Appendix 1. Time Fractals and the Supply/Demand Index

Transcription

1 Trading Regime Analysis: The Probability of Volatility By Murray Gunn Copyright 2009 John Wiley & Sons Ltd. Appendix 1 Time Fractals and the Supply/Demand Index Fractals refer to the phenomenon observed in Elliott Wave and Chaos Theories that natural life evolves in a series of self-similar patterns at each degree of evolution. Time fractals take this phenomenon to a market context by observing that a market evolves in a series of self-similar patterns at each degree of trend or each degree of time. We can see this clearly in the classic stylised Elliott Wave chart that each wave is subdivided into ever smaller and smaller waves of self-similar patterns. Whether we agree with the self-similar theory of fractals or not, the fact that a market can be analysed at each time fractal level means that time fractals are extremely relevant. In fact they are the most relevant pieces of information for the market analyst because if someone asks me the question What do you think about the market outlook? I have to know the degree of time or trend he is talking about. Does he mean the outlook for the next six months, the next month, the next week or the next six hours? If he wants the outlook for the next six months, I will analyse the weekly time fractal chart, but if he wants the outlook for the next week then the weekly chart is quite useless. I would need to analyse the daily, or more probably hourly, chart for that particular analysis. If we think about it in trend-following terms, when we define a trend by the slope of a moving average, then a market can be in an uptrend in the monthly time fractal, a downtrend in the weekly fractal, an uptrend in the daily fractal and a downtrend in the hourly fractal. Figure A1.1 shows the Nikkei 225 stock market index in Japan over the monthly, weekly and daily time fractals respectively as at 16 October Included in the charts is a 21-period moving average of the close and, if we define the trend as the slope of the moving average, we can see that on the monthly fractal the trend is up, on the weekly fractal the trend is down and on the daily fractal the trend is up. On each degree of time, therefore, analysis of a market can and does show different conclusions. In this case let s say you wanted to be a long-term investor and therefore

2 396 Trading Regime Analysis Source: Bloomberg 31/3/1994 Nikkei Monthly Time Fractal Trend Up /5/ /7/ /9/2006 Nikkei Weekly Time Fractal Source: Bloomberg Trend Down /11/ /11/ /10/ /10/ Source: Bloomberg Nikkei Daily Time Fractal Trend Up /8/ /10/2007 Figure A1.1 Nikkei 225 monthly, weekly and daily time fractals wanted to follow the monthly uptrend. Would you be long the market because the monthly trend is up, would you be flat because the weekly trend is down, or would you wait until the weekly trend turned up to go long the market if the monthly trend stayed up? If you were a shorter term investor who wanted to follow the weekly trend, would you be short the market because the weekly trend is up, would you be flat because the daily trend is up, or would you wait until the daily trend turned down to go short the market, given that the weekly trend stayed down? These questions can confuse market participants all the time and is one reason why trading psychology and behaviour occur in the way they do. An investor might have entered a long position in the Nikkei but, when he saw periods when the daily trend turned down, he may have started to feel quite nervous about his position. In fact, even though he is a long-term investor he may have become so nervous while observing the price action at the daily time fractal that he chose to cut his long position only to see the price turn back up and follow the direction of the monthly time fractal, which was up. As we know from the earlier section on psychology, people can get short term

3 Appendix 1: Time Fractals and the Supply/Demand Index 397 when they want to be long term, and can get long term when they want to be short term. It is usually the winning trades that become short term when they should be kept as long term, and it is the losing trades that get long term when they should be kept as short term. In other words, people tend to cut winning trades too early and run losing trades too long when what they should be doing is the exact opposite. An appreciation of time fractal analysis is crucial to having a consistent plan in the markets. If we start jumping around time fractals at a whim, then all we do is confuse ourselves and our analysis, which will then obviously affect our trading or investing. THE SUPPLY/DEMAND INDEX The more experienced I have become in the markets the more I have come to realise that this game of ours should be one of the simplest to play. However, as I personally am all too aware, the art of extracting consistent above-benchmark returns from the markets is clearly the most difficult endeavour any human being could hope to enter into. Nonetheless, a large part of my market analysis has essentially boiled down to the two elements this book is about. Either the market is going to be in a rangetrading environment or it is going to be in a trending environment, and so realising this has allowed me to dismiss many of the gimmicks dressed up as new theories or magic models that appear to make money in every time period (at least in backtesting!). No, either the market is trending or range trading and in my experience most trend-following techniques will find the trend to a greater or lesser extent, and most range-trading techniques will be able to extract value from the range to a greater or lesser extent. In the final analysis, everything in the markets boils down to being either a strategy that benefits from increasing volatility or a strategy that benefits from decreasing volatility. It s a long-volatility or a short-volatility world. Over the years I have therefore come to the conclusion that my personal style bias, in the long time fractals at least, is towards trend following because I am a natural buyer of volatility and, moreover, that my preferred method of measuring the trend is using the venerable moving average, or two moving averages to be precise. As I have mentioned, in my early years in the markets I back-tested and optimised systems and moving averages before I finally realised that optimising systems and parameters is, in my opinion, just plain crazy. Sure we can optimise parameters to find the best fit for a set of data points in a time series but the elephant in the room is always the question of how often to re-optimise. To me it is just data fitting. So I decided a number of years ago that establishing parameters that were consistent and sensible was the most important thing to do, at least in relation to trend following. There is no magic. Moving averages provide a discipline, not a magic black box. I decided that my preferred brand of trend following was the very straightforward continuous moving average cross system where the system is long when the shorter average is above the longer average and vice versa.

4 398 Trading Regime Analysis As far as I am concerned the basic and most important driver of market prices is the balance between the supply of that market and the demand for that market. As I have mentioned previously, the reasons why people want to supply the market or demand the market are quite frankly secondary to the actual act of doing it. Someone could sell a huge chunk of shares that he owns in a perfectly good company because he got out of the bed on the wrong side and was in a bad mood. Or he might need to raise some cash due to outflows from his funds. The reasons for people acting in any market are so numerous that worrying about them and trying to figure out why people are doing what they are doing is, in my view, very futile. What is important, however, is working out where the balance of supply and demand lies and where it might lie in the future. Fundamental analysts do this by coming to a conclusion about the future prospects of the market based on their forecast of future variables that may or may not be correlated with the actual buying and selling of the market and, crucially, making a big assumption that the majority of the market power will also agree with that outlook. Technical analysts do it by analysing the current balance between supply and demand in the market itself and anticipating how that might change, given the current psychology of the market. Technical analysts are trying to work out who has control of the market, the bulls (demanders) or the bears (suppliers) and there are a number of ways this can be done. However, my preferred method is to use moving averages of the closing price, as this method is a standard way to smooth out the noise in the market. I chose a moving average combination of 5 for the short one and 40 for the long one. There was no particular science involved in this and, in fact, it was probably an anti-science thing because I was so fed up concentrating on optimisation issues. I thought, why not choose something that feels sensible and can be used over all time fractals, and so I wanted an average that was short enough to be sensitive (but not too sensitive) and long enough to keep me in big trends (but not so rigid that it missed cycles). I picked 5 and 40 and have used these numbers ever since. To reiterate: there is no magic, only discipline. The philosophy I use is very simple. When the 5-period moving average is below the 40-period moving average, this is showing that supply is dominating demand (therefore there is downward pressure on the market) and when the 5-period moving average is above the 40-period moving average this is showing that demand is dominating supply (therefore there is upward pressure on the market). This is what the difference between a short moving average of the market price and a long moving average of the market price tells me. It gives me the information of when supply and demand dynamics in the market are changing. If the market has been in a long downtrend (supply dominating) then the 5 will be below the 40, and will have been for some time. However, if demand starts to show itself in the market it will show itself first in the 5-period moving average, and if the 5 moves up above the 40, then this information tells me that demand is now so strong that it is dominating supply. A great many people use moving averages but I don t think too many actually stop to think about the information they give us, and I believe the supply and demand

5 Appendix 1: Time Fractals and the Supply/Demand Index 399 information in looking at a short moving average against a long moving average is beautiful in its simplicity. Not only can this simple method tell us when supply is dominating demand or vice versa, but it can also tell us when supply or demand are at extreme points. This concept is measured by the distance the moving averages are away from each other. Figure A1.2 shows the Structural Supply Demand Index for the US Dollar Index from 1990 until Structural refers to the fact that I am using the monthly time fractal for this chart and the solid black line is the monthly closing price of the US Dollar Index, shown on the scale on the right. The scale on the left shows the difference between the 5- and 40-month moving averages on the US Dollar Index and the area graph shows the chart of these values. When the chart is below the zero line it is showing that the 5-month moving average is below the 40-month moving average and that supply is dominating demand. Similarly, when the chart is above the zero line, it is showing that the 5-month moving average is above the 40-month moving average and that demand is dominating supply. Source: Bloomberg L.P Demand Dominating Supply Dominating /10/ /4/ /10/ /4/ /10/ /4/ /10/ /4/2008 Figure A1.2 US Dollar Index (DXY) Structural Supply Demand Index We can see that when supply is dominating demand (and the ratio is declining) the market is in a downtrend, and when demand is dominating supply (and the ratio is increasing) the market is in an uptrend. This is simply what we would expect, given the fact that we are using the first derivative of the price itself (i.e. the moving average) and the movement in the index, as we can see, lags the price movement slightly. However, my belief is that periods of supply dominating and demand dominating can persist for a long time and, when they do, the market in question will experience a strong trend. The absolute key here is how often this persistence of supply or

6 400 Trading Regime Analysis demand occurs and, in actual fact, most of the time the periods of supply dominating or demand dominating will be relatively short, resulting in a choppy price action. However, when the minority periods of supply or demand persisting come along, the market s trend is so strong that it generally compensates, and more, for this period of choppy price action, or whipsaw in the language of trend following. The bottom line here is simply this: What I am presenting is nothing radically new; it is just two moving averages being used in a trend-following system to show when there is likely to be an uptrend or when there is likely to be a downtrend. However, by presenting it in this fashion I hope I can get the point over to the reader that what moving averages show us is the most important element in any analysis of any market: that is, the balance between supply of and demand for that particular market. Supply and demand, as I am sure the fundamentalists and the economists would agree, is the ultimate arbiter of whether a market will go up, down or stay the same. There is, in the final analysis, nothing else to worry about. This is why I use the Supply/Demand Index as my most important and basic analysis technique for gauging market direction. The trend is your friend is one of the all-time truths in market lore. Although, as I have demonstrated in this book, it is possible to identify trend exhaustion and take advantage of that; to continuously fight a persistent trend is one of the most soul (and account) destroying activities anyone could wish to undertake. The trend is your friend. Burn it into your psyche. Using a simple and consistent technique like the Supply/Demand Index enables the analyst to easily gauge supply and demand dynamics across multiple time frames or fractals. This gives the analyst a global view of the supply and demand mix because it will show how different time fractals are interacting with one another to give ideas and signals to trade or invest. For example, Figure A1.3 shows the Supply/Demand Index for the DXY (US Dollar Index) over the Structural (monthly), Cyclical (weekly) and Tactical (daily) time fractals. We can see that presenting the data in this way immediately gives the analyst a sense of where the market is on all fractals at any one point of time. This was snapped on 26 May 2008, and on that date it shows that supply of the US Dollar Index was dominating over all time fractals. Notice that the Tactical Supply/Demand Index had signalled that supply was starting to dominate demand on 19 February On that date, too, the Cyclical and the Structural Supply/Demand Indices were showing supply dominating demand. On this date in February, therefore, the analyst would have seen a situation where supply was dominating in the two higher time fractals and that supply had just begun to dominate demand in the Tactical time fractal. We have a market where downward pressure is prevalent in two larger time fractals and where it is just becoming prevalent in a lower time fractal. When something like this appears it is usually a very good sell signal for the medium-term trader because the probabilities are stacked in favour of the supply pressure. This proved to be the case here with the US Dollar Index falling from around 76 to around 71 a few weeks later. Using Supply/Demand Indices in this way, therefore, can identify some quite profitable opportunities for the trader or investor. In fact, this method of looking

7 Appendix 1: Time Fractals and the Supply/Demand Index 401 TACTICAL CYCLICAL STRUCTURAL Source: Bloomberg L.P /10/ /4/ /10/ /4/ /10/ /4/ /10/ /4/ /12/ /3/ /10/ /5/ /12/ /7/ /2/ /9/ /3/ /10/ /5/ th Feb /1/ /2/ /4/ /6/ Figure A1.3 US Dollar Index (DXY) Supply/Demand Index at two or three time fractals in order to stack the odds in favour of a trade is one that is well known and probably best described by Alexander Elder in his book Trading for a Living. Elder referred to his method as triple or double screen and he used slightly different techniques, but the philosophy is the same. The priority is to identify where the larger scale pressure (or trend) is in relation to the market and this can be determined by looking at the longer time fractals. The analyst can then look at the shorter time fractals to identify when there might be an extension of that pressure, thus producing buy or sell signals. The beauty of the method is that it can be very flexible and used by long-term, medium-term, short-term and even day traders. For instance, look at the Cyclical Supply/Demand Index in Figure A1.3. Around April 2006 the Cyclical Index started to show that supply was dominating demand in the US dollar and this fact in itself is interesting. However, if we look at the Structural Supply/Demand Index for around the same time we can see that, although the supply pressure had weakened considerably from its extreme point around 2004, the overall supply pressure had still not abated (the index was still below zero). So when the analyst noticed that the Cyclical Index was showing that supply was dominating then, allied to the fact that the Structural Index also had supply dominating, it was a very big clue that the downtrend in the US dollar that had started in 2002 was not yet over. If the trader or investor had used this method to enter into a short US dollar position at that time, by the start of 2008 and obviously depending on their money management strategy,

8 402 Trading Regime Analysis he could have been sitting on the correct side of a 20% move in the currency market! This, of course, is far, far easier said than done (a fact to which I can testify!) and we all know that the best trader or investor in the market is Harold Hindsight. Nevertheless, even although I do not use the interaction between the time fractals of the Supply/Demand Index as a mechanical trading system, preferring to use it as a discretionary tool for position sizing, the method seems to offer some value for picking high probability occasions of trend extension. It is not infallible, nothing is, but it makes logical sense to me. TACTICAL CYCLICAL STRUCTURAL Source: Bloomberg L.P /10/ /4/ /10/ /4/ /10/ /4/ /10/ /4/ /3/ /10/ /5/ /12/ /7/ /2/ /9/ /3/ /10/ /5/ /11/ /12/ /1/ /2/ /3/ /4/ /5/2008 Figure A1.4 S&P 500 Index Supply/Demand Index Figure A1.4 shows the S&P 500 equity index in the United States of America and the relevant Supply/Demand Indices. We can see that the primary Supply/Demand Index was showing demand dominating all the way through the mega bull market of the 1990s turning to supply dominating in the downswing and then back to demand dominating from late What is interesting from this long-term index is that the power of the net demand in the upswing from 2003 had not, at the time of writing at least, come anywhere near the power of the net demand during the bull market of the 1990s even though the S&P Index itself had traded back up to the levels seen at the peak back in This is a sign that the underlying strength of the market is not as powerful as it was in the last upswing and is therefore a warning sign for bulls. In fact, the S&P started to fall from re-testing the 2000 peak in 2007 and a study of the Intermediate Supply/Demand Index shows that supply was dominating from around November 2007, having admittedly only very briefly dominated in August

9 Appendix 1: Time Fractals and the Supply/Demand Index Any long-term investor who was long the stock market towards the end of 2007 and was using the methodology of fractal based supply/demand indices, would (should) have been reducing his longs at the very least. We can see from the Minor Supply/Demand Index that supply started to dominate in early 2008, well before the big falls in the market in the middle of January which caused a bit of a minor panic at the time. Therefore, using a simple study of supply and demand in the market can turn out to be quite useful in enabling investors and traders to effectively manage their positions on a dynamic basis, thus being able to either cushion the blow of trend reversals or take advantage of movements in the direction of the bigger trend. AN INVESTMENT PROCESS IS SIMPLY A LONGER TERM (SUBJECTIVE) TRADING SYSTEM A lot of fundamentalist analysts deride those that follow a systematic approach to investing or trading on the basis that those systems tend to be biased towards using price data as the main input. Even for those systems that use fundamental data inputs for making buy and sell decisions, there is criticism that they are inflexible and therefore not as useful as a discretionary methodology. As I have alluded to elsewhere, while I agree that a purely systematic approach can be quite dangerous at times when the human psychology responsible for managing that model is flawed, or when the model does not take into account market depth or liquidity (think LTCM), a systematic approach to investing is, in fact, what we are all trying to achieve. The big institutional pension and mutual fund managers are given money to manage by their customers (such as large corporations or municipalities) and these customers will only consider giving them money if the fund consultants are satisfied that the fund manager in question has fulfilled certain criteria. What the consultants are looking for in a fund manager can very often be summarised as the three P s: performance, people and process. Performance is an obvious one because the customer will want to give his cash to a fund manager with a good track record of performance. In actual fact this is quite a debatable point because many people believe that, as performance tends to be mean reverting over time, you should actually rotate your allocations to fund managers by firing the top performing managers and hiring the worst performing managers. The philosophy is that the top performers over (say) the last three years will tend to start to underperform, whereas the worst performers over the same historical period will now tend to start to outperform. This logic is the same as that behind the Trend Following Performance Indicator. Periods of underperformance will eventually lead to periods of outperformance, and vice versa. The second P stands for people, because fund consultants want to be confident that the people employed to manage and administer funds are (a) competent and (b) likely to be there for some years. A company with a high degree of staff turnover is, at least in the consultant s view, symptomatic of a company with problems in

10 404 Trading Regime Analysis the areas of staff morale and retention. As any business, in particular a service-based business, is nothing without its people, the consultants have to check this box as part of their work. The third P stands for process, and this means the investment process. This is the area where fund managers can really differentiate themselves because it is the area in which their investment style or bias becomes evident. Some managers will have a bias to a fundamentally based process, some to a technically based process and some will have a process that includes both fundamental and technical market analysis. The most important aspects that a consultant will look for, however, is not whether the process is considered to be good in their opinion but that it is logical and, most importantly, consistent. Logic and consistency are therefore the watchwords for fund managers when they are thinking about their investment process, and the process will be written out and followed by the asset class teams involved in managing the funds. A very simple example of a fundamentally driven investment process might involve a list of criteria that have to be thought of or checked before an investment decision is made. That is, each asset class within the fund management operation might have to answer questions such as: What are the drivers of the market or stock in question? What is changing in the (fundamentals) of the market? What is in the price already? What are the triggers for the investment decision? These questions provide a logical and consistent framework for an investment process, albeit one that gives a lot of scope for interpretation and meaning. The answers to these questions from a fundamental perspective will be quite subjective by nature, and one person s answers might be different from another person s answer. In this sense the investment process becomes quite fuzzy. For example, the so-called drivers of the market in question will surely be numerous and their relative importance will differ from one person to the next. I might think that the US dollar is being driven by the price of oil at the moment but my colleague might think that it is interest rate spreads. We can both produce evidence to support our case in the form of correlation coefficients. However, another colleague might say that this is just coincidence and that the real driver of the market is the central banks diversification of their reserves. Also, the answer to the question of what is in the price already is inherently subjective when viewed from a fundamentalist s perspective. In some markets it is easy to identify what is in the price ; for example, in the futures market of the Fed Funds 30-Day interest rate, there is an expiry date and a price, so the analyst can tell instantly the price of the Fed Funds rate at any given time. However, in most other ways the answer to the question of what is in the price? becomes a subjective debate which is impossible to resolve in any meaningful way. A stock might have had a big fall and I may think that the bad news that accompanied it is in the price already. My colleague, however, might disagree with that assessment and the problem is that there is really no way to resolve it. Now consider a more technically minded investment process (for example, a trendfollowing methodology) that follows the same logic as above. The answers to the same questions now become very objective.

11 Appendix 1: Time Fractals and the Supply/Demand Index 405 Q: What are the drivers of the market or stock in question? A: Numerous and varied drivers or reasons exist that cause market participants to act in the market by buying or selling. Q: What is changing in the market? A: The price is constantly changing to reflect the collective discounting of the future by the market crowd as a whole. Q: What is in the price already? A: All that is known or expected by market participants is in the current market price. Q: What are the triggers for the investment decision? A: When the designated short moving average of the price crosses the designated long moving average of the price from below, the decision is to buy/go long, and when the opposite happens the decision is to sell/go short. This shows that an investment process is simply a set of rules that provide a framework within which the investment manager, trader or analyst can work. If it is consistency of investment process that is desired, then a process involving at least some form of technical market analysis (e.g. for the trigger decision) has to be better than using a purely subjective analysis. SUMMARY In this appendix we have looked at the application of fractal-based analysis on a market and how it can help to give the analyst a much broader understanding of the overall market conditions. Combining time fractals is, in my opinion, critical to understanding the context of the entire market, be that in terms of price or volatility direction. We have also examined how we can incorporate basic supply and demand time fractal analysis into an investment process in order to gain a higher degree of consistency.