods). Given the overall bull market over the back test period Sell periods only accounted for 14% of weeks (89/646) compared to Buys at 53% (344/646). Neutral weeks (some 33% of observations) had below index returns but comparable volatility. Overall this is a fantastic result for the MSI tool back test. However, two questions remain in This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved
Page

evaluating the MSI as a market timing tool. Is the performance of the tools chained weeks back ended or clustered, or is it reasonably consistent over the back test period? Secondly how many signals are there and hence what is the likely impact of the costs of implementing market timing decisions objectively? In deriving the MSI equity curves I hope to answer both of these questions. Given we have what appears to be an effective tool, the output or fruits of the tool can be used in two main ways: Firstly to identify periods where the prevailing macro-economic environment has historically been indifferent or negative for equity returns, and avoid market exposure during those periods (a risk lowering process). Here the goal is to achieve market like returns but at a much lower risk, and this is represented in the first equity curve. Secondly, we can use the MSI to identify periods where the prevailing macro-economic environment has historically been most conducive to equity returns (MSI = Buy), and take geared long exposure during those periods. In addition, we can identify periods where the prevailing macro-economic environment has historically been negative for equity returns (MSI = Sell), and maintain short market exposure during those periods. This is a return maximizing process. Obviously by calibrating what exposure taken for each of the MSI Buy, Neutral and Sell signals we can adapt the market timing process to suit a particular mandate long only, absolute return or hedge. Deriving equity curves using the MSI I designed and evaluated the equity curves for 2 potential market timing products: MACROtimer 100 nil a regulated long-only asset allocation fund designed to shift equity exposure between 100% and nil depending on the previous period MSI score. MACROtimer 150 nil (50) a hedge fund (designed to fall into the regulated hedge fund category in the future) with the goal of generating profits in both up and down markets. Equity curves for each product were generated utilizing both the Alsi future and the Dtop future. There were no significant differences between the two, so my evaluation has focused on products built using the more liquid Alsi future which also has a longer history.
Page
9

MACROtimer 100 nil
Potential Structure: Sector: Benchmark: Investment Objective: Collective Investment Scheme (Unit Trust or ETF) Multi-Asset Flexible Fund FT/JSE Top40 Index To deliver market returns at significantly below market volatility, measured over a full market cycle.
Assumptions in back-test Trading Commissions Trading market impact Total Trading Costs 2 index points (0.5 bps) 30 index points (7.5bps) 8bps on actual volume of trades done
Cash holdings earn the Safex Interest on Initial Margin over night interest rate. Exposures adjusted weekly (including the re-investment of interest on collateral) at the Friday closing price based on the MSI score as follows: MSI Score 6, 7, 8 MSI Score 3, 4, 5 MSI Score 0, 1, 2 BUY NEUTRAL SELL 100% Alsi exposure 0% Alsi Exposure 0% Alsi Exposure
Alsi future buy and roll assumes interest on margin account re-invested weekly to maintain 100% exposure. While this is similar to the Top40 total return indices ex Bloomberg (used in the MSI back test) it achieves a slightly higher return, perhaps due to futures trading at persistent discounts to fair value over much of the back test period. Results of back-test 646 weeks ending 03/01/2014 MACROtimer 100 nil Alsi future (buy & roll) Annualised Total Return (net of costs) 19.1% 18.1% Annualised Volatility of Return 11.6% 21.9% Costs (assumptions above) Data net of costs Assumed NIL Average Portfolio Churn 5.2x per annum N/A Largest Draw Down - Size -12.0% -46.1% Largest Draw Down Time to new high 32 weeks 137 weeks 2nd Largest Draw Down - Size -9.8% -36.5% 2nd Largest Draw Down Time to new high 81 weeks 119 weeks Proportion of negative weeks 20% 43% Two charts of the equity curve and benchmark are shown the second chart with a log scale to better illustrate the early volatility of the time series given its extended duration.
Page
10

9
Use Market Timing to lower risk

8

Buy 100% Neu 0% Sell 0%
7
6
5
5
4 4
3
2
2
11
Mkt Situ RHS ALSI ATR 18.2% AVol 21.9% MSI_Equity_Curve 100 nil ATR 19.1% AVol 11.6%
2001/08 2002/04 2002/12 2003/08 2004/04 2005/12 2006/08 2007/04 2007/12 2008/08 2009/04 2010/12 2011/08 2012/04 2012/12 2013/08
2001/12 2002/08 2003/04 2003/12 2004/08 2004/12 2005/04 2005/08 2006/04 2006/12 2007/08 2008/04 2008/12 2009/08 2009/12 2010/04 2010/08 2011/04 2011/12 2012/08 2013/04 2013/12
Page

8
Use Market Timing to lower risk Log Scale
Buy 100% Neu 0% Sell 0%

7
12
Mkt Situ RHS ALSI ATR 18.2% AVol 21.9% MSI_Equity_Curve 100 nil ATR 19.1% AVol 11.6%
0.8
2001/08
2001/12
2003/04
2004/12
2005/04
2006/08
2006/12
2008/04
2008/08
2009/12
2010/04
2011/08
2011/12
2013/04
2013/08
2002/04
2002/08
2002/12
2003/08
2003/12
2004/04
2004/08
2005/08
2005/12
2006/04
2007/04
2007/08
2007/12
2008/12
2009/04
2009/08
2010/08
2010/12
2011/04
2012/04
2012/08
2012/12
Page
2013/12

MACROtimer 150 nil (50)
Potential Structure: Hedge Fund (currently unregulated, but designed for the Retail Hedge Fund category per the National Treasury white paper) Multi-asset Flexible Fund Total return on the FT/JSE Top40 Index To generate a return higher the benchmark at a comparable volatility, measured over a full market cycle. To make profits in both up and down markets.
Sector: Benchmark: Investment Objective:
Assumptions in back-test Trading Commissions Trading market impact Total Trading Costs 2 index points (0.5 bps) 30 index points (7.5 bps) 8bps on actual trades done (generous assumption)
Cash holdings earn the Safex Interest on Initial Margin overnight interest rate. Exposures adjusted weekly (including the re-investment of interest on collateral) at the Friday closing price based on the MSI score as follows: MSI Score 6, 7, 8 MSI Score 3, 4, 5 MSI Score 0, 1, 2 BUY NEUTRAL SELL 150% Alsi exposure 0% Alsi Exposure -50% Alsi Exposure
Alsi future buy and roll (the benchmark) assumes interest on margin account re-invested weekly and futures held adjusted to maintain 100% exposure. Results of back-test 646 weeks ending 03/01/2014 MACROtimer Alsi future 150 nil (50) (buy & roll) Annualised Total Return 24.7% 18.1% Annualised Volatility of Return 18.7% 21.9% Costs (assumptions above) Data net of costs Assumed NIL Average Portfolio Churn 9.2x per annum N/A Largest Draw Down - Size -19.0% -46.1% Largest Draw Down Time to new high 84 weeks 137 weeks 2nd Largest Draw Down - Size -16.4% -36.5% 2nd Largest D.Down Time to new high 38 weeks 118 weeks Proportion of negative weeks 29% 43% Two charts of the equity curve and benchmark are shown the second chart with a log scale to better illustrate the early volatility of the time series given its extended duration. This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved
Page
13

16
15 14
Use Market Timing to boost returns

Buy 150% Neu 0% Sell (50%)
13 12
11 10 5
9 8 7
6 3
5 4 3
2 1
14
Mkt Situ RHS MSI_Equity_Curve 150 nil (50) ATR 24.7% AVol 18.7% ALSI ATR 18.2% AVol 21.9%
2002/04 2002/08 2004/04 2004/08 2004/12 2006/04 2006/08 2006/12 2008/12 2009/04 2010/12 2011/04 2013/04 2013/08 2013/12
2001/08 2001/12 2002/12 2003/04 2003/08 2003/12 2005/04 2005/08 2005/12 2007/04 2007/08 2007/12 2008/04 2008/08 2009/08 2009/12 2010/04 2010/08 2011/08 2011/12 2012/04 2012/08 2012/12
Page
Use Market Timing to boost returns Log Scale

Sell 0,1,2 Cash 3,4,5 Buy 6,7,8
15
0.8
Mkt Situ RHS MSI_Equity_Curve 150 nil (50) ATR 24.7% AVol 18.7% ALSI ATR 18.2% AVol 21.9%
2002/08 2002/12 2004/12 2005/04 2006/12 2007/04 2007/08 2009/04 2009/08 2009/12 2011/04 2011/08 2011/12 2013/08 2013/12
2001/08 2001/12 2002/04 2003/04 2003/08 2003/12 2004/04 2004/08 2005/08 2005/12 2006/04 2006/08 2007/12 2008/04 2008/08 2008/12 2010/04 2010/08 2010/12 2012/04 2012/08 2012/12 2013/04
Page

Commentary on the Equity Curves The shape of the equity curves (best represented by the log scaled charts) indicates the MSI back test is not the result of clustered or back-ended performance. The conservative cost assumptions applied to the actual trades over the period have also not diminished the returns below attractive levels. It is clear the MSI is a viable and objective tool to greatly improve market timing. In addition, it is important to note that the periods when the MSI battles to be effective (represented by draw downs in the equity curve), do not correlate with declines in the overall market, on the contrary the MSI did well during periods of market decline during the back test: 2008 Top40 draw down -46%, MACROtimer 100 nil +6%, MACROtimer 150 nil (50) +34% 2002 Top40 draw down -36%, MACROtimer 100 nil +12%, MACROtimer 150 nil (50) +28% When evaluating weekly returns the correlation with the Top40 index is 0.526 for the 100 nil strategy and 0.264 for the 150 nil (50). The periods of non-correlation have important implications for the practical application of the tool MSI-driven timing products could combine exceptionally well with long-only stock picking portfolios. That is a quantitative benefit, however I believe there could be a much larger qualitative benefit by taking a positive step to address the impact of volatile equity returns (by allocating a slice of assets to a market timing strategy), an investor should then have the behavioral resilience to avoid poorly timed subscriptions and redemptions to their stock picking strategies, which can truly become buy and hold resulting in huge improvements in aggregate investor returns (as opposed to fund returns). Even though the back test period encompasses a great variety of market environments, of course the future may look nothing like it. However, given the common sense basis of factor selection, I believe the MSI is likely to be a robust tool going forward.
Page
16

3) Non-linear application of the MSI protecting against rare events The sound performance of the MSI as a timing tool to either reduce risk and/or enhance returns over the back test period has been clearly shown by the equity curves derived in the previous section. There is one problem however with absolute return mandates (such as most hedge funds) the linear expression of the MSI using index futures does leave exposure to dramatic market moves. If the MSI is positioned long and there is an overnight geo-political event resulting in a huge drop in global risk assets, the back test performance will count for nothing. Of course with mandates that have an equity index benchmark, exposure to rare events is not really an issue (or rather it is a mandate issue rather than a performance issue), but with an absolute return or cash plus benchmark it is a substantial risk to take. However using the MSI in risk-limiting mode, such as in the 100 nil equity curve shown in the previous section, reduces a mandates exposure to rare events by about half despite achieving market-like returns (simply by being in cash about half the time). While this may work for some mandates, a more likely solution is expressing the MSI using options for absolute return mandates, which enables the strategy to maintain a finite portion of capital at risk at all times. The friction costs of options (high cost of options, huge variability of option prices, wide volatility doubles when trading etc.) are substantial, but given the MSI tools insights on volatility of return, I was able to create a process that effectively mitigated this. A lower risk calibration of this process (losses capped at <10% at all times) achieved returns ahead of the Top40 index over the back test period with just 55% of the index volatility. MSI a better predictor of volatility of return than returns themselves?

Position based on Calibrated Score Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts SELL 14% -8.2 -25.5 36.9 15.2 NEU 33% 12.6 -4.6 21.7 0.1
Buy & Hold BUY Top40 53% 28.1 17.2 10.8 15.4 21.7 -6.3
Backtest of 12 years of weekly data
While the back test data above shows how different the volatility of return has been between MSI Sell and Buy periods (buy periods have less than half the volatility), this represents realized or historic volatility. Options are priced off implied volatility, which can be a very different time series. The chart overleaf shows the daily mid implied volatility price (50% delta, near contract, red line inverted scale) together with the MSI. On an eyeball basis the MSI appears to do a good job of predicting the volatility environment.
Page
17

8 10
MSI and Implied Volatility

7 15
20
25
30
35
40
MSI weekly Implied Vol Near 0.5 delta Inverted Scale

1-Feb-03 1-Feb-04 1-Feb-05 1-Feb-06 1-Feb-07 1-Feb-08
45 Implied Vol peaks at 62.5% in 2008 (off chart) 50

1-Feb-09 1-Feb-10 1-Feb-11 1-Feb-12 1-Feb-13
0
1-Feb-02
The Option Back Test Methodology To establish whether an MSI-driven process could be effectively implemented using options I did a second back test of daily data. The MSI signals were still updated weekly, however daily price data and index volatility data was used to price the option positions. The volatility data consisted of the mid implied volatility of Safex listed options on the Top40 index future, 40%, 50% and 60% delta (to enable calculation of the volatility skew) for the next two Safex expiries. The data was daily from Feb 2002 until Dec 2013. It was supplied by Prescient Securities, a leading South African derivative broker. Conservative assumptions were made for volatility doubles (i.e. the costs of trading). All positions were Safex-listed vanilla call or put options or combinations thereof, so no exotic options were used (the aftermarket for more exotic options can be thin or non-existent during periods of market stress). Option values were calculated using Prescients GEARS excel add in, Bloomberg spot price data and the volatility data described above. Reducing Friction Costs associated with Options The back test enabled me to calculate the performance of passive option strategies which allowed me to estimate the friction costs associated with options. This then formed a baseline against which I could evaluate the MSI-driven process. The goal was to both deliver downside protection with finite risk of loss at all times (as can easily be done by consistently applying a variety of option This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved
Page
18

positions) but also to generate sound index-comparable returns with constrained volatility of return (a much tougher ask). Position Structuring in different market and volatility environments The option-driven process has two inputs, the MSI signal (which determines whether to go long, short or sit in cash) and the prevailing level of implied volatility (which determines how the positioning, if any, is structured). The structures used were vanilla call or put options, or simple combinations thereof. All enabled the maximum loss to be determined up front (no open wings) and the size of that potential loss was kept to below 10% at all times, and in fact it averaged 6.2% over the back test period. This does not mean that the MSI option process equity curve cannot suffer a drawdown greater than 10% (if several unprofitable positions are taken sequentially over time it can) rather that the losses resulting from a sudden adverse market move are limited to single digits at all times. The position sizing was also calibrated such that delta-measured exposure would at all times comply with the National Treasurys proposed Retail Hedge Fund limits set out in their white paper. Positioning was designed to benefit from the likely volatility environment as well as capture the potential returns. A matrix of possible positioning is summarized below. Matrix of Non-linear positioning
The equity curve showing the results of applying the MSI objectively in a non-linear fashion utilising Safex-listed index options is illustrated below. Both the daily mark to market curve (in red) and the realized NAV curve (on option sale or expiry, in black) are shown.
Page
19
Page
20

7,000,000
MSI Non-linear (10,6,4)

Log scale
Daily Back Test
80
R6.5m R6.0m
60
40
20
R1m
Page
21
MSI Options ATR Rolling 1 year return % Near Alsi Rolling 1 year return % Near Alsi ATR 16.2% A Vol 26.2% MSI Options ATR 17.0% A Vol 12.3% Realised NAV
1-Feb-03 1-Feb-04 1-Feb-05 1-Feb-06 1-Feb-07 1-Feb-08 1-Feb-09 1-Feb-10 1-Feb-11 1-Feb-12 1-Feb-13
-20
700,000
-40
1-Feb-02

Review of the Non-linear Equity Curve Over the daily back test period the MSI Option strategy delivered market comparable returns (17% per annum net of costs compared to 16.2% for the Alsi over the same period). The volatility of return (based on the daily returns of the marked to market positions) was 12.3% compared to 26.2% for the Alsi (using daily data for both). Note the daily back test start date was some 6 months later than the weekly back test described in earlier sections (due to the availability of volatility data), hence the discrepancy in Alsi returns which achieved 18.2% on the weekly back test (start 8/2001) compared to 16.2% on the daily (start 2/2002) due to a more adverse start point or entry level. The second chart combines the equity curves with rolling 12 month return data, to illustrate the more stable return profile of the MSI Option strategy (worst year on year decline some 12% compared to >40% for the Alsi) and the historic non-correlation during market draw downs. This has important implications for the potential of a market timing product to be combined with an investors existing long equity exposures in stock selection mandates, to deliver a major improvement in risk adjusted returns, and hence in investor returns. For example, as shown earlier the option equity curve has an ATR of 17.0% pa and an AVol of 12.3% pa. A popular equity unit trust with a long and excellent track record, the Nedgroup Investments Rainmaker Fund, has an ATR of 20.0% and AVol of 17.5% pa on a daily basis over the same back test period (brown line overleaf, data ex Bloomberg). If I create a 50/50 combination with our MSIdriven option strategy and rebalance quarterly, the merged portfolio delivers a return of 19.1% pa at an AVol of just 11.5% pa (pale blue line over leaf). This great result is possible as the correlation of the daily returns between Rainmaker and the Option equity curve is only 0.17. Practical Implications of combining an option-driven market timing strategy with a stock picking unit trust The slight reduction in returns and dramatic reduction in volatility achieved by the 50:50 unit trust option strategy combination on back test may not seem that impressive to a non-technical reader. However the practical implications are far reaching. Few investors in a fund with a volatile return history will actually achieve the long-term reported fund returns, most will have received much lower returns due to the timing of their subscriptions and redemptions. By lowering the volatility of return through a blended investment in the option strategy, the average investor may actually receive a higher return as the timing of subscriptions and redemptions becomes less important in determining what you earn.
Page
22
Equity Curves - Log scale with YoY Returns
80%
884 801
70%
654
60%
50%
40%
30%
20%
10%
0%
Option Curve Year on Year %ge cge Nedrainmkr Year on Year %ge cge Option Curve ATR 17.0% A Vol 12.3% Blend 50:50 ATR 19.1% Avol 11.5%
-10%
23
-20%
100
90.0 2/1/2002
Nedrainmkr ATR 20.0% A Vol 17.5%
Page
Correlation between MSI Option Curve and Rainmaker equity just 17%
2/1/2003 2/1/2004 2/1/2005 2/1/2006 2/1/2007 2/1/2008 2/1/2009 2/1/2010 2/1/2011 2/1/2012 2/1/2013
-30%

4) Taking the MSI International - The China Situation Index (CSI) The CSI is my only daily model, and it requires close monitoring with an average of nearly 5 trade signals a month. It is worth it, however, as it has been the most lucrative model over the back test period. It is also by far the most volatile model, requiring a strong stomach to exploit. My live trading over the past year or so appears to confirm the back test potential. Building a model for the worlds most speculative market China was the first market outside South Africa that I tried to model. I had a fascinating visit in 2007 that sparked my interest in China, and given its importance for South African mineral exports, I have followed developments there ever since. After several spectacularly unsuccessful investments in Chinese companies, I decided to try and model the overall market it was clear my bottom-up stock picking efforts were a failure in China and I needed a robust framework to support my process. My standard Situation Index, build from fundamental, sentiment, technical and valuation macroeconomic factors on a weekly basis, was not an effective tool in the Chinese market. I tried many variations before finding a model more suited to the Chinese environment. I now had a model that worked, in fact worked spectacularly well, but I was concerned that I did not understand why this market should be so different. It was only when I read China specialist Michael Pettis November 2013 blog, When are markets rational? that the light went on. To quote from the article: There are also very few value investors in China because most of the tools they require, including good macro data, good financial statements, a clear corporate governance framework, and predictable government behavior, are missing. As a result, the vast majority of investors in China tend to be speculators. If we broadly divide information into fundamental information, which is useful for making long-term value decisions, and technical information, which refers to short-term supply and demand factors, it is easy to see that the Chinese markets provide a lot of the latter and almost none of the former. The ability to make fundamental value decisions requires a great deal of confidence in the quality of economic data and in the predictability of corporate behavior, but in China today there is little such confidence. http://blog.mpettis.com/2013/11/when-are-markets-rational/ My China model is focused on a much shorter time frame, with the CSI being calculated daily rather than weekly. The fundamental factors are omitted, leaving 7 technical, sentiment and valuation factors. Given the strict capital controls & that the government, not the market, sets interest rates and exchange rates, it is little surprise that my normal fundamental factors proved ineffective. Michael Pettis points above further illustrate this. The level of volatility is staggeringly high over the 6.5 years of the back test (some 2000 days), the SHASHR index delivered annualized total returns
Page
24

of 11.6% at an annualized volatility of return of 28.6%, and SHSZ300 index (the base index for the Chinese listed futures) had returns of 17.3% and volatility of 31%. Summary of the CSI Back Test Results SHASHR Index
China Situation Index Back Test

Position based on Calibrated Score Proportion of days Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts SELL 25% -28.8 -40.4 36.2 7.6 NEU 39% 6.3 -5.2 26.3 -2.3
Buy & Hold SHASHR
BUY 36% 59.3 47.7 25.1 -3.5
Index
2044 days 11.6 28.6
Backtest of 6.8 years of daily Bloomberg TR data
The chained returns of Buy periods predicted by the CSI annualize at 59%, compared to 6% for Neutral periods and -28.8% for Sell periods. The differential in volatility between the MSI periods, while significant, is much lower than that of returns, and even the Buy periods had high volatility by global standards (although this data is calculated from daily returns not weekly, which makes a significant upward volatility impact). Implementation of the CSI Only qualifying foreign investors can invest domestically in China (meaning I cannot currently trade the Chinese listed index futures), so I need to execute the models signals using a variety of Hong Kong listed ETFs, Singapore-listed index futures and US-listed ETFs. Fortunately the model appears sufficiently robust for the second derivative positioning to still be effective. Given the underlying markets volatility, it is no surprise that the equity curves obtained by objectively applying the CSI over the back test period are also very volatile. In practice I have mitigated the volatility somewhat by employing a mix of linear and non-linear positioning (using options) to express the CSI signals. The CSI gives an average of 5 trade signals a month, so profits are significantly impacted by the active cost of execution, but fortunately have historically been high enough to still deliver attractive net returns. I have live traded the CSI for just over a year and achieved similar results to that anticipated by the back test over that period. My trading is not entirely objective (I dont follow every signal) but when I do trade it is always with the signal (i.e. as indicated by the CSI score). An equity curve showing objective implementation of the CSI using a US-listed ETF is shown overleaf.
Page
25

600 550 6 500 450 400 350 300 250 200 150 1 100 CSI RHS CAF US Equity ATR 2.5% A Vol 41.8% CAF Equity Curve ATR 26.5% A Vol 33%
9/27/2007 9/27/2008 9/27/2009 9/27/2010 9/27/2011 9/27/2012 9/27/2013
CSI Equity Curve - US listed ETF
26
Page
50
9/27/2006

5) The land of the rising sun the Japanese Situation Index Like the MSI the Japanese Situation Index (JSI) is also made up of fundamental, sentiment, technical and valuation factors. However in contrast to South Africa, Japanese equities have had a torrid period over the back test time frame. The situation index methodology has done well in this very different environment, once again appearing to provide a significant edge in predicting returns and also had a more muted success in predicting volatility of return. Summary of the JSI Back Test Results
Japan Situation Index Backtest Summary

Position based on Calibrated Score Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts SELL 51% -12.1 -15.7 22.7 2.4 NEU 28% 12.2 8.6 18.4 -1.9
Buy & Hold BUY 21% 38.1 34.5 15.3 -5.0 Topix 627 3.6 20.3
wks
Backtest of 12 years of weekly Bloomberg data
Japan has had a very different equity environment to most global markets over the 12 year back test period. Annualised returns were only 3.6% pa with a volatility of return of 20.3% (Topix total return, weekly Bloomberg data). In contrast to South Africa and Australia (where >50% of the back test period was a Buy per the situation indices) only 21% of the time was the JSI showing a buy, compared to over 50% of the time a sell. Annualised returns in those few buy periods were excellent, and they were achieved with a significantly lower volatility of return. In contrast sell periods showed negative returns and higher volatility (refer table above). Why Japan? Japan caught my attention because of an interesting standoff the Governor of the Bank of Japan has promised to achieve 2% inflation. Japanese households, with some USD15 trillion of financial assets, have chosen to so far largely ignore this promise. The BoJ flow of funds data shows an overwhelming preference for Yen cash deposits and investments backed by domestic bonds, and very little equities or unhedged foreign securities (together only some 13% of household financial assets on a look through basis http://www.scribd.com/doc/139032397/Japan-Follow-UpInstitutional-Exposures-2013-05 ). The potential for massive domestic buying of equities should Governor Kuroda San start gaining credibility attracted my interest. Where else in the world can you inflation protect your portfolio and significantly increase your income yield at the same time?! (Equities and foreign bonds give you much higher yields than Yen deposits and bonds). JSI Factor Analysis
27
Page
However would the situation index methodology work in a market where equity Buy and Hold has been totally discredited as a strategy? Can my macro economic factors adapt to a zero interest rate policy environment? The answer was yes. What was different was the relative importance of This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved

fundamental & sentiment factors relative to technical factors compared to other markets. I believe this is due to the technical factors adding the most value in trending markets, and the choppy movements of the Topix index over the back test period were the antithesis of this. Of course that is not to say the market will not trend in the future, so a robust model needs to retain the technical factors.
Sentiment & Fundamental Factors

Score/4 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 9% -14.8 -18.4 25.9 5.6 1 27% -10.4 -14.0 19.4 -0.9 2 33% -0.1 -3.7 22.8 2.5 3 23% 20.1 16.5 16.6 -3.7
Buy & Hold 4 9% 50.4 46.8 13.9 -6.3 Topix 627 3.6 20.3
wks
Technical Factors
Score/3 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 41% -3.4 -6.9 22.9 2.6 1 9% -27.8 -31.4 18.0 -2.3 2 11% 1.4 -2.2 20.7 0.4 3 40% 21.0 17.4 17.4 -2.8
Buy & Hold Topix 627 3.6 20.3
wks
Valuation often appears to be far more diagnostic of returns at the sector or individual stock level than for the market as a whole. This certainly appears to be the case with the Topix index where over the back test period valuation provided only moderate market timing edge. Also while other factors have a stable inverse relationship between predicted return and predicted volatility of return, periods of cheap valuation (i.e. a high valuation score) are often periods of high volatility. This is another important reason to consider valuation separately.
Valuation Factors
Score/2 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 7% -7.8 -11.4 15.5 -4.7 1 41% 3.7 0.1 19.0 -1.3
Buy & Hold 2 51% 5.2 1.6 21.8 1.5 Topix 627 3.6 20.3
wks
Page
28
In summary the situation index methodology has adapted well to the Japanese environment, and is a highly effective tool to assist in market timing of Japanese equities. Graphic representations of the JSI and the Topix index, followed by the JSI equity curve are overleaf.

2000
7
JSI TPX Index

1800
6
1600
1400
1200
1000
800
600
Page
29
JSI Equity Curve - Topix Index Log Scale
1237
6
700 5
153
1
100
JSI exc Valn TPX Index ATR 3.6% A Vol 20.3% Equity Curve ATR 23.3% A Vol 19.0%
0
Page
30
70

6) The Australian Situation Index (ASI) Australia was an obvious international market to try the situation index methodology. The economy appears to have several similarities to SA and the likelihood of macro-economic factors being diagnostic in equity returns was good. This proved to be the case with a sound back test result as illustrated in the table below. In fact like the MSI, the ASI appears to be predictive of both equity returns and volatility of returns, providing interesting prospects for option-based strategy implementation. Summary of the ASI Back Test Results
Australian Situation Index Backtest Summary

Position based on Calibrated Score Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts SELL 14% -17.6 -28.0 27.5 11.7 NEU 24% 9.3 -1.1 14.7 -1.0
Buy & Hold BUY 61% 18.4 8.0 12.1 -3.7 AS51 639 10.4 15.8
wks
Over the past 12 years the ASI spent more than 60% of the time with a buy signal, and only 14% of the time with a sell signal a similar but more extreme pattern to South Africa and the MSI. The 2008/2009 draw down was more severe than in SA, with the ASI declining 49% and taking over 5 years to reach a subsequent new high. When calibrating the ASI one needs to take cognisance of the historic bull market in a flatter market future the returns associated with different macroeconomic environments may all take a step lower. ASI Factor Analysis The Situation Index methodology relies on the combination of factors rather than the individual categories. In different market and macro-economic environments different factors are important, and a factors impact on the overall score at certain times is the key. This aspect is ignored when looking at factor categories in isolation, as illustrated below, which rarely line up as symmetrically as they do for the ASI, both for returns and volatility of returns. In each category, like the situation indices themselves, a higher score is more bullish.
Fundamental Factors
Score/2 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 39% 4.3 -6.2 18.3 2.5 1 41% 9.3 -1.1 12.7 -3.1
Buy & Hold 2 20% 21.9 11.5 15.7 -0.1 AS51 639 10.4 15.8
wks
Page
31

The fundamental factors can be thought of as slow twitch, changes in factor scores often do not happen for years at a time. An exception to this is if the factor is sitting close to its calibration line (between 0 and 1), then a relatively insignificant move can change the score. In most of the situation indices the fundamental factors are based on proprietary models that estimate the current level of GDP growth and inflation (the big problem with raw reported economic data is the time lag so factors based off spot prices are essential).
Sentiment Factors
Buy & Hold 2 50% 19.0 8.6 12.6 -3.2 AS51 639 wks 10.4 15.8
The sentiment factors have historically changed much faster than fundamental, and can be appropriately categorized as medium twitch. Technical factors, however, are a mix of fast twitch and medium twitch, and change most often amongst the various factors. The technical factors are designed to highlight trending markets rather than mean reversion opportunities, hence a rangetrading environment often results in a decline in the situation index equity curve with poorly timed signals. It follows that in the technical factor analysis the market will often mean revert from short term moves, and these will influence the returns at each technical factor score. That said the ASI has spent a considerable portion of its back test history in sustained trends.
Technical Factors
Score/4 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 16% -3.6 -14.0 25.8 10.0 1 12% 11.7 1.3 17.9 2.1 2 13% -0.5 -10.9 13.1 -2.7 3 18% 15.8 5.4 13.7 -2.1
Buy & Hold 4 42% 16.7 6.2 10.7 -5.1 AS51 639 wks 10.4 15.8
berg data
The valuation Conundrum I have had much greater success with valuation factors at a sector or individual stock level. This is probably due to the offsetting impact of different sectors within the overall index for much of a market cycle. Only rarely is everything expensive or everything cheap. For example recently the resources sector has been historically cheap, but industrials are near all-time rating highs. While valuation measures clearly are important, for many situation indices they are best looked at separately from other factors. The prime reason for this is that they tend to reduce the effectiveness This material is provided for informational purposes only. This material may not be suitable for all investors and is not intended to be an offer, or the solicitation of any offer, to buy or sell any securities. Refer to the disclaimer at the end. Copyright MACROtimer.com 2014 All Rights Reserved
Page
32

of the volatility of return predictions of a situation index. Periods of very attractive valuation (i.e. after deep declines in the market) are often associated with very elevated levels of volatility of return.
Valuation Factors
Score/2 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 16% -0.1 -10.6 15.4 -0.4 1 62% 11.7 1.2 14.6 -1.1
Buy & Hold 2 22% 14.2 3.8 18.8 3.0 AS51 639 wks 10.4 15.8
In some of my situation indices my solution has been to exclude the valuation factors from the index itself, and rather to vary the bet size when implementing buy or sell signals depending on whether they are aligned with or contrary to the valuation score (e.g. if the ASI signals BUY and the valuation score is 2/2 the position size would be larger than a BUY signal with a valuation score of 0/2, and similarly on the sell side). Conclusion The back test of the ASI has given an excellent result, and I believe it could be a valuable tool to provide an objective quantitative framework for Australian investment processes that draw on top down views, whether to determine asset allocation for balanced long only mandates, or net equity exposures for long/short equity hedge funds. In addition the ASI can, together with the other situation indices or alone, be a profit generator in its own right (beta timing as alpha) in the global macro environment. Implementation is cheap and easy, with deep listed index futures, and importantly options on futures, available. The futures trading hours (pretty much always traded) make a linear long position very flash crash vulnerable as they trade during the US market time zone. (The near future contract declined more than 10% intra-day during the 6 May 2010 flash crash). Note that the iShares US-listed Australian ETF declined more than 17% intraday during the same event, so it is also unsuitable for very geared linear longs. The ASIs insights into volatility will really assist with option-based implementation to mitigate the flash crash issue. Charts showing the ASI score and the ASX200 index and the ASI score with implied volatility levels are overleaf, followed the ASI Equity curve derived from the ASX200 index.
Page
33
350
The ASI and ASX200 Index Total Return

Weekly Data (AUD)
7 300
6
250 5
200
3 150
2
100 1
34
ASI /8 ASX200 ATR 10.4% A Vol 15.8%

50 0
Page

50
Implied Vol peaks at 68.5% Q4 2008

45 ASI /8 inverted RHS Implied Vol Historic Vol 90d 40
ASI and Implied Volatility
-2
-1
1
35
2 30 3
25
4 20
5 15
10
35
Page
ASI Equity Curve - ASX200 Index Log Scale
8 810 7
5 333 4
100
Page
36
85
ASI /8 ASX200 ATR 10.4% A Vol 15.8% ASI ATR 19.1% A Vol 15.7%

7) The Korea Situation Index (KSI) The KSI model has delivered a sound back test, with big return and volatility of return disparities between the forecast Buy, Neutral & Sell periods. This includes the past 3 years or so which have been a narrow range trading environment (the Kospi200 index has traded within a 50 point range). This has impacted the equity curve based on the KSI, which is approaching a 2.5 year draw down. Despite this the KSI-driven equity curve back test delivers over 6 percentage points of alpha per annum with only 80% of market risk over the 12 year back test.
Korean Situation Index Backtest Summary

Position based on Calibrated Score Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts SELL 19% -3.2 -13.8 35.9 11.7 NEU 42% 4.6 -6.1 23.3 -0.9
Buy & Hold BUY KOSPI2 39% 622 wks 25.6 10.7 14.9 17.3 24.3 -6.9
KSI Factor Analysis The KSI is structured in a very similar way to the MSI & ASI, with slow twitch fundamental factors, medium twitch sentiment factors and medium & fast twitch technical factors.
Fundamental Factors
Score/2 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 17% -28.9 -39.6 32.5 8.3 1 56% 21.2 10.5 23.4 -0.9
Buy & Hold 2 KOSPI2 27% 622 wks 37.8 10.7 27.1 21.3 24.3 -3.0
Sentiment Factors
Page
37

The technical factors appear very influenced by the recent range trading environment, and so when viewed in isolation do not appear to add value.
Technical Factors
Score/4 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts 0 14% 21.8 11.1 38.6 14.3 1 18% 11.9 1.2 21.5 -2.7 2 22% -12.6 -23.3 25.2 0.9 3 19% 39.4 28.7 16.1 -8.1
erg data
As with the ASI the valuation factors highlight how when the market is cheap (i.e. 2/2 on valuation score) after a big price decline it is also very volatile. Because of this, and our desire to get a better insight on predicted volatility, the valuation factors are analysed separately and used to determine bet size.
Valuation Factors
Score/2 Proportion of weeks Chained Ann. Total Return % Differential to Buy & Hold %age pts Chained Ann. Volatility of Return % Differential to Buy & Hold %age pts
Implementation
Buy & Hold 0 21% 2.0 -8.7 21.4 -2.9 1 50% 10.0 -0.6 19.3 -5.0 2 KOSPI2 28% 622 wks 18.9 10.7 8.2 32.8 24.3 8.5
The South Korean market has a liquid index future, and liquid screen-traded index options, which affords cheap linear or non-linear implementation. The US-listed ETF is liquid & also has an active options market. Being USD denominated the ETF has the advantage of giving you an implied currency bet, i.e. long KRWUSD when you are long the ETF and short KRWUSD when you are short the ETF. Given the behavior of the KRW as a risk asset this positioning is appropriate, and while the volatility of the ETF and the equity curve derived from the ETF is higher, the alpha from applying the KSI objectively increases to some 10 percentage points pa (compared to 6% pa with the Kospi200 TR Index). Geared linear long positions in the index futures and the ETF must be considered flash crash vulnerable as they are traded through to the US time zone. The charts overleaf show the KSI and the Kospi200 Index, and the equity curve derived from the Kospi200 index.
Page
38
Page
39
KSI Equity Curve- Kospi200 Index Log Scale
8 675 7
5 336 4
100
40
KSI RHS Kospi200 ATR 10.7% A Vol 24.3% KSI ATR 17.3% A Vol 19.2%
2/1/2003 2/1/2004 2/1/2005 2/1/2006 2/1/2007 2/1/2008 2/1/2009 2/1/2010 2/1/2011 2/1/2012 2/1/2013
Page
70 2/1/2002

8) Combined Asian SI equity curve the benefits of diversification In evaluating the Situation Indices as a potential objective market timing tool, it is appropriate to analyse the favourable impact from diversification by applying capital across all four international models and markets. To simplify the currency impacts I used US listed ETFs for each of Australia, China, South Korea and Japan. I first created an equally weighted benchmark from these ETFs (rebalanced quarterly). The Korean and Australian ETFs have a history stretching back to 2002, and then the Japanese and Chinese ETFs were added from early 2006. This passive benchmark achieved a USD return of 11.3% pa at an annualized volatility of return of 23.5% to the end of 2013. I then divided capital equally between market timing models based on each countrys situation index. The trades were implemented using the US-listed ETFs, and capital also rebalanced between the models quarterly. This back test delivered a startlingly good result, with an annualized return of 27.5% at a volatility of 18.8% (also USD). This was with net exposures averaging 93% for periods the back test was net long and -38% for periods it was net short. Gross exposure was <200% at all times, and net exposure extremes were between 180% and -100% at all times (i.e. would classify for the Retail Hedge Fund classification per the SA Treasury white paper). The key behind this result was the relatively low levels of correlation between the different equity curves derived from each situation index. This is illustrated in the table below.
Correlations ASI KSI JSI ASI KSI 1 0.396 1 JSI 0.220 0.143 1 CSI 0.080 0.116 -0.046
While it is clear that during a major risk event historical non-correlation would count for little, in as far as the combined equity curve has off-setting positions (long and short positioning) it provides real protection. Note that the back test period includes both trending markets and dramatic sell offs. The composite benchmark (blue line) and equity curve are illustrated overleaf, together with the net exposures taken in deriving the equity curve. Data was from Bloomberg.
Page
41

2000
ETF based Asian Equity Curve - USD

CSI & JSI added H1 2006
200
1800
1804
150
1600
1400
100
1200 50 1000
0
800
600
400
Net Exposure % ETF passive ATR 11.3% A Vol 23.5% ETF Equity Curve ATR 27.5% A Vol 18.8% Ave Long 93 % Ave Short -38 % KSI & ASI from 2002
-50
359
-100
200
Page
42
0 2/1/2002
-150
2/1/2003 2/1/2004 2/1/2005 2/1/2006 2/1/2007 2/1/2008 2/1/2009 2/1/2010 2/1/2011 2/1/2012 2/1/2013

Disclaimer Kevin Cousins is a portfolio manager at Matrix Fund Managers. ("MatrixFM"). This article is prepared by Kevin as an outside business activity. As such, MatrixFM does not review or approve materials presented herein. The opinions and any recommendations expressed in this article are those of the author and do not reflect the opinions or recommendations of MatrixFM. None of the information or opinions expressed in this article constitutes a solicitation for the purchase or sale of any security or other instrument. Nothing in this article constitutes investment advice and any recommendations that may be contained herein have not been based upon a consideration of the investment objectives, financial situation or particular needs of any specific recipient. Any purchase or sale activity in any securities or other instrument should be based upon your own analysis and conclusions. Either MatrixFM or Kevin Cousins may hold or control long or short positions in the securities or instruments mentioned. Advice and any recommendations that may be contained herein have not been based upon a consideration of the investment objectives, financial situation or particular needs of any specific recipient. Any purchase or sale activity in any securities or other instrument should be based upon your own analysis and conclusions. Either MatrixFM or Kevin Cousins may hold or control long or short positions in the securities or instruments mentioned.
Page
43

Situation Index Handbook 2014

Uploaded by

Copyright:

Available Formats

Situation Index Handbook 2014

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Situation Index Handbook 2014

Uploaded by

Copyright:

Available Formats

The Situation Index Handbook

The Situation Index Handbook

The Situation Index Handbook

The Situation Index Handbook

The Situation Index Handbook

The Situation Index Handbook

Mkt Sitn Index RHS /8

Top40 Index ATR 17.2%

The Situation Index Handbook

Market Situation Index

Backtest of 12.5 years of weekly data 17 August 2001 to 3 January 2014

The Situation Index Handbook

Market Situation Index

Backtest of 12 years of weekly data

The Situation Index Handbook

The Situation Index Handbook

The Situation Index Handbook

Use Market Timing to lower risk

Market Situation Index

The Situation Index Handbook

Market Situation Index

Use Market Timing to lower risk Log Scale

Buy 100% Neu 0% Sell 0%

The Situation Index Handbook

Sector: Benchmark: Investment Objective:

The Situation Index Handbook

Use Market Timing to boost returns

Market Situation Index

The Situation Index Handbook

Use Market Timing to boost returns Log Scale

Market Situation Index

The Situation Index Handbook

The Situation Index Handbook

Market Situation Index

Backtest of 12 years of weekly data

The Situation Index Handbook

MSI and Implied Volatility

MSI weekly Implied Vol Near 0.5 delta Inverted Scale

45 Implied Vol peaks at 62.5% in 2008 (off chart) 50

The Situation Index Handbook

The Situation Index Handbook

The Situation Index Handbook

MSI Non-linear (10,6,4)

The Situation Index Handbook

The Situation Index Handbook

Equity Curves - Log scale with YoY Returns

Nedrainmkr ATR 20.0% A Vol 17.5%

The Situation Index Handbook

The Situation Index Handbook

China Situation Index Back Test

Buy & Hold SHASHR

BUY 36% 59.3 47.7 25.1 -3.5

2044 days 11.6 28.6

Backtest of 6.8 years of daily Bloomberg TR data

The Situation Index Handbook

CSI Equity Curve - US listed ETF

The Situation Index Handbook

Japan Situation Index Backtest Summary

Backtest of 12 years of weekly Bloomberg data

The Situation Index Handbook