JULI 10, Opportunities for technical models in sideways markets

Transcription

1 JULI 10, 9 ECONOMIC SITUATION AND STRATEGY Opportunities for technical models in sideways markets The stock markets are characterized at present by considerable uncertainties. Will businesses and consumers who expect current conditions to improve ultimately prove right, or will the world economy not recover so quickly? While it is quite typical of cyclical turning points that expectations improve first and clear recovery tendencies only become discernible later in real economic data, the danger exists that expectations have run too far ahead of actual development. Uncertainty among investors is therefore especially great at times when a cyclical turning point might be imminent and visibility on the markets is usually low Dez 89 Germany: Ifo business climate and industrial production Dez 91 Dez 93 Dez 95 Dez 97 Ifo business climate Dez 99 Dez 01 Dez 03 Dez 05 Dez 07 Industrial production (y/y; r.h.s.) Dez 09 10% 5% 0% -5% -10% -15% -20% -25% Technical indicators and models can provide valuable assistance then. Market technical analysis operates on the assumption that past price data can be used to generate buy or sell signals for the present and future. It has long been criticized by academics, but is used successfully by practitioners in day-to-day business. Assuming efficient markets, so say the theoreticians, no profits can be realized from past prices because all known information is already contained in current prices. Today s price developments are therefore only attributable to news that was not anticipated by the market, and not to prices of the past. This view even remained widespread when it was shown in some academic studies that excess returns could be systematically achieved with certain technical models. Newer branches of research in which human behavior is explicitly modeled are now also belatedly supplying different ways of explaining the effectiveness of technical models. When, for example, wellinformed investors buy a stock, other market observers can only see that the price for the relevant security advances. From that, they might conclude that information exists that justify a price increase and then start buying the stock themselves. The price would then advance again, and a chain reaction might get started ( herding behavior ). If one manages to tap such a price pattern with technical analysis, then profits can also be generated from information derived from past price data. One may generally distinguish between two different approaches in the practical application of technical analysis. Some technical analysts swear by the use of charts of past price data and try to discern patterns and formations that have provided reliable clues in the past about whether prices will continue to rise or fall and whether a price is supported from a certain level or encounters resistance. There are many different approaches, and it is left to the experience and good judgment of the technical analyst to make concrete trading recommendations. Other market participants use technical indicators to make statements about market development. Under technical indicators, various notions are subsumed that are supposed to permit inferences about future market developments from price development alone. Well-known and often-used indicators include, for example, the MACD, the Williams %R, and the RSI. The advantage of such indicators consists in the fact that they can often be formalized at little expense and can therefore be tested extensively. However, it should be mentioned that there are at once several different possible ways of calibrating even a single indicator. Thus, in some sources, one finds just for the Williams %R an oscillator concept used mainly for stock market trading various standard settings and formations that can be used to generate buy and sell signals. Nevertheless, once calibrated, the indicator yields clear trading signals whose quality can then be tested. In this report, we therefore want to limit ourselves to this part of market technical analysis. The simplest way obtain buy and sell signals with the help of market technical analysis is simply to use a single indicator with the usual settings. To be able to make statements about the quality of the signals, the indicator should be tested in advance on historical data. If one obtains a buy signal, then one enters the market at a certain delay and, as appropriate, with realistic transactions costs until the indicator gives a sell signal. This is shown as an example in the chart below, in which the results of an MACD trading strategy are depicted against the. The parameters for the MACD have been set to the usual values of 12 and 26 days for the two moving averages and to 9 days for the trigger line, and all three averages have been smoothed exponentially. An entry signal occurs when the MACD cuts above the trigger line from below, and one gets an exit signal when the MACD crosses below the trigger line from above. It has been assumed that after a buy signal, the investment is made in the stock index at the closing price of the relevant day. If the system gives a sell signal, then the capital is likewise withdrawn from the stock market at the closing price and is invested in the money market. We have assumed an interest rate for the money market that is oriented to the 1-year REXP. Those are realistic trading conditions for an institutional investor. 1

2 buy-and-hold versus MACD strategy MACD strategy At the current edge of the time series, an investor who has traded as described above with the support of the MACD would actually be about even with the. However, the investor would have had to show significantly worse performance for years relative to the benchmark. It may also be shown for other technical concepts that constant outperformance compared with a buy-and-hold strategy has seldom been attainable in the past by using a single indicator, especially if longer periods are considered than in the example above. Better results may often be achieved if, instead of a single indicator, several are used at once and a buy signal requires confirmation from further indicators. However, sometimes such concepts also do not work over relatively long periods. Although not always supplying equally good results should also be conceded to technical analysis, its sometimes moderate success is rather attributable to a different deficiency. Among the indicators, some rely on a continuing trend and others assume a trend reversal. While one can often observe price corrections after brief increases in sideways markets and a mean-reversion effect actually occurs, prices mainly move in one direction in pronounced bullish or bearish phases, and investors would be ill-advised in such situations to bet already on a trend reversal after a short time. At the same time, trend concepts often supply false signals in sideways phases because the current price always fluctuates around the sideways-moving market. To be able to exploit the strengths of both concepts better, it is a good idea to distinguish between sideways markets and trend markets and then use the appropriate indicators for the buy or sell recommendations. One can likewise make such a distinction on the basis of historical price data. The simplest way to identify trends (but one also subject to weaknesses) is to use the distance between a moving average and the current price as a measuring stick. If the price is too far away from the moving average, then one is in a trend phase; if the two are close together, then one is in a sideways market. On the other hand, sophisticated and more selective indicators are based on the direction in which prices are moving and the efficiency of the price movement. The direction of prices is solely a matter of what the ratio is of days with advancing prices to days with declining prices. The more that one of the two possibilities predominates, the more strongly the direction of movement indicates a trend. The efficiency of movement has to do with the question whether there is a clear price movement. In simplified terms, an efficient price increase would be, for example, one that continues from the beginning to the end of the trading day without the price dropping below its opening level or rising above its closing level in the course of the day. A trend exists when there is a clear direction of movement and the price movement occurs efficiently. The more pronounced the movement and efficiency are, the more pronounced the trend is. If there is no clear direction of movement and/or the price movements are not efficient, then it is a sideways market. If one uses such a method of trend identification by choosing between two indicators that can respectively play out their strengths primarily in trend markets and in sideways markets, improvements may again be achieved in comparison with a simple combination of two technical indicators. Another parameter for the performance of technical systems is the setting used for the individual indicator. It is easy to show, at least for historical price trends, that astronomical profits would have been possible with adjusted parameter settings for the indicators buy-and-hold versus optimized MACD strategy MACD strategy Even on the above-described MACD strategy with only one indicator, a substantially more buoyant course is taken with the right parameter settings (264, 95, and 113). Unfortunately, a naive optimization of a technical system to past data ( curve fitting ) often has the result that equally astronomical losses occur in real time. Nevertheless, there are possibilities of improving the parameter settings of a technical model. We have constructed a large number of time series for that purpose, which consist of randomly combined snippets of the original yield data

3 and constructed price series time series 1 time series 2 time series 3 Typical price patterns remain, but the individual series differ from one another considerably. Constructed price series thus exhibit booms and busts that are in some cases significantly heftier than any that have so far existed in reality. If one adjusts a technical system so that it functions well on a large number of such series, then curve fitting no longer comes up. Instead, one obtains an adjustment to typical price patterns. It is then no longer a prerequisite for such a system to function well that past prices recur as exactly as possible in the future, but rather that the forces that have produced the determined price patterns in the past also affect the market in the future. Considering that the markets are moved by people and fundamental human behavior does not change very fast, we regard this assumption as extremely plausible. We have set up and tested such a technical system. It consists of several indicators that assume a continuation of the trend and several indicators that rely on a trend reversal. A trend identification switches between the two groups. All parameters have been set so that they will achieve good performance in interplay with as many constructed price trends as possible. The results are depicted in the chart below investor must endure a daily loss of more than 4.6% with a probability of 1%, the figure for an investor with technical support would come to only 2.5%. The technically supported investment strategy also comes out ahead from the standpoint of how much volatility has been paid for the excess return in comparison with a risk-free asset. The annualized Sharpe ratio stood at more than 0.58 compared with the roughly 0.4 that would have been achieved on a investment. Overall, support from technical models can give investors additional clues as to where things are headed. We therefore consider it ideal in such market phases as today s to consider technical aspects in addition to comprehensively analyzing fundamental market and economic data in order to be able to identify cyclical turning points early. and technical supported trading strategy technical system The comparison between a straight buy-and-hold and the technically supported trading strategy shows that loss limitation is possible to a certain extent in bearish times. The technical strategy performs significantly better from the standpoint of risk. While in the period presented a 3

4 Weekly outlook for the period of July 13-17, 9 Feb. Mar. Apr. May June July Release DE: ZEW economic expectations July 14 DE: ZEW economic conditions July 14 EU16: Industrial production, m/m -2.8% -2.5% -1.9% 2.2% July 14 EU16: Industrial production, y/y -15.9% -18.4% % July 14 EU16: Consumer prices, m/m 0.4% 0.4% 0.4% 0.0% 0.3% July 15 EU16: Consumer prices, y/y 1.2% 0.6% 0.6% 0.0% -0.1% July 15 EU16: Core inflation, m/m 0.4% 0.6% 0.4% 0.0% 0.1% July 15 EU16: Core inflation, y/y 1.7% 1.5% 1.7% 1.5% 1.6% July 15 MMW estimates in red. Chart of the week: New orders and industrial production Germany: New orders and industrial production (Jan 1992 = ) Jan 92 Jan 94 Jan 96 Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jan 08 Jan 10 Industrial production New orders This week s chart shows the development of new orders and industrial production in Germany since the beginning of the 1990s. As we can see, industrial production and new orders are still at their 1999 levels despite the recovery in recent months. In any case, a stabilizing of new orders over several months may now be observed, and industrial production has increased lately by monthly comparison. Notably, at the current edge of the time series, industrial production is not being carried by one sector, but intermediate goods and capital goods as well as consumer durables are contributing to a welcome development. Another positive aspect is that new orders of intermediate goods also show a significant plus in monthly comparison, which points to continuing improvement in the months ahead. However, these figures come as no big surprise. There has been more or less consensus among economists for many months that stabilization or even slight growth may be expected from the third quarter onward. The purchasing manager indexes also suggest that the world economy is on the verge of stabilizing. There is nevertheless little reason for euphoria, since stabilization or even slight growth mean in the context of low utilization rates for manufacturers that earnings will remain low and will hardly be able to increase. That is why attention is focusing more on the reporting season that has just begun and will give information about how companies judge current conditions and expectations for the coming months. These data will presumably be more important for the market in the coming days than the release of macroeconomic time series. 4

5 As of Change versus Stock markets -1 week -1 month -3 months YTD Dow Jones ,2% -6,6% 1,2% -6,8% S&P ,5% -6,3% 3,0% -2,3% Nasdaq ,4% -5,8% 6,1% 11,1% Wilshire ,9% -6,6% 3,3% -0,6% ,7% -7,4% 3,1% -3,7% M 5.5-2,7% -4,1% 10,8% -0,9% Tec 613-0,9% -3,3% 14,8% 20,6% EuroStoxx ,6% -6,7% 2,9% -5,5% Stoxx ,8% -4,9% 6,4% -2,1% Nikkei ,3% -5,1% 4,2% 4,9% Topix 874-5,1% -4,8% 3,8% 1,7% Bond markets 3 months Euribor 1, months Treasury Bill 0, year US Treasuries 3, year Bunds 3, year JGB 1, US mortgage rate 5, IBOXX AAA, 4, IBOXX BBB, 6, ML US High Yield 13, JPM EMBI+, Index 441-0,2% 0,5% 5,5% 12,5% Convertible Bonds, Exane ,4% -0,9% 4,7% 7,7% Commodities CRB Index 389,14-2,4% -7,5% 3,4% 7,2% MG Base Metal Index 247,46-2,8% -2,6% 12,5% 44,7% Crude oil Brent 61,88-7,7% -8,8% 20,8% 56,5% Gold 914,75-1,9% -3,8% 4,0% 6,1% Freight rates Baltic Dry Index ,3% -14,2% 104,2% 289,9% Currencies Financial markets at a glance EUR/ USD 1,3990-0,1% 0,2% 5,4% 0,5% EUR/ GBP 0,8611 0,5% 0,1% -4,4% -10,9% EUR/ JPY 130,05-3,2% -4,9% -2,3% 3,1% EUR/ CHF 1,5119-0,7% -0,3% -0,9% 1,8% USD/JPY 92,95-3,2% -4,6% -7,5% 2,4% Carsten Klude Dr. Christian Jasperneite Matthias Thiel cklude@mmwarburg.com cjasperneite@mmwarburg.com mthiel@mmwarburg.com The reports, tables, and charts presented are based on reliable information from publicly accessible sources. We, however, make no guarantee of accuracy. The content is protected by copyright. M.M.WARBURG INVESTMENT RESEARCH 5