Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr Multiple-Length Stochastics by Stuart Meibuhr Plotting different indicators together and combining their analyses is a popular analytical method for timing trades. But how about using one indicator and varying its parameters? Stuart Meibuhr takes stochastics and varies the length of the lookback for trades of different time frames . T he popular concept "if one is good, two are better" has often been used by technicians in the selection of indicators in the belief that combining the signals of different indicators is a more reliable strategy than following just one. But this idea has rarely been applied to the use of multiple-length indicators, with the exception of perhaps multiple-length moving averages. Few other indicators have been studied for their value as multiple-length trading tools. In a recent STOCKS & COMMODITIES article, Barbara Star showed that the crossover of the seven-day and the 14-day commodity channel index (CCI, short and intermediate term) could serve as an early warning system forewarning of impending price reversal for live cattle. Star sometimes also used the 28-day CCI (long term) and even suggested that the three CCI plots be overlaid, with buying or selling then occurring when all the plots turn at the same time above or below the 100 level. One logical conclusion stemming from these charts was that a long-term stochastic could help the investor or trader remain on the correct side of the market. In contrast, Martin Pring used multiple-length rate of change (ROC) indicators on monthly and daily price data for the Standard & Poor's 500 index and other markets. He showed that when three ROC plots peak or bottom together, significant moves occur in the S&P 500. With monthly data, he used six-, 12- and Article Text Copyright (c) Technical Analysis Inc. 1 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr 24-month ROC curves that were smoothed by six- and nine-month moving averages. For daily data, he used 10-, 15- and 30-day ROC curves. Walter Bressert, in his book The Power of Oscillator/Cycle Combinations, calculated the relative strength index (RSI) indicator for different lengths using the Fibonacci series up to 21 on daily Treasury bond data. When all the RSI curves peaked above the 70 level at the same time, the T-bond price would also tend to make a significant peak. A similar situation existed for lows when the RSI curves bottomed below the 30 line. These multiple-length indicators appear to reflect the importance of the interaction of several different time cycles. When these cycles are peaking (or bottoming), all the different length indicators are also peaking or bottoming, and so, a low-risk trading opportunity exists. How the trader takes advantage of that opportunity reflects the needs of the individual trader; securities, options (puts or calls), LEAPs (long-term equity appreciation securities) and futures are all valid trading vehicles. Traders select the vehicles that best meet their trading time frames for the best possible dollar gain per dollar invested. And so, it would behoove us to examine the benefits that can be gleaned from the use of multiple-length stochastic plots (see sidebar, "Calculating stochastics"). In comparing the many charts, I realized that at certain times, many of the stochastics reached an extreme value simultaneously. That led me to overlay the stochastic plots. SIMPLY STOCHASTICS An article in the January 1990 S&C examined the use of a 14-bar stochastic on a monthly T-bond chart over a 12-year period. I was intrigued, so I charted the 14-bar stochastic on monthly data for both the Dow Jones Industrial Average (DJIA) and the OEX. Figure 1 shows the 14-bar stochastic for the DJIA and Figure 2 shows the same stochastic for the OEX. I then charted the 59-bar stochastic on weekly DJIA data, as there are about 59 weeks in 14 months. The weekly stochastic plot was more jagged, but the same times were identified by the extreme values. Moreover, the same time periods were identified by a 294-bar stochastic plotted on daily data (there are about 294 trading days in 14 months). One logical conclusion stemming from these charts was that a long-term stochastic could help the investor or trader remain on the correct side of the market. A 294-bar stochastic slowed by 21 units crossed below the 80 level on Wednesday, October 14, before the severe market tumble on Black Monday, October 19, 1987! Figure 3 shows the daily OEX; the vertical marking connects the day of the stochastic crossover with the same day on the index. How much shorter could the stochastic become before it no longer served as a useful trading tool? I experimented with various stochastic time periods, down to five. In comparing the charts, I realized that at certain times, many of the stochastics reached an extreme value simultaneously. That inspired me to overlay the stochastic plots. Now, for more than a year, I have been overlaying multiple-length stochastic plots as a means of identifying low-risk trading opportunities in the OEX and in stocks . These low-risk times occur when each of the stochastic plots are at an extreme value (above 80 or below 20), similar to the strategies that Article Text Copyright (c) Technical Analysis Inc. 2 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr use extreme readings of the CCI and RSI plots. I decided empirically to use four stochastics — 150/5, 50/5, 13/5 and 5/5 — more or less at random; there is nothing magical about these numbers. I did not optimize the parameters. These four sets of numbers represent the long term, intermediate term, short term and very short term, and the two components of each set represent the %K and %D values of the stochastic. To minimize clutter in the charts, I didn't plot the slowed %D (sometimes used as a trigger). All stochastics were plotted with simple moving averages rather than with exponential averages because simple averages calculate faster than exponential averages. Although calculation speed is not an issue when using MetaStock Professional, it did become an issue when I screened the database of stocks using another program. With MetaStock, using simple moving averages for the stochastic calculations meant using 150/5/1/s (the formula setup) as the conditions for the long-term plot. I wrote several MetaStock macros to perform the charting procedure for two, three or four stochastic plots overlaid. SOME SPECIFICS Figure 4 shows the OEX and the three longest stochastic plots overlaid; the dashed line corresponds to the 13/5 stochastic, the dotted line represents the 50/5 stochastic and the solid line is the 150/5 stochastic. In general, when all three stochastic plots are above 80 or below 20, profitable trading opportunities exist. The best (most likely profitable) trade was to sell an index option. Sell a call when the three plots are above 80 and sell a put when the three plots are below 20; other traders might prefer to buy the opposite option. The trigger to open the trade is when the 13/5 stochastic crosses either the 80 or the 20 line. Does this method work every time? No, because no indicator does. Does this catch every top or bottom? Of course not! Some trading opportunities will be lost, but it is better to make errors of omission rather than errors of commission. And using stochastic multiple lengths could increase my trading success rate. STOCK TRADING Do the multiple stochastic plots help identify buy and sell opportunities for stocks? Yes. First, you must determine the time frame you expect your trade to run. For example, if you are an intermediate-term trader, you might choose to follow the 50/5 stochastic line for crossovers of the 80 or 20 line. The trade is taken (open or closed) when this crossover occurs, but only if the next-longer stochastic line is also above 80 (for a sell signal) or below 20 (for a buy signal) . This signal works best with cyclical price movement but it should be confirmed by one or more of your other favorite indicators. This trading tool does not appear to be successful either with utility stocks or interest rate-sensitive stocks because of the recent strong underlying trend. It works equally well with low-priced (above, say, $2-3) and high-priced stocks. Figure 5 shows two stochastic plots (150/5 and 50/5) for Alcoa (AA). The 50/5 stochastic plot crossed down through the 80 line several times and yet only the last crossover was successful. Not all trades will work. This technique would not have allowed the trader to ride with the entire move from mid-December to the end of May, because the stochastics indicator is an oscillator and not a trend-following method. A short-term trader can see several opportunities for trading options each time both stochastic plots turn up or down. Figure 6 shows similar data for Chevron (CHV). Three buy signals were given, in mid-December, early February and early March. Certainly, the signal in March would have been the most profitable. This is another reminder of the importance of using stop-loss orders to manage risk. If the trader was long a call Article Text Copyright (c) Technical Analysis Inc. 3 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr purchased in March, then the first sell signal in mid-June should have caused the position to be closed. A short-term call buyer probably would have had a profitable trade in mid-December when both stochastics crossed the 20 line on the same day, but this signal was premature for the bottom. Does this method work every time? No, because no indicator does. Some trading opportunities will be lost, but it is better to make errors of omission rather than errors of commission. An example of a low-priced stock is Bally (BLY), which is shown in Figure 7 with its two stochastic plots. Two sell signals were given, one in late January and one in late February. The late February signal would have been profitable. Both signals indicated that any long position must be closed. A buy signal was given shortly before the beginning of this chart. The crossover of the 20 line by the 50/5 stochastic in early June is not a valid buy signal because the 150/5 stochastic had not yet fallen below the 20 line. This only proves that this technique will miss some profitable opportunities. I f a large number of stocks are followed, it is impractical to chart each one daily to find out if any have the desired stochastic pattern. Let the computer do that job. Several technical analysis screening programs exist that will determine if a security meets any desired stochastic pattern. I initially used TechniFilter Plus 6.0 from RTR Software but now have updated to version 7.0, which calculates much faster. Usually, only a very small number of securities — less than 1% of all the stocks in the database per day — display a desired stochastic pattern. On some days, no stocks exhibit the desired pattern. Once the program identifies a security with a desirable pattern, I subject it to a detailed technical analysis before making any buy or sell recommendation. The formula in TechniFilter Plus for the 50/5 stochastic is: ((C–LN50)/(HM50–LN50))A5 where LN50 is the lowest low of the past 50 days, HM50 is the highest high for the past 50 days and A5 is a five-day simple average of the 50-day stochastic. Using TechniFilter and MetaStock together, I have been successful in finding tradable stocks by using a stochastic-based filtering system. Two short-term stochastics can be used to identify the best times to sell uncovered OEX options. When both stochastics are above 80 or below 20, the OEX has reached short-term cyclical tops and bottoms during those times when the market is not in a strong up or down move. Such sideways movement occurs often, or at least it did during 1991 and 1992. Figure 8 shows two stochastic lines, 10/3 and 30/3. A +2 or –2 indicates that both stochastics are above 80 or below 20, respectively. A value of +1 or –1 shows that only the short stochastic is beyond its extreme value. While another indicator is needed to determine the exact day to place the trade, these spikes in the stochastic lines, if they endure more than one day, clearly correspond to turning points. To be successful with this trading method, good money management techniques are necessary to control the occasional losses. Article Text Copyright (c) Technical Analysis Inc. 4 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr In MetaStock, the formula to produce the chart in Figure 8 is: if(stoch(10,3),>,80,{AND}if(stoch(30,3),>,80,+2,+1), if(stoch(10,3),<,20,if(stoch(30,3),<,20,–2,–1),0)) I encourage the reader to experiment with values other than 10 or 30. I have explored several ideas on the use of stochastic plots for timing trades from short to long term, for stocks and options. Despite the versatility of multiple-length stochastic plots, other indicators should be used to confirm the signals from the stochastics. Further research may improve on the findings presented here. Stuart Meibuhr trades stocks and options for his own account. Recently moved to Scottsdale, AZ, he has lectured and taught on computerized investment topics for the past 10 years. He may now be reached at (602) 443-4926. Bressert, Walter [1991]. The Power of Oscillator/Cycle Combinations, Walter Bressert & Associates, Tucson, AZ. Hartle, Thom [1990]. "Stochastics and long-term trends," Technical Analysis of STOCKS & COMMODITIES, Volume 8: January. Pring, Martin [1992]. "Rate of change," STOCKS & COMMODITIES, August. Star, Barbara [1992]. "The commodity channel index," STOCKS & COMMODITIES, February. Figures Copyright (c) Technical Analysis Inc. 5 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr FIGURE 1: MONTHLY DOW JONES INDUSTRIAL AVERAGE, 14-BAR STOCHASTIC. Values below 20 indicate buying opportunities. The stochastics indicator stayed above 80 during most of the bullish phases. FIGURE 2: MONTHLY CHART OEX (S&P 100 INDEX) 14-BAR STOCHASTIC. The 14-period stochastic turned up from 20 at opportunistic times. Figures Copyright (c) Technical Analysis Inc. 6 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr FIGURE 3: DAILY CHART OEX FROM MAY 1987 THROUGH JUNE 1988 AND THE 294-BAR STOCHASTIC SLOWED BY 21 BARS. The long-term stochastic gave a sell signal several days before Black Monday in October 1987. Figures Copyright (c) Technical Analysis Inc. 7 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr FIGURE 4: DAILY CHART FOR THE OEX FROM OCTOBER 1991 THROUGH JULY 1992 SHOWING THREE STOCHASTIC PLOTS: THE 150/5 (SOLID LINE), 50/5 (DOTTED LINE) AND 13/5 (DASHED LINE). Important turns in the market occurred when all three indicators were either above 80 or below 20. Figures Copyright (c) Technical Analysis Inc. 8 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr FIGURE 5: DAILY CHART ALCOA FROM OCTOBER 1991 THROUGH AUGUST 5, 1992, SHOWING THE 150/5 AND THE 50/5 STOCHASTIC PLOTS. A successful buy signal appeared in December and a successful sell was indicated in June. As with most techniques, stop-loss orders can be used to help avoid large losses in case of false signals. Figures Copyright (c) Technical Analysis Inc. 9 Stocks & Commodities V. 10:12 (501-506): Multiple-Length Stochastics by Stuart Meibuhr FIGURE 6: CHEVRON (CHV) DAILY (OCTOBER 1992-AUGUST 5,1992) SHOWING THE 150/5 AND THE 50/5 STOCHASTIC PLOTS. A good buy was indicated in March. FIGURE 7: DAILY CHART BALLY FROM OCTOBER 1991 THROUGH AUGUST 5, 1992, SHOWING THE 150/5 AND THE 50/5 STOCHASTIC PLOTS. The best signal happened with the downturn in late February. Figures Copyright (c) Technical Analysis Inc. 10